In order to understand the seeming ease with which the Japanese could make and break contact with their opponents at will, the reader should have some comprehension of the extent to which western New Britain remained terra incognita to the U.S. invaders.

Maps attempted to show only the approximate routes of only the most important trails, none of them visible to aerial observation, and gave no indication of the maze of secondary trails that angled off in all directions to serve the gardens, or former gardens, of the local inhabitants. Place names of native villages were spotted liberally, and guides became increasingly available, but seldom did any two authorities agree as to village locations or even which village boasted which usually unpronounceable name. For these reasons localities shown on the accompanying maps are more indicative than actual.

What ensued after the capture of Hill 660 developed many aspects of a gigantic game of blindman's buff, with the Marines "it." Familiarity with the country gave the Japanese the inestimable advantage of being able to choose their ground, whether to fight or to withdraw. Their opponents, on the other hand, must painstaking explore dozens of false leads in order to orient themselves, and this in jungled, mountainous, swampy terrain where the incredibly dense rain forest hid landmarks so thoroughly that a patrol might pass within a few yards of its objective without ever discovering it.

And after 21 January the Japanese enjoyed still a further advantage: they knew what they were going to do and how they were going to do it, while their pursuers could only guess.

Patrol activity in western New Britain took many forms and was practiced on an extremely wide scale. In addition to such normal functions as maintaining security and obtaining intelligence, patrols had an important responsibility in searching out and bringing in Melanesian natives of the region who had fled into the hills to escape the Japanese. In all, some hundreds of sick, emaciated refugees came into the U.S. lines for rehabilitation as a result of these humanitarian efforts: to serve as guides, carriers and laborers (thereby depriving the Japanese of their similar services) and to be repatriated to their villages as soon as practicable. Such results were highly gratifying, and the natives showed proper appreciation. However, inasmuch as these activities were essentially non-military in nature and performed under

AS THE JAPANESE RETREATED the natives returned to their villages. Here an ANGAU representative recruits laborers for the Marines.

direction of Australian officers¹ sent along for the purpose, they are better omitted from a tactical account of the operation.

Following the capture of Hill 660 and subsequent loss of contact with the enemy, patrol operations assumed a new pattern and a new importance: to seek out and destroy, or at least drive off, the vanished Japanese preparatory to completing the division's secand major mission--seizure of the line Borgen Bay-Itni River.²

This took the essential form of a two-pronged probe, converging on the region where General Matsuda's headquarters were known to be, or to have been. These movements occurred simultaneously but in the interests of clarity will be taken up in turn.

Operations in the High Ground

Patrols had been active in front of the airdrome perimeter ever since its establishment,

but owing to the maze of native tracks and cross tracks the Marines were some time in locating definitely the main route over which the Japanese survivors had escaped the region: an ascending trail that skirted the eastern flank of Mt. Talawe. On 19 January a strong three-day patrol made up of personnel of Company L, 1st Marines, followed this upward past Mts. Gulu, Langla (the volcano) and Munlulu, then down to the high saddle north of Mt. Tangi. There, at a trail junction,³ they surprised a group of some 20 Japanese and eight armed natives, who fled eastward after losing six of the Japanese killed.

Four abandoned bunkers guarded the junction, indicating its importance in the minds of the Japanese, and discarded enemy packs, equipment and medical supplies were found in the vicinity. The patrol scouted the east-west trail that formed the junction for 300 yards in each direction, making no further contacts but discovering a Japanese telephone wire, which was cut. The trail itself proved exceptional for that region: wide enough to accommodate a jeep along many stretches and showing signs of considerable recent use.

The report of the patrol leader, Captain Ronald J. Slay, CO of Company L, served to convince the division command that he had found precisely what he had set out to find: the east-west Government Trail which the maps showed as running inland from Sag Sag to join the main north-south trail at a village called Agulupella, a native name no more unpronounceable than many thereabouts, which was to figure prominently in orders and reports for some time to come. Plans were immediately made to exploit the discovery.

On 22 January two patrols, each consisting of a reinforced company, set out from the airdrome perimeter with the mission of converging from two directions at the trail junction found on the 19th. A composite force⁴ of about company strength made up of elements of the 1st Battalion, 1st Marines, under Captain Nikolai S. Stevenson, followed Captain Slay's route southward over the steep mountain trail and reached the vicinity of Mt. Gulu, northernmost of the projecting shoulders of Mt. Talawe, where the patrol set up an ambush for the night.

