Chapter 18: Precision Bombardment Campaign
What with the Columbus Day tradition, it may have been a lucky omen that the first B-29 landed at Saipan on 12 October 1944. She was Joltin’ Josie, the Pacific Pioneer, piloted by Brig. Gen. Haywood S. Hansell, Jr. “Possum” Hansell was already a veteran in the B-29 program. As air member of the JPS he had perhaps done more than any other to plan the VHB assault on Japan. More recently, as Arnold’s chief of staff in the Twentieth Air Force, he had directed from Washington the operations of XX Bomber Command in CBI. Given command of XXI Bomber Command on 29 August, he was now slated to control at closer hand the VHB offensive from the Marianas.
The arrival of a commanding general is an impressive event even when he emerges feet first from the belly of a bomber, but the first appearance of a Superfortress is something else again and Joltin’ Josie shamelessly stole the show. According to Brief, the breezy magazine that supplied news and pin-up girls for AAF personnel in the Pacific, “The war just about stopped dead in its tracks the day Joltin’ Josie arrived. ... The first of the B-29’s had been inspected by every big gear and ogled from afar by every small fry for 5,000 miles. She was a sensation.”1 Certainly the war effort was halted momentarily at Isley Field. A group historical officer reported that as the huge bomber swept in with its fighter escort, “a great cheer went up, and all work stopped as men shaded their eyes to watch the plane pass over. ... The thrill that went through all was almost electric in effect.”2
There was just cause for elation. For Hansell, the safe arrival was a token of the success of the plan he had steered through the JCS, whereby POA forces had bypassed Truk to seize the Marianas as a
base area for the VLR bombardment of Japan. For the men who greeted him, the coming of the B-29 was the first tangible evidence that their labor in steaming heat and tropical rains was to bear fruit. Yet Joltin’ Josie’s landing was still no more than a token. One airplane, even a Superfortress, did not constitute a striking force, and there was much prosaic work to be done before XXI Bomber Command could drop its first bomb on Tokyo. Isley Field was as yet hardly fit for minimum operation and other VHB fields were months away from completion. The decision to concentrate B-29 bases in the Marianas had seemed to promise the best opportunity for an early and sustained offensive against Japanese industry, but determined enemy opposition on Saipan and competition from Navy construction projects had delayed base development.* The unprecedented decision to operate each VHB wing – with its 12,000 men and 180 aircraft – from a single field would save time.3 but in mid-October preparations to receive men and planes were sadly in retard.
Though the B-29’s of XX Bomber Command had been operating from staging fields at Chengtu in China for four months and though much had been done to perfect the Superfort as a weapon, little had been accomplished toward defeating the enemy. The distances involved had proved frustrating even for the VLR planes of a global air force: the distance from Chengtu to appropriate targets; the distance from Washington to command headquarters at Kharagpur; and above all, the distance to Chengtu from Kharagpur, whence supplies for each mission had to be flown in over the long and dangerous Hump route. By comparison, the Marianas offered many advantages. From Saipan, B-29’s could reach all important industrial areas in the home islands, and supplies could be brought in by ship according to a predictable schedule. Yet distance was still formidable even in the Marianas, situated nearly 5,000 miles from San Francisco and more than 1,200 from Tokyo; distance complicated problems inherent in the global command system of the Twentieth Air Force which not even the teleconference could obviate. Headquarters staffs during the move overseas became widely separated: Hansell brought with him only a small advanced echelon, and it took weeks for the whole of XXI Bomber Command Headquarters to complete the trip from Peterson Field, Colorado;4 when Brig. Gen. Emmett O’Donnell arrived at Saipan on 20 October to open the 73rd Bombardment Wing (VH)
* See above, pp. 512-20.
Headquarters, he concluded a transfer begun on 18 July.5 Like headquarters personnel, combat aircraft were strung out between the States and Saipan, and the problem of maintaining a constant flow of B-29’s was to continue.
It was planned that ATC ferry the planes to Saipan from Mather Field, California, with stopovers at John Rodgers Airport near Honolulu and at Kwajalein, on a schedule which called for delivery of 5 B-29’s a day from 20 October until the 73rd’s authorized strength of 180 was reached.6 Planes arrived, however, at the rate of only two or three a day; by 4 November the cumulative shortage at Saipan was seventeen and at Mather Field, thirty. The difficulty lay in the availability of aircraft rather than in the ferrying job – ATC was prepared to deliver seven planes a day, had they been ready for the fly-out.7 Lt. Gen. Millard F. Harmon, deputy commander of the Twentieth Air Force, kept needling Washington, warning that unless the prescribed schedule were maintained, Hansell could not follow operational plans.8 In spite of Harmon’s importunity, the shortage continued: by 15 November the 73rd had only 90 B-29’s and by the 22nd, 2 days before the first mission over Japan, only 118.9
The 738’s personnel had begun to come in earlier. An advanced air echelon of wing headquarters arrived at Saipan on 24 August, the advanced ground echelon on 16 September.10 The ground echelons of the four VHB groups (497th, 498th, 499th, and Sooth) and their constituent squadrons arrived in September, the air echelons in late October and early November.11 The 73rd Wing had been activated on 27 November 1943 and its crews were, by wartime standards, comparatively well trained. But of necessity most instruction had been at the combat-crew level, with little time for unit training. The transition from practice flights in peaceful skies over the prairies of Kansas and Nebraska to combat missions over Japan was at best a hard one for inexperienced crews. They badly needed the intensive unit training program which O’Donnell inaugurated soon after his arrival in Saipan: this included general theater indoctrination, formation flying, rendezvous, communications, and combat missions against targets less vigorously defended than those in the main islands of Japan.12
Preliminary Operation
For the first training mission, Hansell chose as an objective bypassed and oft-bombed Truk. The strike, involving a moderately
long overwater flight and exposure to light Japanese defenses, would help keep the atoll neutralized while the Seventh Air Force’s veteran 11th Group, whose B-24’s had long been engaged in that task, moved from Kwajalein to Gum. The mission was scheduled for 26 October, but on the 25th it appeared that the B-29’s might be called on to help out in the decisive naval engagement then being fought in the Philippine Sea. Hansell canceled the shakedown against Truk and ordered the 73rd to stand by, with twenty B-29’s on two-hour notice.13 Although later there would be more than enough requests for tactical help from the B-29’s, there was none on the 26th; Harmon, whose job it was to coordinate B-29 and theater operations, told Hansell to get on with the Truk mission.14 This maiden effort was but a modest affair, compared with XX Bomber Command’s first strike of ninety-eight Superforts. On the morning of 28 October XXI Bomber Command, still short of planes, could send only 18 B-29’s of the 497th and 498th Groups, each plane loaded with 6,000 pounds of GP bombs and 5,440 gallons of gas.15 Hansell, a crack pilot who had repeatedly led his heavies in missions over Europe, went along only to have his plane become one of four aborts. As it developed, this was Hansell’s last chance to lead a mission, for Brig. Gen. Lauris Norstad, his successor as chief of staff of the Twentieth Air Force, was soon to invoke a regulation forbidding the VHB’s commanding general from flying over enemy territory.16 Fourteen B-29’s got over the Dublon submarine pens at Truk to unload from 25,000 feet with indifferent success: the 497th got about half its bombs in the area, the 498th less than is per cent. As a training mission, however, it was not bad, and the enemy added a touch of realism by firing a few rounds of flak and sending up one Zeke which remained circumspectly out of range.17
Two days later the same groups again sent eighteen bombers against Truk. Bombing results were even less satisfactory than on the 28th. The first formation made a visual tun as planned, but the second, finding the target obscured by clouds, missed completely when the radar on the lead plane malfunctioned. Again enemy reaction was limited to a single interceptor which failed to close and an occasional burst of flak.18 A third strike at Truk was dispatched on 2 November with crews briefed for radar bombing, but the mission served only to emphasize a well-known fact – that the crews still had much to learn. Two of the three squadrons participating mistook thunderheads for
Ruo Island, designated as the initial point, and the visible bomb bursts were so widely scattered as to defy efforts to draw a bomb plot.19
Despite unsatisfactory performances at Truk, O’Donnell decided to send his green crews against Iwo Jima, a more formidable target. Iwo had been hit by carrier planes in a number of strikes in June and July, and since August had been regularly visited by Marianas-based B-24’s of Task Force 59. On 2 November, the day of the third go at Truk, the Japanese had struck back in a bombing and strafing attack on Isley Field; although no damage was done to the B-29’s parked there, the raiders were thought to have staged through Iwo Jima, some 725 miles to the north, and the fear of other attacks added incentive to O’Donnell’s design.*20 Two airfields on Iwo were named as targets, and operational plans were drafted so as to provide experience in daylight visual bombing and to test night landing facilities at Isley. Though this mission, run on 5 November, was considered successful as a training exercise, as at Truk results were less than impressive – strike photos showed only about a fourth of the visible bomb bursts within 1,000 feet of aiming point.21 A follow-up attack by seventeen B-29’s on the 8th miscarried when one squadron had to jettison its bombs in the ocean, and the other dropped its load through a hole in the undercast. On the first Iwo strike a dozen fighters had come up but had not attacked; on the second, five Zekes and three other planes challenged and one flipped a phosphorus bomb onto a Superfort to score the first – but very minor – battle damage suffered by XXI Bomber Command.22
For the sixth and last preliminary training mission, O’Donnell sent crews from the hitherto untried Sooth Group in an Armistice Day strike against the Dublon submarine pens at Truk. The eight planes which completed the mission found good weather over the target, where they bombed visually from 25,000 feet with somewhat more accuracy than had been achieved before and without any damage from the feeble opposition they met. But even on this raid the bombing could be rated no better than “fair.”23
Truk was used by XXI Bomber Command until the end of the war as a practice field for newly arrived combat units, which ran in all some thirty-two practice missions against the atoll. No B-29 was lost: neither the forty guns which made up the AA defense nor the patched-up fighters there could harm the Superfortresses at their conventional
* See below, p. 581.
