Twelve lookout stations stood along the shore of Saipan, built the year before the Pacific War began, funded by a Japanese naval appropriation that officially listed them as lighthouses. Each came with barracks, a command post, an ammunition shed, and a man with binoculars watching a vast stretch of empty water. But that money bought far more than lookouts. Between 1934 and the autumn of 1940, the Japanese government appropriated at least 14.
5 million yen for construction in the Marianas alone. A further grant the following September covered barracks, baths, latrines, kitchens, infirmaries, storehouses, workshops, torpedo sheds, garages, and air-raid shelters, and the order authorizing it stated in writing that the buildings were for a base force and a defense force, both designated as fighting units under Japanese naval organization. The naval unit already on the ground spent the second half of that year on a task with nothing defensive about it: planning and preparing the invasion of Guam. For most of the war that followed, there was little on the horizon to watch.
The naval base force on the island numbered under a thousand men in the spring of 1943 and had not reached 1,500 by February of the following year. Then, on June 8, 1944, with the island a week away from the arrival of the largest fleet yet assembled in the Pacific, the headquarters those lookouts answered to issued an order moving 2,500 soldiers off Saipan to other islands in the group. That order was a command acting on its best reckoning of where the Americans would strike. The Japanese naval high command had settled on the Palaus or the Carolines, somewhere to the southwest.
After the American carrier raid on the Palaus at the end of March, the loss of Hollandia in April, and the fighting around Biak in May, the high command believed Saipan would not be targeted until later in the year. As late as June 14, with the American bombardment already falling on the cane fields, Vice Admiral Chuichi Nagumo was still putting it in writing that the Americans would land in the Marianas this month or next. Saipan, beneath all of this, was a sugar island with three plantations and two mills, each mill capable of processing 1,200 tons of cane a day. The molasses went into alcohol, and the alcohol went into synthetic Scotch whiskey, port wine, and several other bottled products made for sale back in Japan.
But what the Japanese officers stationed there had been putting on paper through the spring told a different story from the one their own headquarters believed. Six of those papers survive. Read backwards from the invasion, they form a sequence. The last of them is dated May 31, two weeks and a day before the landing.
It went out over the signature of the chief of staff of the 31st Army, Major General Keiji Igata, and it was a survey of the defenses of the various islands. Igata wrote that fortifications could not be strengthened to any worthwhile degree without materials fit for permanent building, then listed them, because listing them was the entire point: cement, steel reinforcement for the cement, barbed wire, lumber, none of which could be had in the islands. Without those materials, he wrote, the number of soldiers standing on the ground made no difference, because men with nothing to build with could only fold their arms. He called the situation unbearable and asked for the matter to be dealt with at once.
A staff officer is permitted to say a great deal in a paper of that kind, and Igata said nearly all of it. The one thing he did not say was what would have to be true for his request to be answerable. Before that came the paper of May 20, headed as the outline of the defensive plan of the Northern Marianas force. It set out what the garrison would do.
Units would build strength enough to destroy an American landing force at the water’s edge. The islands would become a fortress, and the airfields would be held. The plan admitted of itself, in its own text: “Although tardy,” it promised field positions finished in the first ten days of June, with permanent works to follow afterwards at the important points. Nothing inland was mentioned anywhere.
Everything in the plan happened on the beach, which on a flat coral atoll was orthodox Japanese practice, but on a volcanic island fourteen miles long with a 1,500-foot mountain in the middle of it, it left the best ground on Saipan unfortified. The officer who drafted it was working against a clock he did not control. The army’s own program called for field positions within a month of a unit’s arrival and permanent works within three months. The engineers responsible for the permanent works, the 7th and 16th independent engineer regiments, did not reach the island until April.
The division meant to hold Saipan, the 43rd, did not arrive until the first week of June. There is a concrete measure of what the beach scheme meant on the ground. On May 25, the commander of the 136th Infantry Regiment issued an operations order for the central sector, which covered most of the shoreline where the 2nd Marine Division would come ashore three weeks later. Four rifle companies and an infantry gun company went into the front line on the beaches.
The reserve for that entire sector amounted to two-thirds of a single company. One item in the plan was built in quantity. It called for dummy emplacements laid between the real strong points and between the coastline and the support line, thickest of all along the stretches where a landing was least likely, so that American gunfire would go into empty ground. The Americans photographed a good number of them, and the official history contains a picture of a dummy searchlight platform on the east coast that is convincing.
The paper before that is dated May 10, and it is the one that answers Igata. Headed as the state of supplies and materials in the Southern Marianas, it set out the freight position in a single line. One ship in three was being sunk. Another was being damaged.
Every item on Igata’s list, every bag of cement and every coil of wire, had to arrive from Japan by sea, and the sea lanes were no longer safe. The answer to the last warning was already in the files, written three weeks before the warning went out. So the question stops being whether the warnings were clear. They could not have been much clearer.
The question is where they went and what sat on the other end. The Japanese command arrangement in the Central Pacific in 1944 did not have one obvious upward address. On paper, the 31st Army answered to the Army section of Imperial General Headquarters in Tokyo. In March, a new naval command, the Central Pacific Area Fleet under Nagumo, went in above both the Army and Navy forces across the Mandated Islands and the Bonins.