Simultaneously Captain Preston S. Parish embarked a similar composite company of the 7th Marines in landing craft and proceeded to Sag Sag, terminus of the East-West Government Trail, near the Anglican mission where the Reverend Sub-Lieutenant Wiedeman, who accompanied the patrol, had held forth some years before the war. The men bivouacked about one and a half miles inland of that place and moved on to Aipati on the 24th. Neither patrol made contact with the Japanese, but Parish gained intelligence in nearby native villages⁵ indicating the recent presence of several small parties.

The next day, however, told a different story for the Stevenson patrol. Hardly had the men got off to a good start along their difficult route than a hidden machine gun opened on the point at a range of about 30 yards near Mt. Langla. Only a small volume of rifle fire supported it, leading to the belief that no more than six or seven Japanese had set the ambush. The advance guard deployed and drove these off in a brief fire

fight, without loss to the Marines, nor to their opponents so far as they could tell.

The patrol proceeded more cautiously after that, but was ambushed again at about 1500 in almost precisely the same manner, and quite likely by precisely the same Japanese. This time the advance guard reacted more quickly and killed two of the enemy before the rest took to their heels.⁶ With his men tired and short of water, Stevenson did not attempt to pursue but selected favorable terrain and set up an ambush of his own for the night.

On 24 January the patrol pushed on to a point about 1,500 yards south of Mt. Munlulu, where it was suddenly pinned down before a strongly situated enemy force estimated at platoon strength, stiffened with two heavy machine guns. Unable to neutralize this fire, and unable to close with--or even see--the Japanese positions in that rugged terrain, Stevenson finally managed to break contact and fell back to more favorable territory in hopes that his opponents, encouraged by their success, might feel optimistic enough to attack him.⁷

THE PARISH PATROL embarks for Sag Sag.

Map 15
Western Patrols
January-February 1944

A MARINE PATROL gingerly uses a log to cross a stream.

In light of Captain Slay's experience a few days before, this resistance came as a rude surprise. The three fire fights had seriously depleted Stevenson's ammunition, he was short of food,⁸ and more than five miles of most difficult mountain country now separated him from his base. Prudently he drew back to Mt. Gulu on the 25th, where he and his patrol were relieved the following day by a fresh group from the 1st Marines: Captain George P. Hunt's Company K.⁹

Meanwhile the Parish patrol had been working its way eastward, combing the country thoroughly and keeping the trail open to the rear as the most practicable supply route. It encountered no Japanese west of the trail junction, and intelligence obtained from natives convinced the commanding officer that the enemy had withdrawn from the area. He learned of Stevenson's predicament and moved to his assistance but was unable to make contact.¹⁰ However, on the morning of 27 January Captain Hunt broke through the now abandoned position that had held up his predecessor and reached the junction, where he found a group of Parish's force in occupancy.

Here he learned that a small patrol reconnoitering to the east had been ambushed the previous day short of the village of Niapaua, near the junction of the important government trails. Borrowing a machine gun platoon from Parish, Hunt set off in that direction. Some distance farther on his advance scouts discovered an estimated 50 Japanese in position on the far side of a stream that cut across the trail. As nightfall was approaching, he set up a perimeter 700 yards short of that point and prepared to attack in the morning.

This attack began at 0700 on the 28th and soon drove off the Japanese previously discovered. However, these proved to constitute only an outpost line, and presently the advancing Marines came upon an enemy force estimated to number at least one reinforced company armed with both light and heavy machine guns and mortars " . . . strongly entrenched in pillboxes on cliffs commanding all areas of approach . . . . Terrain precipitous and very dense jungle which hampers visibility."¹¹ Unable to maneuver, Hunt engaged in a fruitless fire fight for three hours, during which his group suffered 15 casualties,¹² then decided to pull back beyond mortar range. The Japanese, mistaking the motives of this withdrawal, swarmed out of their positions and attempted to pursue, only to be stopped by strong rear guard action, thus completing the temporary stalemate.