bombing altitude. There was little left at Truk to destroy, but Japanese officers interrogated after the war credited the B-29’s with preventing their forces from rebuilding air installations destroyed in earlier attacks by Navy and AAF planes, and rated B-29 bombing accuracy as “excellent.”.24 Certainly this was not true of the first training expeditions, either at Truk or at Iwo, but that was of little moment.25 What counted was that some crews from all groups except the 499th had been over an enemy target and had faced enemy opposition without suffering casualties or appreciable damage to aircraft. Crews had learned something that could not be taught from the book. More could have been learned in additional milk-run missions, but in mid-November XXI Bomber Command was behind schedule for its bomber offensive against the Japanese homeland. The next time the Superforts left Isley Field they would be Tokyo-bound.
Target Selection
Tokyo was a natural choice for XXI Bomber Command’s first strategic mission: except for the Doolittle raid in April 1942 the capital city had experienced no air attack, and a successful strike would have important psychological effects within Japan and among the Allied nations. But the choice was also justified by material considerations based on intensive study of Japanese industry. In earlier chapters it has been shown that the MATTERHORN plan for strategic bombardment by XX Bomber Command’s B-29’s staging through fields at Chengtu was guided by a report of the Committee of Operations Analysts issued on 11 November 1943.* Deliberately restricted to economic objectives, this report had listed six profitable target systems: merchant shipping, steel production, antifriction bearings industry, urban industrial areas, aircraft plants, and the electronics industry. Although no internal priorities were given, by strong implication the steel industry was preferred above the others, and since that industry alone offered important targets within range of Chengtu, XX Bomber Command had during the early months of its campaign gone out against steel plants in Kyushu and Manchuria.
By autumn of 1944 the strategic situation in the Pacific had changed greatly. The decision of Allied leaders at the Cairo conference to speed up the war against Japan had already produced important results, and as the offensive gained momentum it had appeared
* See above, pp. 26-28.
that the steel industry, classified as a “long-term” objective, should be passed over for targets offering more immediate results. At Arnold’s request the COA had prepared another report in two parts, based on the alternate assumptions that the war could be won by aerial and naval blockade alone or by those means plus an invasion of the home islands. This report was submitted on 10 October, two days before Hansell landed at Saipan.26 Its recommendations for XX Bomber Command and the resulting changes in target selection have been described in a previous chapter;” the main concern of the COA, however, was with operations of B-29’s based in the Marianas.
Under the first assumption, the air-sea war, the COA recommended attacks against Japanese shipping (to include a comprehensive VLR mining campaign), the aircraft industry, and Japanese urban industrial areas. Mining was to be a continuing program, but after completion of planned attacks against aircraft factories and urban areas, the strategic target program was to be reviewed in search of more profitable objectives – the food supply, for instance, might be vulnerable to air attack. Recommendations under the second assumption were similar, though the order of items and the emphasis were changed: “4 an attack on the aircraft industry and on urban industrial areas and an intensification of the attack on shipping by all available means, including mining by VLR aircraft where operationally feasible.” Although first priority was given to the aircraft industry, the COA believed that, if possible, mining operations should be conducted concurrently. For the area attacks, six cities were named – Tokyo, Yokohama, Kawasaki, Nagoya, Kobe, and Osaka – but the attacks were to be delivered only when they could be done in force and within a short time.27 Because the JCS believed that an invasion of Japan would be necessary, it was the latter set of recommendations which was to guide the Twentieth Air Force in its choice of targets; the lessons of the Combined Bomber Offensive in Europe had had a sobering effect, and no person of authority in the AAF urged the probability of a victory by air power alone. There was no disagreement with the COA opinion that shipping losses had so threatened Japan’s industrial balance that it would be profitable to neglect such long-term objectives as steel in order to cripple those industries geared directly to the military machine. On the score of military importance and of vulnerability, moreover, no industry seemed so attractive an objective as that engaged
* See above, pp. 133-35.
in production of aircraft. This target system, merely listed as one of six in the report of 11 November 1943,was now given first priority.28
The Japanese aircraft industry had got its start just after World War I, using for the most part patents obtained from American, British, and German firms. Total output in 1930 was 445 planes and in 1936 was 1,181. Thereafter the China “incident” and plans for further aggression brought a marked increase in production: in 1941, 5,088 military planes were built. After Pearl Harbor this expansion was stepped up, with production figures mounting to 8,861 in 1942, 16,693 in 1943, and 28,180 in 1944. As in the United States, the mere recital of totals does not tell the whole story of expansion, for planes increased in weight and improved in performance, and the ratio of combat types to trainers was sharply increased.29 From the beginning a few big names dominated the industry: four companies – Nakajima, Mitsubishi, Kawasaki, and Tachikawa – turned out more than two-thirds of all the aircraft built during the war years, and three-fourths of all combat types were produced by the first three of these.30 Concentrated in and around the cities of Tokyo, Nagoya, and Osaka, the aircraft industry was a target system highly susceptible to attack. The Joint Target Group (JTG) at Washington approved in general the recommendations of the COA and emphatically indorsed the priority given the aircraft industry and the delay in urban area attacks (save for one test incendiary mission against a small city district). The JTG believed, however, that the VLR mining campaign should also be delayed, except in tactical support of other operations, until “Japanese airpower is destroyed, or until it is calculated that a blockade of Japan can be imposed, which will more decisively affect the enemy war capability than attacks on other target systems.”31 With this modification in timing, the COA report was to guide the initial campaign of XXI Bomber Command.