The Army objected to serving under an admiral. That argument ended on March 15 in a compromise by which each island would be run by whichever officer on it was senior, and Nagumo and Lieutenant General Hideyoshi Obata agreed between themselves, out loud and without writing it down, that neither would take full authority over the other. It left the theater without a supreme commander. Igata’s report of May 31 is addressed to the chief of staff of the Central Pacific Area Fleet, whose headquarters sat a short walk from his own on the same island.
A Japanese naval commander taken prisoner on Saipan afterwards called that headquarters an administrative shell with no tactical meaning, an organization that never got past being an arrangement on paper. The next paper has no date on it at all in the surviving copy. It is a 31st Army table showing the ammunition needed at once for anti-aircraft guns and machine cannon, the sort of table somebody types after doing the arithmetic and finding the answer frightening. The allowance for those weapons on Saipan was three units of fire.
The regiment on the island and its two attached companies did not hold one. A unit of fire was 4,800 rounds. The same shortage ran through every area the 31st Army was responsible for. None of this stayed abstract.
American troops moving inland in June would find heavy weapons on Saipan, from 200-mm mortars up to 5-inch coastal guns, along with searchlights and anti-aircraft pieces, sitting in naval depots chained to railway flatcars, packed in grease, beside gun pits dug halfway and then abandoned. The garrison had no horses and almost nothing that could travel across country. The only machines for shifting artillery were rear-drive trucks built for roads, so guns that men could not drag by hand stayed where they were. The pieces that did get into position had been sighted in with care on the obvious targets, especially the breaks in the reef.
But each battery had done that sighting alone, with no common survey control behind it. Japanese artillery on Saipan was never able to mass its fire. The airfields told the same story in a different currency. In March, both the Combined Fleet and the Central Pacific Area Fleet issued orders for a rapid build-up of air bases, the aim being a large number of rough fields at speed, brought to a finished state later.
The program came to fourteen airfields and two seaplane bases across the group. Normal facilities were to be complete by April. When the Americans arrived in June, nine of the fourteen were operational. In the first week of March, the papers stopped being requests and became an alarm.
On March 3, the 31st Army issued an emergency plan for the defense of the archipelago, and at that date, the archipelago was all but undefended. The following day, the Army abolished civil government on Guam and brought in martial law across the islands a few days later. Americans later took excerpts from the Army Headquarters Staff Diary covering late winter into March, and the diary put the position without decoration. The American strikes had alarmed the Central Command.
After the carrier attack on the Marianas on February 22 and 23 had shown how thin the defenses were, Tokyo expected an invasion within the month. The same diary shows the headquarters trying to turn the 8th Expeditionary Unit North into the Marianas instead of letting it go on to Truk. The headquarters thought about that and then let the unit go on south. Tokyo was not asleep, and Tokyo did not ignore its officers in the islands.
Tokyo had already answered in February, at once and at the highest level. The staff of the 31st Army had been getting ready to move out to the Marianas when the Marshalls fell. Once the islands had been hit from the air, Prime Minister Hideki Tojo ordered them to leave for Saipan the very next day. Between January and the first days of June, the Japanese put about 45,000 soldiers aboard ships bound for the Marianas.
Tokyo had answered, and the answer was an army, and the army went by sea. At least 12,000 of those men were aboard hulls that were torpedoed on the way out. Many of the ones pulled from the water were carried on and landed anyway on whatever island the surviving ships were nearest. A large share of the reinforcement came ashore without weapons, without papers, belonging to units that were somewhere else entirely.
Survivors of one convoy bound for Yap were still sitting on Saipan in June waiting for shipping. So were men of a regiment that had been going to the Palaus. In May, the army renamed and reorganized its units across the whole area, so that a garrison headquarters could not always say with confidence what it held. At that point, the six papers stopped being six unrelated problems.
Every officer who wrote one of them was describing the same American capability from a different angle. The supply paper of May 10 saw a freight ratio. Igata saw an absence of cement. The ammunition table saw rounds not held.
The Northern Marianas plan saw time that had run out. The staff diary saw aircraft over the islands in February. What all of them were looking at was an American campaign against the Marianas that had been running for four months before a single rifleman came ashore, fought at sea and in the air against shipping by submarine crews and carrier pilots that almost nobody in the garrisons would ever set eyes on. Most of what is known about any of this survives because a Japanese clerk filed a paper and later an American translator gave that paper a number.
These papers are known today as American translations of captured Japanese originals, given translation numbers by Pacific Ocean Areas Intelligence or by the landing force in the field. The surviving copies show the addressee far better than they show the routing afterwards. Igata’s report went sideways to a naval headquarters on his own island rather than to Tokyo. The ammunition table carries no date at all, so its place in the sequence is a reasonable guess and nothing firmer.
Which of these specific warnings reached which desk in the Army section in Tokyo, in what form and in what order, is not something the surviving papers establish. The first of the six is the plainest, and it asks for nothing. It is an arrival report written in the first week of March about what had just come ashore. The 18th Infantry Regiment had sailed from Ujina in Japan on February 26, 1944, aboard three troop ships, part of the 29th Division being shifted out of Manchuria to hold the islands.
3,080 men were aboard the ship that was hit. 1,688 of them came out of the sea alive. What the regiment still possessed when it stood on Saipan went up the chain as an inventory: seven rifles, one grenade thrower, two light machine guns, and 150 bayonets. The ship was the Sakito Maru, hit late in the afternoon of February 29, 1944, by an American submarine that did not come home afterwards and never reported the attack.
Three and a half months later, the Americans landed.