The remainder of the Parish patrol joined Hunt in bivouac the following day. Major William J. Piper, Jr., executive officer of 3/7, had brought up reinforcements to Parish the previous evening and now assumed command of the combined group. Resuming the advance on the morning of 30 January, the Marines found the strong Japanese positions abandoned and pressed on without further incident to Niapaua and Agulupella where Piper had orders to stand by to form part of a larger force, designated the Gilnit Group (see below).

There they joined company with a strong patrol from the 5th Marines, which had worked its way inland from the Borgen Bay area, and learned that once more, save for a few wretched stragglers, the entire Japanese 65th Brigade had vanished into thin air.

Eastern Area Patrols

As with their compatriots on the airdrome perimeter, the principal interest of Marines operating inland of Borgen Bay and along its shores lay in consolidating their defense positions. At first their patrols worked about 1,000 yards in front of the lines, primarily for security purposes, with the result that they made no important contact with the vanished Japanese until 20 January.

Hindsight shows that greater aggressiveness might have prevented the tactically successful withdrawal of the enemy,¹³ but the reason this was not employed should require little explanation. The BACKHANDER Force had accomplished its basic mission quickly

THIS PATROL is scouting Japanese positions in the Natamo area. A few moments after the picture was made an enemy machine gun opened fire, killing two of the Marines and wounding several others.

GENERAL SHEPHERD AND COLONEL SELDEN discuss the fighting at Natamo. (Army Photo.)

and economically; its secondary mission--seizure of the line Borgen Bay-Itni River--would better await consolidation and reorganization. Further acquisitions of terrain could have no possible value; and the difficult nature of this terrain made it far more practicable for defense than for attack, as the Marines had discovered in their advance to Hill 660. Estimates indicated presence of the Japanese in sufficient strength to cause plenty of trouble: troops as familiar with the features of the region as the Marines were ignorant. Nothing in U.S. plans or orders stipulated pursuit of the enemy beyond boundaries already gained, and to attempt this under the circumstances prevailing would risk further loss of life to no useful end.

The contact on 20 January occurred when a patrol¹⁴ from 1/5, probing along the southern shore of Borgen Bay, drew heavy automatic-weapons fire from Natamo Point. The patrol immediately deployed along the western bank of a small estuary partially separating the base of the point from the mainland and engaged the enemy in a two-hour fire fight. With mortar shells expended and machine-gun ammunition running low, the leader radioed a request that artillery fire be placed on the point. Under cover of this, the patrol withdrew without casualties and returned to the perimeter, on the way destroying two enemy dual purpose 75mm guns, evidently abandoned by the Japanese in order to consolidate their strength on the point.¹⁵

It so happened that a map of the point had been captured on 3 January that showed numerous light and heavy machine-gun positions and indicated that it would be defended by approximately one platoon. Accordingly a reinforced company prepared to attack and made contact again on the 22d but found opposition too strong to permit crossing the estuary. When two days working over by artillery and Fifth Air Force A-20's failed to destroy the indicated positions or produce any perceptible diminution in the volume of fire returned, first estimate of enemy strength was revised upward to company level, armed with 20mm automatic cannon and at least one 37mm and one 75mm as well as the machine guns previously spotted.

On 23 January another company of the 1st Battalion, reinforced by medium tanks, crossed Borgen Bay by boat and landed on the east bank of the Twin Fork River. With the further support of artillery and a rocket-launching DUKW, Companies C and D, under Major H. T. A. Richmond, battalion executive officer, captured the point that afternoon.

The following day, in a series of short, sharp skirmishes, they secured a stretch of shore line extending 500 yards eastward to the left bank of the wide, deep Natamo River. The Japanese held the opposite bank in sufficient strength to repulse all attempts

to cross near the mouth for the next four days, even with support of half-tracks and the rocket DUKW, and they succeeded in bringing an artillery piece to bear on the beach positions and landing craft supporting the operation. Meanwhile Marine patrols explored upstream in an attempt to find a practicable ford, but not until 28 January, with all of 1/5 operating in the area, did the battalion manage to gain a firm hold east of the river.