Hansell and O’Donnell had called the shots on the training missions; under the peculiar command system of the Twentieth Air Force, the JCS, through Arnold and Norstad, now took over. The first target directive, sent to Hansell on 11 November,32 set the primary mission of the Twentieth Air Force as the destruction of Japanese aircraft engine and assembly plants and major overhaul and repair facilities: XX Bomber Command was to continue the campaign it began in strikes against Okayama and Omura; XXI Bomber Command was to launch
its attacks against the homeland in November. The latter’s primary targets, to be attacked by precision methods, were listed in the following order of priority:
1. Principal Engine Manufacturers:
Mitsubishi Jukogyo, Nagoya Hatsudoki
Nakajima Hikoki, Musashino Seisakusho
Kawasaki Kokuki, Akashi
Nakajima Hikoki, Tama Seisakusho
2. Principal Aircraft Component and Assembly Plants:
Nakajima Hikoki, Ota Seisakusho, Takasaki area
Kawasaki Kokuki, Kagamigahara, Nagoya area
Nakajima Hikoki, Koizumi Seisakusho, Takasaki area
Mitsubishi Jukogyo Kokuki, Nagoya area
Aichi Tokei Denki, Eitoku, Nagoya area
Secondary and last resort targets, all suitable for radar bombing, were listed in two groups: 1) port areas (Osaka, Nagoya, Tokyo, Kawasaki, Yokohama, Shimonoseki, Kure, Hiroshima, Kobe, Nagasaki, Sasebo, and Yokosuka); and 2) urban areas (Hiroshima, Kure, Niigata, Yawata, Tobata, Wakamatsu, Kurasaki, Kokura, Fukuoka, Nagasaki, Omuta, Moji, Kurume, and Nobeoka) .
The secondary mission of the Twentieth Air Force was support of planned Pacific operations, with targets to be selected by the Washington headquarters in coordination with other commands concerned. XXI Bomber Command was to perform all photography necessary to assigned operations and such special photo missions as might be directed. Finally, Norstad indicated there would probably be test incendiary attacks against some of the urban and port areas listed as alternate targets.33 In view of postwar criticisms of target selection in ETO where assembly plants were given preference over aeroengine factories,* it is important to notice the reversal of that policy in this listing of priorities, dictated by the apparent concentration of production in a few factories.
SAN ANTONIO I
To secure the maximum effect from the opening blow of the VHB campaign, Tokyo had been chosen. According to Norstad’s teleconference of 11 November, transmitting the JCS directive, the specific target was Nakajima’s Musashino plant, like all of the Twentieth’s
* See Vol. III, pp. 793-94.
targets described in a special target information sheet. It was located at the edge of a crowded suburb in the northwest part of Tokyo, some ten miles from the Emperor’s Palace.34 In assigning this plant a priority second only to Mitsubishi at Nagoya, the JCS had assumed that Nakajima’s plant at nearby Tama (carrying fourth priority) was a separate establishment; later the two were considered as one target, a judgment confirmed when postwar investigation revealed that the two factories, originally independent, had been merged in October 1943. Estimates that Musashino-Tama (or Musashi, to use the correct name for the merged plants) produced 30 to 40 per cent of all Japanese combat aircraft engines were reasonably close to the true figure of 27 per cent. Nakajima had manufactured a number of air-cooled models purchased from the Wright Aeronautical Company in 1937; by late 1944 the plant was concentrating on army and navy versions of the 1,130-horsepower Ha-35 and the 2,000-horsepower Ha-45. Six major assembly plants were dependent completely or in part on Musashi’s output; its destruction, then, was expected to have early and significant results on delivery of military planes.35
Lamenting the paucity of target intelligence, the COA had recommended in its report of 10 October an early photographic coverage of Japan’s industrial centers by VLR planes.36Although the photo-reconnaissance B-29’s of XX Bomber Command were already engaged in that task, the most important areas lay beyond their range, and the task fell to Hansell’s 3rd Photo Reconnaissance Squadron, 2 of whose F-13A’s arrived at Saipan on 30 October after a 33-hour, 2-stop flight from Mather Field. At 0550 on 1 November a crew headed by Capt. Ralph D. Steakley took off for Japan. His F-13, first U.S. plane over Tokyo since April 1942, droned above the city at 32,000 feet in clear weather, its cameras recording long-hidden industrial secrets. After nearly fourteen hours in the air, the plane returned unscratched. The crew received well-merited decorations, and the command got needed negatives from which the squadron’s understaffed photo-lab unit hurriedly turned out 7,000 prints.37 Reconnaissance missions continued as more F-13’s trickled in; one was lost in a mission to Nagoya on 21 November, but there were nine on hand at the end of the month. Before the first strike at Tokyo, on 24 November, the 3rd P/R Squadron flew seventeen single-plane missions over the home islands, of which eight were for weather reconnaissance. When enemy opposition permitted, an F-13 could stay over the target for about an
hour on missions with a 1,500-mile radius, and neither fighters nor flak were effective: of 100 fighters airborne on 7 November, for instance, only 2 got within 1,000 yards of the high- and fast-flying F-13’s; the heavy flak encountered on every mission did no damage.38 Weather was a more formidable opponent, however, and a third of all photo missions in November and December were thwarted by undercast and another third partially spoiled. Yet by 11 November Hansell felt that he had adequate coverage of the most important aircraft plants in the Tokyo area, as well as naval and harbor installations at Tokyo, Yokohama, Yokosuka, and Kawasaki.39
Meanwhile, on 30 October Hansell had submitted to Arnold his operational plan for the first strike at Tokyo, coded SAN ANTONIO I. This called for daylight, visual bombing from 30,000 feet by 10 to 12 squadrons of 9 to 11 planes, each carrying 5,000 pounds of bombs (30 per cent M76 incendiaries and 70 per cent 500-pound GP’s) and 8,070 gallons of gas.40 Enemy opposition could not be accurately forecast. Washington had estimated first-line fighter strength in Japan at 1,114 on 12 October, at 608 on 2 November; the 73rd’s field orders indicated that 400 to 500 fighters in the Tokyo-Nagoya area could be sent up against the B-29’s.41 These figures were all too high; the Japanese order of battle for the home islands in November included only 375 fighters.* Preliminary analysis of photos had revealed at least 150 heavy AA guns between Tokyo and Funabashi, so that intense and accurate fire was predicted for the Tokyo and Tokyo Bay areas, on the Chiba peninsula, and along Sagami Bay.42 Information concerning the enemy’s radar was vague. After some debate between Hansell and Harmon over countermeasures, Norstad decided that XXI Bomber Command should send out prestrike ferret missions as well as a D-day diversion to Nagoya by “window”-dropping F-13’s; jamming by RCM transmitters on the B-29’s was to be deferred until radar operators were better trained.43 Air-sea rescue precautions were elaborate: for SAN ANTONIO I, 5 lifeguard submarines, assigned by the Commander, Submarines, Pacific (ComSubPac) and under operational control of Task Group 17.7, were stationed between Honshu and Iwo Jima; one destroyer was to patrol south of Iwo and another was to be dispatched for duty in the area 100 to 150 miles north of Saipan; 1 PBM search plane was to be on station just south of Iwo, while 3 PBM’s and 6 PB2Y’s were held on alert at Saipan.44
* See above, p. 172.
Selection of a D-day involved much discussion. Both Washington and Hansell wanted an early date, but the latter was hesitant to send fewer than 100 B-29’s over Tokyo, and the slow deliveries from Mather Field put a brake on his plans.45 On 16 October, after talking with Hansell at Saipan, Harmon told Arnold that SAN ANTONIO could be run on or about 10 November.46 Shortly thereafter, Harmon, Hansell, and a representative of Nimitz’ headquarters worked up a plan to coordinate the B-29 mission with a carrier strike (HOTFOOT).47 Beginning on 12 November, the Navy planes were to hit the Tokyo area, and by the 17th when the B-29’s would arrive, Japanese fighter defense should have been badly hurt. Hansell liked this idea and timing, which would help him get his minimum of 100 B-29’s forSAN ANTONIO; though Norstad was less enthusiastic, he concurred on 28 October.48 Four days later Arnold qualified this approval, informing Harmon that the double strike was on only if a firm date could be set for HOTFOOT; otherwise XXI Bomber Command was to go on alone when enough B-29’s were on hand.49 Nimitz, his carriers heavily committed in the Philippines, was forced to put off the strike and could not underwrite any early date.50 Hansell still wanted more time and the carrier help, and Harmon was willing to maintain daily contact with CINCPOA in hopes of getting a D-day for HOTFOOT which could coordinate with the 20 November deadline for the B-29 mission.51 On the 12th Nimitz indefinitely postponed HOTFOOT, and SAN ANTONIO was set up as an independent strike for the 17th.52
Long before dawn on that day Isley Field was alive with preparations. The plan was a widely spread secret on Saipan, and vehicles bearing personnel of all ranks poured into the great base. Twenty-four war correspondents, representing every important news outlet in the United States, were on hand to give the command its first spate of publicity. Movie cameramen got set to catch the Superforts as they rolled down the runway and became airborne. “Rosey” O’Donnell climbed into his B-29 under a barrage of photo-flash bulbs. Everything was set but the weather. That element, which was to prove the worst hindrance to the B-29 campaign against Japan, was giving an ominous preview of things to come. In the long run, target weather was to cause the most trouble, but on the morning of 17 November it was base weather that held up the show. Most unusually and perversely, the prevailing easterly wind had veered around into the
southwest, so that the customary take-off run would have to be reversed. At Isley this involved an uphill pull, hazardous at any time for combat-loaded B-29’s, and suicidal in the steady rain which was falling as H-hour approached. A wing operations jeep notified airplane commanders that take-off would be delayed one hour, but after sixty minutes the rain showed no signs of abating. The mission was postponed for twenty-four hours.53
On the morning of the 18th weather forced another day’s delay. This schedule-and-scratch routine became SOP through a week of frustration during which nerves were frayed to the breaking point. In Washington, Arnold and his staff grew impatient; so did the men who had worked through Saipan’s summer heat and autumn rains to build the base. The combat crews, their enthusiasm dulled by repeated postponements, griped that the B-29 was the best plane that never left the ground. For a long week the clouds hung low over Isley. Then the wind veered again, and the 24th dawned clear. This was it.54
At 0615 the first plane rolled down the strip: it was Dauntless Dotty, with O’Donnell at the wheel and Maj. Robert K. Morgan, erstwhile pilot of the famed Memphis Belle, in the co-pilot’s seat. The great silver plane used every inch of the black-topped runway and a short stretch of the coral extension before pulling up; then, skimming the water beyond, it passed out of sight.55 The long delay had built up Hansell’s force, and Dauntless Dotty was followed by 110 other B-29’s, carrying in all 277.5 tons of bombs.56 Hansell had originally planned two routes outbound, straddling the Bonins and so plotted as to get the four combat groups over the target simultaneously in a converging attack from east and west.57 With typhoon conditions between Saipan and Iwo and with weather planes reporting the storm moving northeastward, Hansell instead sent all his planes up the western track, briefed to attack along a west-to-east axis.58 To fit these changed tactics, Hansell switched the diversionary mission from Nagoya to the Tokyo area. The F-13’s were to enter the Japanese radar screen from the southeast as the first B-29’s went in from the southwest; after dispensing “rope” the F-13’s were to photograph target damage by the early groups, and a lone F-13 was to follow the last combat group to get final photo coverage.59
En route to Tokyo seventeen B-29’s aborted. Six Superforts were unable to bomb because of mechanical failures, and the weather over Tokyo made bombing difficult for the others. Formations flying at
altitudes of from 27,000 to 33,000 feet were swept into a 120-knot wind which gave the bombers a ground speed of about 445 miles per hour;60 below, an undercast almost completely obscured the target. Only twenty-four planes bombed the Musashino plant; sixty-four unloaded on dock and urban areas. Thirty-five of the aircraft that bombed had to do so by radar. The Japanese fighter defense was less fierce than had been feared and much less effective than that which had been met by AAF formations over Germany. Intelligence officers, consolidating crew reports, figured that about 125 Jap fighters had been up – a mixture of Tojos, Zekes, Tonys, Nicks, Irvings, and some unidentified planes – of which the B-29 gunners claimed 7 destroyed, 18 probables, and 9 damaged. As usual in Japanese interceptions, there seemed to be no coordinated plan of attack, and pilots varied in skill, aggressiveness, and tactics used. The one U.S. combat loss occurred when an enemy pilot drove his damaged Tony into the tail of a B-29 in what looked like a deliberate ramming; with elevator and right horizontal stabilizer shorn off, the Superfort crashed into the sea twenty miles off the Honshu coast with the loss of all aboard. Flak was meager to moderate, and generally inaccurate.
The planes came back in formation, as planned, to a point opposite Iwo Jima and thence on home individually. Air-sea rescue precautions paid off when the whole crew of one B-29 was saved after running out of fuel and ditching The others came straggling in, landing between 1926 and 2259 hours. Because of congestion at Isley, two groups went on to Guam, returning to Saipan next day. The total cost was not great by standards used for unescorted bomber missions in the ETO: two B-29’s destroyed, eight damaged by enemy action and three by accidental hits from B-29 guns; one man killed, eleven missing, and four injured. But bombing results of the mission were not encouraging. The F-13’s, like the bombers, were hindered by clouds and their strike photos showed only sixteen bomb bursts in the target area.61Actually, the bombing was somewhat better than this incomplete coverage indicated, but still not good. A bomb plot prepared by the Musashi management and known to Americans only after the war showed 48 bombs (including 3 duds) in the factory area; 1 percent of the building area and 2.4 per cent of the machinery were damaged, and casualties included 57 killed and 75 injured.62The intangible results were more important. The XXI Bomber Command had struck the toughest target area in Japan under bad weather conditions
without excessive losses.63And the appearance over Tokyo of the B-29’s brought home to the Japanese people, in more compelling fashion than the earlier strikes at outlying cities by XX Bomber Command, the impotence of the Japanese air forces to cope with the strategic bombers. This was the beginning of the disillusionment that made the government’s propaganda progressively less effective.64
High-Altitude Precision Bombing
SAN ANTONIO I set the pattern for the next three months of operations by XXI Bomber Command. There were deviations – more strikes at Iwo Jima’s airfields, a night area attack on Tokyo, experimental incendiary raids on urban areas at Tokyo, Nagoya, and Kobe – but until 9 March the command was concerned primarily with daylight, high-altitude, precision attacks delivered against aircraft factories in Japan according to orthodox AAF doctrines.
Because of the slight damage wrought on 24 November, Hansell decided to send his planes back to Musashino in a second major effort (SAN ANTONIO II), the operational plan substantially the same as on the first strike, except for a change in the method of withdrawal. To exploit any weakness in the enemy’s defenses that had resulted from the first mission, he set an early D-day, 27 November.65 The 73rd Wing had by now 119 Superforts but only 87 were scheduled for the mission. Actually, eighty-one were airborne on the 27th; nineteen of these aborted and the others found the target completely hidden by 10/10 cloud. As briefed, they dropped by radar on the secondary targets, dock and urban areas at Tokyo, and on Hamamatsu, Shizuoka, Numazu, and Osaka. One B-29 was lost with its entire crew, when it ditched on the return trip.66 Although the F-13’s were unable to take strike photos, it was later realized that Musashino was intact in spite of the 192 sorties sent against it, and there was little hope that any important damage had been done elsewhere. The two missions, though not destructive, provoked angry reaction from the Japanese. Staging down from Iwo Jima, they heckled Isley Field in small but effective raids: in two attacks they destroyed four B-29’s, seriously damaged six, and inflicted lesser damage to twenty-two others. Hansell was worried enough to disperse his force by sending some of his B-29’s from crowded Isley to safer Guam, and to plan joint air-sea strikes at Iwo in which the Superforts might help neutralize the staging fields.67
Meanwhile, Hansell informed Washington of his intentions: maximum strikes against a top-priority target when weather would permit visual bombing and he had sixty planes ready to go; raids against secondary targets with thirty or more planes when weather prevented precision bombing; and nightly weather-strike missions.68 He wanted to improve on SAN ANTONIO II with a big daylight show within the next two days, but the weather forecasts were pessimistic and there were not enough B-29’s in commission to meet his minimum standard. To keep up the pressure on the enemy – and in part to relieve congestion at Isley, which on the record was a more dangerous location for a B-29 than the sky over Honshu – he ran a night radar mission of some thirty planes on 29 November. The target was the dock and industrial areas in Tokyo and again the results were negligible.69
For 3 December the command scheduled a maximum daylight mission against the Nakajima Aircraft Plant at Ota, forty miles northwest of Tokyo.70 By D minus 1 weather reports were forbidding: at bombing altitudes over Ota, winds were reaching velocities of 180 miles per hour or more. At 0130 on the 3rd it was decided that the only hope for the day was to go back to Musashino where visible bombing might be possible.71 Crews had already been briefed twice for the target; the 73rd Wing hurriedly cut field orders and by 0945 eighty-six bombers were heading for Tokyo. Seventy-six got over the city to find clear weather but high winds; 59 planes bombed visually from a mean altitude of 28,700 feet with poor results.72 Musashi’s records indicate that twenty-six bombs fell in the plant area with some small damage to buildings and equipment and almost none to machinery; Japanese casualties were moderately high. Strike photos, the command’s only source of information, seemed to show even less damage, and for these slight results the command had paid dearly, with six B-29’s lost and six damaged.73
It was ten days before XXI Bomber Command went back to Japan in force, though the B-29’s did participate in a joint Army-Navy attack against Iwo Jima on 8 December.* On 6 December Arnold’s headquarters issued a new Air Estimate and Plans for Twentieth Air Force operations,74 but one which differed little from the similar paper drawn up in November either in concept or in target priorities. MacArthur’s campaign in the Philippines was behind schedule, and
* See below, p. 584.