During the same period, patrols had been working cautiously inland as well. On 23 January, a platoon from 2/1, attached to 5th Marines¹⁶ and holding the right of the eastern perimeter, followed the main trail to within 1,000 yards of Magairapua, known to have contained the headquarters of Colonel Katayama's 141st Infantry. Here it got diverted onto a secondary trail where it encountered a Japanese force of estimated equal strength, killing four and driving off the rest. However, the next day a three-man reconnaissance patrol from 2/5 succeeded in entering the village, only to find a large bivouac area evacuated by the Japanese in apparently good order.

These events, combined with those taking place simultaneously behind Mt. Talawe, convinced division intelligence officers that they were facing no isolated pockets of resistance, but integral components of a single well planned operation: that the force in the Natamo region and that in front of the Hunt-Parish patrols had been placed where they were deliberately to fight holding actions in order to cover the withdrawal of the Japanese main body.¹⁷

But whither that body had withdrawn remained a mystery. Its most direct and obvious route lay along the north coast trail toward Talasea and Cape Hoskins, and the fight in the Natamo region might well be a covering action for such a move. But several alternatives existed. The Japanese might head due south over the Government Trail to Cape Bushing, known to have been strongly garrisoned at one time, on the slim hope of being evacuated by sea. Or they might join the force still operating against Arawe. Or they might go on along the south coast to Gasmata where an overland trail connected with Talasea and Cape Hoskins well beyond reach of American interception. That those groups detected withdrawing invariably retired to the southward led to considerable speculation that they had chosen one of the three latter courses.

The division command determined, therefore, to push the pursuit more vigorously. The plan called for troops moving inland from the eastern perimeter to join the Hunt-Parish group in the vicinity of the trail junction village of Agulupella. If they could cut off any Japanese in the process, all the better, but the main purpose was to form the combined forces into a super-patrol and move southward across New Britain to Gilnit on the Itni River, thereby completing the operation's secondary mission.¹⁸

Accordingly, on 27 January 1/5 dispatched a combat patrol to the southwest over a wide unmapped corduroyed trail which the battalion had found some 500 yards east of Natamo Point. Moving inland, these Marines found a number of recently used bivouac areas and some abandoned equipment, leading to the belief that they had discovered the enemy withdrawal route from this region. This belief appeared confirmed when, about 3,200 yards from the beach, they got into a vigorous

fire fight with a force estimated to number 100, which they took to be the rear guard of a larger force. While they were attempting to break contact, the Japanese observed a machine gunner leave his piece and tried to rush it. The assistant gunner, however, was still in position and mowed down 15 of the attackers, whereat the patrol returned to its parent unit.

At the same time Company E, reinforced, advanced from the 2d Battalion, 5th Marines perimeter toward Magairapua. This patrol made two contacts with the enemy, who retired southward in both cases, and the company bivouacked that night 200 yards short of Colonel Katayama's former command post.

On the 28th, the day that Captain Hunt had his vicious three-hour fire fight west of Niapaua, Major William H. Barba started out to his assistance with 1/5 reinforced, less one company,¹⁹ over the corduroyed trail. This unit encountered only trifling resistance and went on past the scene of the previous day's fire fight, but was held up by mortar and machine-gun fire some distance beyond. The Japanese withdrew, again toward the south, and the battalion bivouacked on the scene with six casualties, reaching Nakarop the following day without incident.²⁰

MATSUDA'S COMMAND POST yielded boxes of documents and clothing.

There the 1st Battalion joined Company E, which had arrived earlier, and was joined in turn by a group designated the ADC patrol. This consisted of Company G, 5th Marines, strongly reinforced with supporting weapons and accompanied by a number of native bearers, under command of Major Charles R. Baker, 2d Battalion executive officer. It had started out that morning and reached the scene over Government Trail via Magairapua.²¹ Patrols from Nakarop made contact with the Hunt-Parish group (now commanded by Major Piper), and the two groups rendezvoused at Agulupella on 30 January, as previously mentioned.

Nakarop turned out to be the place name which Allied translators had transliterated from Japanese sources as "Egaroppu," nerve center of 65th Brigade operations. Tents still stood in abandoned bivouac areas, and the village contained both native and Japanese-built structures. The largest of the latter carried a sign translated: "Matsuda Butai Army Command Principal Place."²² It showed traces of elaborate radio installations recently removed and contained an American-made telephone switchboard labeled "Glory Division." But of the former occupants there were no live traces at all.