the Twentieth was obligated to support Pacific operations-specifically, Third Fleet operations in advance of the Mindoro invasion set for 15 December.75 According to accepted AAF doctrine, this could best be done by attacking an important industrial objective. The choice for primary visual target was the Mitsubishi Aircraft Engine Works at Nagoya, and the same company’s aircraft works was named as radar target (rather than secondary, as Norstad had suggested) ; strays, it was hoped, would spill into crowded Nagoya, Japan’s second city and an industrial center of great importance.76 The engine works, still in top priority for XXI Bomber Command,77 lay in the northeast section of Nagoya, about two and a half miles from Nagoya Castle. The plant was considered by the JTG as a single target,78though it actually consisted of three separate but closely related units of the vast complex comprising the Mitsubishi Heavy Industries, Ltd.: 1) the No. 2 Engine Works, responsible for research, design, and manufacture of prototype engines; 2) the No. 4 Engine Works, which between 1939 and 1945 manufactured 44,004 engines, the most important model being the Ha-102, a 1,000-horsepower motor used on the Nick and Dinah 2; and 3) the No. 10 Engine Works, which furnished castings and forgings for all Mitsubishi engine plants.79
On the 13th, the 73rd Wing was able to get ninety bombers up, most of them carrying ten 500-pound GP’s but one squadron from each group loaded with incendiary clusters. As on previous missions, a number of planes failed to reach the primary target: sixteen B-29’s aborted and three bombed targets of opportunity. Japanese resistance was lively and, in all, four B-29’s were lost, thirty-one damaged.80 The bombing, if of less than pickle-barrel precision, showed improvement. Strike photos indicated that 16 per cent of the bombs dropped had fallen within 1,000 feet of the aiming point and that 17.8 per cent of the roofed area had been destroyed.81 Although this in itself was encouraging, had intelligence officers been able to read from their photos the whole story, there would have been even more optimism on Saipan. At the No. 4 Engine Works an assembly shop and 7 auxiliary buildings were destroyed, and an assembly shop and 11 buildings were damaged; at the No. 2 Engine Works a prototype engine-manufacturing shop and 2 other shops were damaged; and personnel losses ran to 246 killed and 105 injured. For the first time XXI Bomber Command had made an appreciable dent in the aircraft industry. Plant officials calculated that the attack reduced productive
capacity from 1,600 to 1,200 engines per month; after 13 December parts were no longer machined at No. 4, and engine production was limited to assembling parts on hand and those received from other plants. Mitsubishi officials had been considering the advisability of dispersing the Nagoya facilities ever since the U.S. conquest of Saipan. After the strike of 13 December the transfer of equipment to underground sites began, but even at the end of the war the movement had not progressed far enough to allow production in the new plants.82
Nagoya had only a brief respite. On 18 December Hansell dispatched eighty-nine bombers against the Mitsubishi Aircraft Works, the giant assembly plant which used most of the engines produced in the No. 4 Engine Works. Located on reclaimed land at the northeast corner of Nagoya harbor,83 it was, like the engine works, composed of three integrated plants: 1) the No. 1 Airframe Works for research and experimental engineering; 2) the No. 3 Airframe Works, which built navy planes – Zeke and Jack fighters and Betty bombers; and 3) the No. 5 Airframe Works, which manufactured bombers and reconnaissance and transport planes for the army. Large, compact, and conspicuous, this complex offered an excellent visual target, and the proximity of the harbor’s shore line made it suitable for radar strikes as well.84 All of this was fortunate-or should have been. On this 18 December attack many planes, as usual, failed to follow the flight plan so that only sixty-three planes bombed the primary target. Cloud cover was heavy and forty-four of these dropped by radar, to add considerably to the damage caused by an earthquake on 7 December. Though few bombs were plotted in the area, 17.8 per cent of the roofed area appeared to have been destroyed.85 The No. 3 Works suffered extensive damage to the sheet-metal, heat-treatment, fuselage-assembly, and final-assembly shops, and at No. 5, approximately 50 per cent of the total assembly area was damaged. Casualties, in dead and injured, amounted to 464. In spite of the damage, however, production loss amounted to only about ten days’ work.86
Again on 22 December XXI Bomber Command went back to Nagoya in a repeat attack on Mitsubishi’s engine works. This time the mission was planned as a daylight incendiary mission, with each B-29 carrying 2.75 tons of M76’s and no high explosives.87 The change in tactics resulted from recent correspondence between Washington and Saipan which seems significant in the light of later changes in the command setup. Convinced from an early date that Japanese cities
were highly inflammable.88 Arnold’s headquarters had suggested in the target directive of 11 November that test incendiary raids be conducted before running the mass attacks on six selected cities.* A small raid against Tokyo on the night of 29/30 November had been unsuccessful. On other missions some incendiaries had been used, but Washington seemed to think too much reliance was being placed on HE bombs.89
On 18 December, just after LeMay’s XX Bomber Command had burned the heart out of Hankow’s military storage area with fire bombs.† Norstad requested a full-scale incendiary attack on Nagoya as soon as 100 B-29’s were ready to go – this as an “urgent requirement” for planning purposes.90 Hansell protested strongly: he had “with great difficulty implanted the principle that our mission is the destruction of primary targets by sustained and determined attacks using precision bombing methods both visual and radar”; now as this doctrine was “beginning to get results” on the aircraft industry, pressure to divert his force to area bombing threatened to undermine the progress made. Hansell would, however, consider the message as an order modifying his original directive.91 Norstad replied immediately in a conciliatory message: the aircraft industry continued to carry an “overriding priority” and the test fire raid, he reiterated, was simply a “special requirement resulting from the necessity of future planning.”92 Mollified, Hansell promised to run the desired mission as soon as possible after completing missions already scheduled.93
This reply went out on 21 December. The Nagoya mission on the next day, though using only incendiaries, was not in fulfillment of Norstad’s request; it involved only 78 bombers dispatched instead of 100 and it was planned as a daylight precision attack. The weather turned bad, however, and before the last formations were over Nagoya the target was covered by 10/10 cloud. Only forty-eight planes bombed the Mitsubishi plant and they had to drop by radar; strike photos were few and revealed little.94 Actually there was not much damage to reveal: 252 fire bombs fell in the area of the No. 4 Works, damaging a few buildings but hurting no machine tools and causing no loss to production.95
The last mission Hansell had on his December docket was a return trip to the first homeland target, Nakajima-Musashino near Tokyo, on 27 December. By any reasonable standards the attack was a failure.
* See above, p. 554.
† See above, pp. 142-44.
Only 54 per cent of the B-29’s participating – thirty-nine out of seventy-two – bombed Musashino, and the twenty-six bombs in the area plotted by Japanese observers did little damage other than setting fire to a hospital.96 This unhappy accident may have given point to a favorite Japanese propaganda line – that the American devils were chiefly interested in destroying hospitals, schools, and private homes. And the fact that most of the small damage done was by incendiaries rather than by the high explosives dropped was not a decisive argument in the current debate over tactics.