This combination of patrols, now grown to sizable proportions, was designated the Gilnit Group and ordered to stand by in the Agulupella neighborhood preparatory to executing its cross-island mission to the Itni River. As a result of an on-the-scene investigation, its acting commanding officer informed Division Headquarters as follows "Natives report Japs left Agulupella area 28Jan in mass evacuation (possibly 700) and

GENERAL MATSUDA'S HEADQUARTERS following discovery and occupation by the Marines. (Army Photo.)

followed Aisalmipua-Kakumo Road toward Borgen Bay and Cape Gauffre. Many sick and wounded; extreme poverty, reduced to eating native dogs and looting native gardens."²³ Another patrol the same day reported native intelligence indicating that the enemy's immediate objective was Kokopo, just east of Cape Gauffre.

This convincing evidence of the general direction of the withdrawal was shortly confirmed by capture of the actual withdrawal order issued to General Matsuda by 17th Division headquarters, and Matsuda's own order directing withdrawal in four echelons via the Aisalmipua-Kakumo trail toward Kopopo.²⁴ But the immediate value of this

IN A SERIES of shore-to-shore operations the 5th Marines jumped from Natamo to Iboki Plantation.

information was not so great as might appear on the surface. Nobody knew where Aisalmipua and Kakumo might be, since they were spotted at widely separated points on different maps, and even the natives could not agree which was which. So again the dispatch with which the pursuit was pressed left something to be desired.

A patrol from Company E, 1st Marines, operating out of Old Natamo, explored the coastal track northward along the eastern shore of Borgen Bay, reaching Alaido on 1 February and Namuramuga the following day.²⁵ It reported evidence of some Japanese having passed this way, but so few as to indicate that the main withdrawal route lay elsewhere. So two other patrols probed inland simultaneously through the jungle in an effort to locate this and intercept as many of the retiring enemy as possible. On 2 February they reached the trail in question at widely separated points, to the east and southwest of Old Natamo. Discarded gear, abandoned weapons and bloody bandages verified the nature of their find. But they encountered only 19 Japanese stragglers, sick, wounded and dying.²⁶

Clearly, such cohesive fighting units as General Matsuda still possessed had gained a substantial head start, how substantial no one could be certain. Accordingly, with his opposite number's intentions and course now clear, the Commanding General, 1st Marine Division, ordered vigorous pursuit as far as Iboki, designated in the captured orders as rallying point for the withdrawing force.

Map 16
Patrols Eastward to Iboki

The 5th Marines assumed this mission, relieving the 2d Battalion, 1st Marines to rejoin its parent regiment in the airdrome area. Further attempts at interception would be made by means of a series of leapfrogging amphibous landings: placing detachments of Marines ashore at successive points along the north coast to commence patrol operations from there. Landing craft operated by the Boat Battalion, 533d Engineer Boat and Shore Regiment, of the Army's 2d Engineer Special Brigade²⁷ were made available to carry out the amphibious phase, and the division's little Cub artillery spotter planes for observation and liaison.

High seas frustrated the first landing attempt: at Namuramunga on 4 February. In consequence, two patrols from 2/5 proceeded overland from Old Natamo, via the coastal track and main withdrawal route respectively, with orders to reach Kokopo, where recently captured Japanese indicated the fleeing remnants expected to pick up supplies. The coastal patrol got as far as Mambak, killing eight stragglers and taking three prisoners, but both patrols were ordered back to the perimeter on 6 February without reaching their destination.

The following day a patrol from 2/5 landed successfully at Aliado with orders to follow the coastal track to the mouth of the Gurissu River through Kokopo and Gorissi. A Cub

COLONEL SELDEN AND LIEUTENANT COLONEL BARBA look over the latest catch of Japanese prisoners at Iboki Plantation.

plane observer reported Marines in the first named place on 8 February.²⁸ But whatever supplies had existed at either village had been either carried off or destroyed by retiring elements farther forward, and the only Japanese encountered were either dead or in various stages of dying of starvation or disease, or a combination of the two. The patrol completed its mission on the 10th, killing 16 stragglers and bringing six prisoners back to the Borgen Bay perimeter. It also brought back an unverifiable native report that General Matsuda and at least some troops had been been evacuated by barge ten or twelve days before.