Norstad’s test mission was run on 3 January, when ninety-seven B-29’s got off for Nagoya. Each plane carried a mixed load of bombs – 14 x 350-pound M18 IB clusters fuzed to open at 8,000 feet and one 420-pound fragmentation cluster fuzed to open 1,000 feet below releasing altitude, What with aborts and planes straying from course, only fifty-seven bombed the urban area designated as primary target, most of them releasing visually though cloud cover was rated as 6/10. Some fires were started but there was no holocaust. Smoke rising to 20,000 feet combined with cloud to make observation of results impossible for the attackers.97 As a test, then, the mission was inconclusive. To the citizens of Nagoya, who were better informed than intelligence officers of XXI Bomber Command, the damage seemed slight. For the Japanese, it was unfortunate that they formed, on the basis of this ineffective raid, grossly exaggerated ideas of the efficiency of their fire-prevention system.98
Having satisfied Norstad’s requirement, the command returned to its program of precision bombing against aircraft factories. Performance for the most part was of a piece with what had gone before. On 9 January 1945 seventy-two B-29’s were sent against Musashino near Tokyo. High winds broke up the formations so that only eighteen planes were able to bomb the target; twenty-four bombs, widely scattered in the plant area, destroyed one warehouse and damaged two others-a slight return for the effort expended and the six B-29’s lost.99 At the Mitsubishi Aircraft Works at Nagoya on the 14th, precision bombing was again less than precise. Seventy-three B-29’s were airborne and forty bombed, getting four GP’s – one ton – into the No. 5 Works area and damaging three buildings.100
An attack on 19 January was a welcome interlude in this litany of failure. The target was virgin, a plant of the Kawasaki Aircraft Industries Company located two miles northwest of Akashi, a village
on the Inland Sea some twelve miles west of Kobe. The Akashi works housed Kawasaki’s general headquarters and one of the company’s two large production units, which built the twin-engine fighters Nick and Randy and engines for Tony, Oscar, and Frank fighters. Smaller than Nakajima and Mitsubishi, Kawasaki in 1944 delivered 17 per cent of Japan’s combat airframes and 12 per cent of its combat engines.101
Against Akashi, Hansell sent seventy-seven B-29’s, plus three others in a diversionary strike. With good weather, 62 bulled it through to the Kawasaki factory, dumped 155 tons of GP’s, and then returned with no losses. Interpreting strike photos, intelligence officers estimated that 38 per cent of the roofed area showed major damage.102 This was an understatement. Every important building in both the engine and airframe branches had been hit and production was cut by 90 per cent. Indeed, the Kawasaki Company liquidated the combined plant and dispersed the machine tools, which had suffered only slightly, to other sites. The Akashi shops were given temporary repairs at the cost of 226 tons of critical materials and over 9,000,000 yen, but the installation was used thereafter only for limited assembly jobs.103 It is a pity that the full results of this mission could not have been known to Hansell. His first completely successful B-29 attack, Akashi seemed to epitomize the doctrines of precision bombardment he had championed – and it was his last strike of the war. On the next day he was succeeded as head of XXI Bomber Command by Maj. Gen. Curtis E. LeMay, who was transferred from his similar post with the XX in the CBI.
The relief of a commanding general during a lagging campaign is seldom a pleasant affair for the parties concerned; often there are factors involved too delicate to commit to writing even in a top-secret “eyes-only” message. In this instance there were personal relations that must have made the decision a hard one for Arnold: Hansell had been for a while his top planner and something of a protégé, and Arnold was not without a streak of sentiment for his “boys.” Nonetheless, he seems to have made up his mind during December to replace Hansell. On the 7th Norstad wrote Hansell, apropos of the delays in mounting SAN ANTONIO I:
I knew you would worry about the Chief’s feelings at that time since you know him well enough to realize he would be very much keyed up until the first show was over. He was impatient, but his impatience was directed against
the circumstances and not against you. You were not “on the pan” at any time. I think I can best illustrate his attitude by telling you his reaction to the fourth and fifth postponement. After he had indicated that he was disturbed, I made a statement to the effect that I didn’t think it a good thing to put the heat on you under the circumstances. He replied, “Who said anything about putting the heat on Possum?” in a rather irritated manner.104
But by the turn of the year the heat was on. Arnold sent Norstad to the Marianas to break the news and ordered LeMay to go there for a conference.105 Norstad arrived at Guam on 6 January and told Hansell of his impending relief;106 LeMay flew in from Chengtu next day.107 Although it is likely that no formal record was kept of the conference, one might guess that it involved a certain amount of embarrassment for all.108 Three young generals – Hansell, the eldest, was forty-one – were arranging for a turnover in what was the most coveted operational command job in the AAF. Norstad was Hansell’s friend, had worked with him in the same office, and had succeeded him as chief of staff in the Twentieth; LeMay had served as a group commander in Hansell’s heavy bombardment wing in the United Kingdom in 1943. But whatever personal feelings may have been involved, the business was soon settled. LeMay flew back to Kharagpur, taking with him Brig. Gen. Roger M. Ramey, Hansell’s chief of staff, to head up XX Bomber Cornmand.109 Returning to Guam with a small group of staff officers, LeMay assumed command of XXI Bomber Command on 20 January.110
Before leaving for the States – where by his own request he was to take over a minor job in the B-29 training program – Hansell rendered an account of his stewardship in a ten-page letter to Arnold.111 Earlier Hansell had admitted dissatisfaction with the performance of his command,112 and he was still far from content though he cited statistics to show that in comparison with the 58th Wing in the CBI the record of the 73rd “doesn’t look too bad.” He listed four major problems that had confronted him: 1) converting the 73rd Wing from a preference for radar night bombing to a belief in precision bombing; t) improving bombing accuracy, which was “deplorable”; 3) reducing the abortive rate, which had reached 21 per cent of sorties; 4) reducing the number of aircraft ditching and improving air-sea rescue. In each case he thought that remedial actions were already taking effect. His main fault, Hansell felt, had been in driving his crews too hard; in the absence of depot facilities and adequate maintenance this had resulted in excessive aborts and losses at sea.
Hansell’s letter, actually an answer to criticisms explicit or implied, sheds some light on the reasons for his relief. He was not responsible for the unexpectedly bad weather, the slow build-up of forces, or the retarded development of installations, and another man might have done no better under the circumstances. But XXI Bomber Command had not got the expected results and Arnold was not a patient man. Perhaps in the last analysis Hansell’s chief fault was in adhering too strictly to the “book” – to doctrines of precision bombardment which he had helped formulate – in the face of a growing interest in area incendiary bombing evinced by Arnold’s headquarters. Whatever the cause, out went Hansell, the brilliant planner, and in came LeMay, widely recognized as a driving operator. LeMay had gone to Kharagpur as a trouble shooter when XX Bomber Command’s operation had lagged; now, with that command withdrawing from its Chengtu base; his job in CBI was washed up and he was coming to XXI Bomber Command in a similar role. He was to find operational conditions in the Marianas, in spite of obvious difficulties, immeasurably better than in CBI, and he lost little time in exploiting the potentials of his new command.
LeMay’s first two missions, however, showed little variation from the familiar pattern. On the 23 January mission against the Mitsubishi engine plant at Nagoya, 9/10 cloud so obscured the city that only 28 out of 73 planes bombed, getting 4 GP’s and 144 incendiaries in the target area and causing some damage to an assembly shop, a prototype shop, and an office building.113Four days later a planned attack on Musashino was completely spoiled by clouds and high winds over Tokyo.114 On 28 January LeMay suggested turning from Tokyo and Nagoya to targets less hotly defended; specifically, he recommended the Mitsubishi Aircraft Works at Tamashima.115 Norstad replied that LeMay had “fullest latitude” in mixing his blows so as to disperse enemy defenders, even to the extent of hitting lower-priority targets when “tactical consideration” warranted. Tamashha was so unimportant, however, that it was dubious that the Japanese would make any significant changes in fighter deployment for its defense. Norstad thought an incendiary attack on Kobe would be more fruitful; it would furnish information lacking after the inconclusive test raid on Nagoya and if successful might cause the enemy to thin out his fighter defenses.116 Accordingly, LeMay scheduled his first February mission as an incendiary attack on the port and
built-up areas of Kobe, which, with a prewar population of 1,001,200, was Japan’s sixth largest city and most important port. Its yards housed the Empire’s largest concentration of shipbuilding and marine engine capacity; it was an important rail junction and its key industrial plants – steel, railway equipment, machinery, rubber, and ordnance – were closely integrated into the city’s transportation system. Though of comparatively recent growth and hence more modern in construction than many of Japan’s cities, Kobe’s congested business and factory districts and adjacent residential areas were considered highly vulnerable to incendiaries. For the test attack, the target would be the core of the city where the population averaged about 100,000 per square mile.117
For the first time XXI Bomber Command was to send planes from two wings against the home islands. Early planning had called for the deployment of three wings in the Marianas; later, after long debate,* the total was set at five, the two extra wings being diverted from contemplated deployment in the Philippines.118 The lag between plans and execution, however, was considerable. The 73rd Wing was still short of authorized strength in aircraft – its daily average for January was 137 B-29’s on hand, and at the end of the month there were only 157 out of the promised 180.119 Aircraft and crews of the 313th Bombardment Wing (VH), commanded by Brig. Gen. John H. Davies, began to arrive at North Field, Tinian, on 27 December; two groups, the 504th and 505th’ were on hand by 1 January, but the 6th and 7th were not at station until 28 February. At the end of January the 313th had 122 B-29’s.120 Hansell had found the 504th and 505th so deficient in unit training that he started them off in a training program, designed to last four or five weeks and including a thirty-three-plane mission to Truk. By early February it seemed that the groups could be sent to one of the less formidable homeland targets.121
On the basis of favorable weather forecasts, LeMay decided on 3 February to run the mission next day. Not satisfied with the concentration achieved with the M69 bombs used in the Nagoya test, he loaded his planes with E28 500-pound incendiary clusters topped off with frag clusters. Including 38 from the 313th Wing, 129 planes were airborne, but only 69 got through to the target where they dropped 159.2 tons of incendiaries and 13.6 tons of frags from altitudes ranging between 24,500 and 27,000 feet. About 200 enemy
See above, p. 523.
fighters attacked, proving that Kobe was not a soft touch. They shot down one B-29 and damaged thirty-five; another burned upon landing at Saipan. The results, however, were far more encouraging than at Nagoya, for poststrike photos showed damage to 2,651,000 square feet of built-up area.122 Postwar information, agreeing roughly with this estimate, added details: in the area bombed – the industrial southwestern district of Kobe – 1,039 buildings were destroyed or seriously damaged, and although casualties were only moderate, 4,350 persons were rendered homeless. Local war production was hit hard. Of a dozen factories accounting for 90 per cent of Kobe’s essential war industry, five received damage of varying degrees of severity. One of the two major shipyards had to reduce operations by half. Production of fabric and synthetic rubber was completely wiped out and other industries suffered greatly.123
After Kobe, XXI Bomber Command returned to precision attacks. For a while, most of the missions were coordinated with the amphibious assault on Iwo Jima (DETACHMENT),* but support of that operation was incidental to the primary mission of destroying the aircraft industry. Target for the first of the post-Kobe mainland attacks was Nakajima’s Ota plant, given highest priority among the assembly plants in the 11 November directive. Ota’s importance stemmed from the fact that it was concentrating on the manufacture of a very effective fighter, the Ki-84, called Frank by the Americans. The plant had reached a production peak of 300 planes in December 1944, and although the output had declined to a rate of less than 100 per month by February, this fact was unknown to U.S. intelligence officers.124 The one attack which had previously been scheduled for Ota (on 3 December) had been canceled because of weather.† Forecasts for 10 February, however, were favorable, and on that morning 118 planes took off, loaded with 500-pound GP’s and M76 incendiaries in a weight ratio of 4 to 1. Weather over Ota was even better than predicted and eighty-four planes bombed the Nakajima plant. Bombing accuracy, however, was not impressive: only seven incendiaries and ninety-seven GPs (of which forty-three were duds!) fell in the factory area. Nevertheless, eleven of the plant’s thirty-seven buildings were damaged and seventy-four Franks were destroyed. The fact that most of this damage was done by the few incendiaries that fell in the
See below, pp. 589-90.
† See above, p. 561.
target area lends some credence to the opinion expressed after the war by Ota officials that a heavier concentration of fire bombs would have destroyed the plant. Heavy losses – twelve B-29’s lost and twenty-nine damaged – reflected the increasing effectiveness of Japanese defenses.125
Both at Washington and Guam the AAF had showed a disinclination to divert the B-29’s to tactical support of ground or sea operations – for example, Arnold and Hansell had resisted MacArthur’s efforts to have XXI Bomber Command strike Okinawa airfields to aid his Luzon campaign. The campaign for Iwo Jima, however, offered a unique opportunity in that B-29 attacks against homeland industrial targets could be considered as indirect support for the amphibious assault.* By agreement with CINCPOA, LeMay scheduled a mission against Tokyo for 15 February, D minus 4 at Iwo. Since weather forecasts ruled out Tokyo but were favorable to Nagoya, the command sent 117 B-29’s back to visit Mitsubishi’s engine works at that city. On the way, they hit an unexpected cold front that broke up the formations, so that only thirty-three planes bombed the primary target; they caused a fair amount of superficial damage, particularly in the as yet untouched No. 10 Works, but did not greatly affect production. Most of the others unloaded on Hamamatsu, apparently with considerable effect.126
On 19 February, D-day at Iwo, LeMay directed his planes to the familiar Musashino target, hoping to discourage air reinforcements for the beleaguered island. Two days before, the Navy had staged a carrier attack in which low-flying planes, using small bombs and rockets, had done substantial damage at Musashino.127 But on the 19th, weather again baffled the B-29’s; frontal conditions made the trip out difficult and thick clouds hid Musashino completely. Crews had been briefed to hit Tokyo port and urban areas as a secondary target, and 119 B-29’s (out of 150 dispatched) dropped on those districts by radar. Visible damage covered a total area of 102,600 square feet and included 2 important objectives, a spinning mill and the Sumida River railroad yard and bridge.128 The mission merely added to the cumulative evidence that precision bombing under existing conditions was not paying dividends. Though still unwilling to launch the all-out incendiary attacks against major cities until the B-29 force had been built up, Washington wanted more experimentation with fire bombs. On 12 February Norstad reminded LeMay of the inconclusive results obtained
* See below, pp. 589-90.
in two test incendiary missions and directed another “major incendiary attack” to secure more accurate planning data. Norstad recommended Nagoya as target and accorded the mission a priority second only to the two top-billed engine plants.129
The interest in area fire bombing was also shown in a new target directive issued on 19 February. Although the primary mission remained the “destruction of Japanese engine aircraft plants” (Nakajima-Musashino in Tokyo, Mitsubishi-Hatsudoki in Nagoya, Aichi in Nagoya, and Mitsubishi’s No. 6 Engine Plant in Shizuoka – in that order), for “secondary visual attack or for diversionary reasons,” missions were to be directed 1) against selected urban areas in incendiary tests as directed, and 2) against the principal aircraft assembly plants. If radar conditions prevailed, primary targets were to be urban areas in Nagoya, Osaka, Kawasaki, and Tokyo, in that order. In fulfilling its other mission-support of Pacific operations – the command was to attack only its listed targets unless specifically directed by Washington.130 This new target directive thus included one important revision – it elevated “test” incendiary raids to a priority higher than assembly plants. And this was a clue to the nature of future operations.
LeMay was already committed to a mission in support of DETACHMENT on D plus 4 or 5, a maximum strike in either the Tokyo or the Nagoya area. In either case, the target would be an engine plant – Musashino if weather permitted a visual strike, or Mitsubishi Hatsudoki if radar conditions prevailed.131 As the target date approached, however, forecasts for cloud over Tokyo and high winds over Nagoya made either alternative unattractive. LeMay then decided to hit Tokyo on the 25th with a maximum fire-bomb mission, using one 500-pound GP in each B-29 for spotting purposes and filling up to capacity with E46 incendiaries.132
The number of planes dispatched per mission had increased during the last month as more B-29’s came in and as maintenance facilities improved. The mission of 25 February was by far the largest yet sent out by XXI Bomber Command, with 231 Superforts airborne. This increase in force was made possible by the participation of planes from the newly arrived 314th Bombardment Wing (VH), commanded by Col. Carl R. Storrie.133 The wing had been assembling at North Field on Guam since 8 February, when air echelons of the 19th and 29th Groups had flown in. On the 25th the wing still had less than a third of its authorized aircraft and had completed only part
of its shakedown training, but LeMay was confident that the crews assigned to the Tokyo mission would make out.134 In all, 172 planes got over the target and dropped 453.7 tons of bombs. Results, like the size of the effort, exceeded anything achieved before. Heavy undercast precluded strike photos, but prints obtained from later reconnaissance missions revealed that about a square mile of the urban area had been destroyed or damaged.135 Specifically, to cite Tokyo police records available after the war, 27,970 buildings had been destroyed and casualties had been numerous.136
This was the “conclusive” test of the fire bomb that the Twentieth Air Force had been asking for and the lessons learned were soon to be exploited. But on 4 March LeMay sent his planes over the familiar route to Musashino – now called by its proper designation, Musashino Tama (Musashi) – for another precision strike. This top-listed target, which had been visited seven times by the B-29’s and once by carrier planes, still stood virtually intact. Repeated failures, as Norstad had irritably reminded LeMay, lent added importance to the target.137 The force dispatched on the 4th was the largest ever directed against the target – 192 planes. But Nakajima’s luck held. The area was again heavily clouded and the planes dropped elsewhere by radar – 159 in urban areas of Tokyo and 17 on last resort targets. Results in all cases were unobserved.138
This eighth fiasco at Musashi marked the end of a well-defined phase of XXI Bomber Command’s operations. The effort to knock out the Japanese aircraft industry by high-altitude, daylight precision bombing of carefully selected targets had failed. Production of aircraft engines, not grossly off scheduled programs when XXI Bomber Command came to the Marianas, fell off sharply during the last two months of 1944 and production of aircraft declined slightly during the same period. In neither case, however, could the shortages be accounted for by destruction wrought by the B-29’s; indeed, the output of Franks at Ota, it has been shown above, had decreased from 300 a month to 100 before the first air strike.* Not one of the nine high-priority targets had been destroyed, although Akashi had been effectively crippled and production had been slowed down at Mitsubishi’s engine and assembly plants at Nagoya and at Nakajima-Ota. Musashi had suffered only 4 per cent damage after 835 B-29 sorties had been sent against it; Navy planes had done more harm in a single
* See above, p. 570.
strike.139 Probably the indirect effects of the B-29 raids were most important: with the fall of Saipan, Japanese industrialists had begun to lose confidence in their supposed immunity from air attack. Although under governmental pressure for increased production, they began, with the first attacks on Nakajima and Mitsubishi, to hunt for underground or forest cover, and the official directive for dispersal issued in mid-January merely served to quicken a process already well under way. This radical change in manufacturing techniques, never as efficiently conducted as the similar movement in Germany had been, explains in part the slowdown in production in late 1944 and early 1945.140
The indirect results – and indeed some of the direct results – of the B-29 campaign were not thoroughly appreciated by XXI Bomber Command at the time, and a balancing of visible damage against the effort expended was discouraging. In 22 missions involving 2,148 sorties the command had dropped on Japan 5,398 tons of bombs. Only about half of the planes had bombed primary targets.141 Losses had been high, rising in January to 5.7 per cent of bombers airborne. Bombing from altitudes in the neighborhood of 30,000 feet, the Superforts suffered relatively little from flak. Fighter interception, however, was often aggressive and effective, the more so because the restricted pattern of B-29 attacks allowed the enemy to concentrate his fighters in the Tokyo-Nagoya area. The long overwater trip to target and back, without a friendly base en route for refueling or repair, took its toll of wounded or malfunctioning planes.142 These difficulties were reflected in the statistics of losses incurred through February: twenty-nine B-29’s were lost to enemy fighters, one to flak, nine to a combination of fighters and flak, twenty-one to operational difficulties, and fifteen to unknown causes.143
After V-J Day, the United States Strategic Bombing Survey analyzed the unsatisfactory performance of XXI Bomber Command during its first three months of operations. They concluded that the failure stemmed in part from a tactical error, the continued adherence to the conventional doctrines of precision bombing. But many of the contributing factors lay beyond the control of the command’s leaders, and in general these were the factors described by Hansell in his final report to Arnold on 14 January. Like its sister organization in the CBI and VIII Bomber Command in the ETO, XXI Bomber Command had experienced the usual troubles of a pioneering organization. The supporting
services – maintenance, supply, weather, communications, reconnaissance, and air-sea rescue – had developed slowly in an area newly won from the enemy and under a theater command whose first interest was not strategic bombardment.144
During the whole period the command operated with inadequate facilities. Though the deployment of B-29’s was consistently behind schedule, it still outpaced the efforts of Navy construction battalions and aviation engineer units to provide proper bases. Each successive wing to arrive had to initiate operations from a single runway while others were being prepared. For inexperienced crews it was an arduous task merely to get a force airborne, to start and check their engines and follow the tight take-off schedule without overheating or fouling the engines by excessive ground idling time. Assembly, difficult at best on high-altitude missions, was complicated by congestion on the single-strip fields; the long interval between first and last plane airborne (238 minutes on one mission, 193 on another) increased the already heavy fuel consumption required by long missions and the climb to bombing altitudes of 30,000 feet. Thus the B-29’s theoretical bomb load of ten tons was reduced to an actual three tons, and the slender fuel reserve made it impossible to check navigational errors or to compensate for headwinds. With the cheerless prospect of a night return over vast stretches of enemy waters as gas-tank gauges sank, many crews were forced to turn back short of the objective or to bomb targets of opportunity.145
Operational hazards were aggravated by maintenance difficulties. Hardstands were overcrowded, group and depot installations were completed slowly. Initially, each wing’s service groups contained so many inexperienced mechanics that intensive training programs had to be conducted simultaneously with routine combat maintenance. An improper balance of supplies hampered repairs as well as operations. Maintenance failures (and, for a while, poor technical inspection) were responsible for most of the numerous aborts: for 90 percent in November, 85 per cent in December, 66 per cent in January, and 64 per cent in February.146 In terms of total effort, the planes aborting because, of maintenance failure amounted to 25 per cent of all those scheduled in November and December, 23 per cent in January, and 16 per cent in February.147
The most serious obstacle to successful bombardment was weather. Severe frontal conditions, frequently encountered on the trip north
from the Marianas, increased fuel consumption, scattered formations, and made navigation so difficult that many crews missed the landfall entirely. Over the target, crews rarely found atmospheric conditions suitable for precision bombing. The proportion of planes bombing visually had diminished progressively through the winter months: 45 per cent in December, 38 per cent in January, 19 per cent in February.148 Radar bombing seldom proved successful. The AW/APQ-13 radar often malfunctioned at extreme altitudes, and radar operators were in general not sufficiently trained to get maximum results; even under skilled hands its accuracy was not up to the requirements of precision bombing.149 Cloud cover as a hindrance to bombardment was familiar to the AAF from bitter experience over Europe, but the tremendous winds encountered at bombing altitudes over Japan offered a Novel and most disconcerting problem. With wind velocities reaching 200 knots and more, drift was difficult to correct and bomb runs had to be charted directly upwind or downwind. Attacking Japan’s best-defended cities directly in the teeth of a 200-knot wind was unthinkable; going downwind the B-29’s reached ground speeds in excess of 500 miles per hour, in which case neither bombsights nor bombardiers could function properly. Moreover, the high winds made it impossible for crews to make a second pass if the run-in failed; if a navigational error brought a plane in downwind from target it might not be able to attack at all.150
Some of these difficulties decreased as the command grew to full stature, and the efforts begun by Hansell and continued by LeMay had brought improvement-in the rate of aborts, for example. But bombardment results were still far short of expectations, and by early March XXI Bomber Command had come to a crisis. The crisis would be solved by a radical change in bombardment tactics.