The 3d Battalion, 5th Marines (less Company I) moved into Gorissi by boat on 12 February and set up a radar station. A patrol from this battalion reached the El River on the 16th but found it impassible. A second patrol, therefore, leapfrogged the first by sea, landing beyond the next river to the east and arriving at Arimega Plantation on the 19th. Karai-ai, key Japanese supply point west of Iboki, now lay only a short jump ahead, and 3/5 seized it by a combined overland and overwater operation on 21 February.²⁹

IBOKI BEACHHEAD, February 1944.

At this point 1/5 leapfrogged 3/5. In the words of the battalion commander:

Our mission was to advance and patrol the area to a native village just west of Iboki Plantation. 2-5 would then leapfrog 1-5 and seize Iboki Plantation. We had 2 LCM's to assist us in the move. We accomplished the mission by sending a platoon size patrol along the coastal trail to a designated landing beach, while other patrols ranged inland from the battalion bivouac. In the afternoon, we would move 1 to 2 companies by LCM to the landing beach designated as the coastal patrol's objective and set up a perimeter defense for the night. The next morning fresh patrols would be sent forward and the LCM's used to bring forward the remainder of the battalion.³⁰

Patrols headed by the R-2 scouts reached Ketenge Anchorage on 22 February and pressed on. Trouble was anticipated at Iboki as captured enemy orders named it as the main supply base and rallying point for the retiring Japanese. However, both the 1/5 coastal patrol and two companies proceeding by water mistook Iboki for the designated battalion objective and arrived there simultaneously on the 24th in advance of 2/5, to find the supplies systematically evacuated and the place deserted save for a handful of stragglers.³¹ By the 27th the regimental command post and all remaining elements of the 5th Marines, reinforced, were firmly established in the immediate area and patrolling in all directions.

Summarizing these operations, the division reported:

For the first time in the history of the 1st Mar Div a reinforced regiment had conducted a shore-to-shore operation along the coast of enemy-held territory. Using only 10 LCM's and jungle trails they had transported and marched 5000 men with their intendant supplies and equipment, for a distance of 60 miles around and over some of the worst jungle terrain in the world.³²

What made this accomplishment the more remarkable was the somewhat impromptu manner in which it was undertaken initially. Colonel John T. Selden, 5th Marines commanding officer, had no inkling that any such move impended and had hiked across the island to join Colonel Puller's Itni patrol (see second sub-chapter following). As he tells the story:

. . . was one day's march from my CP on the return trip when [J. S.] Stankus came dashing in with a message . . . that something was up that would affect the 5th . . . Arrived at Division CP at eight or nine o'clock [in the evening] to receive orders to put my show on the road. . . That was all the advance notice I had. The next morning, we were on the move. To have accomplished my march four days prior to the deadline without loss or even having a man wounded was, in our estimation, quite a feat.³³

However, owing to the long delay in its execution, the move failed to catch Matsuda or to intercept any organized remnants of his force. To understand the extent to which this delay played into the enemy's hands, it will be well to see what the Japanese were doing all this time.

The Katabasis³⁴ of Jiro Sato

Capture of the 65th Brigade files brought Marine intelligence officers their first intimation of the presence of an officer and a unit not previously identified as being in western New Britain: Colonel Jiro Sato and his 51st Reconnaissance Regiment.³⁵

Initially this unit had not been integral to the brigade. Matsuda fell heir to it when it became stranded in western New Britain en route to join its parent organization: the 51st Division, then operating on New Guinea. The general used it to occupy Rooke Island, to the west across Dampier Strait, in which

capacity the "regiment" numbered approximately 500 officers and men. On 6 December, however, about half of this strength was shipped to the Cape Bushing area where most of the personnel were incorporated into the newly formed 3d Battalion, 141st Infantry, on 12 December.³⁶ (See Chapter IV).

On 7 January, with the Japanese needing every man for the Borgen Bay fighting, Colonel Sato embarked the remainder of his command and set out to join the main body. His route lay via Cape Bushing, up the Itni River to Nigol, thence northward on foot over the Government Trail to Nakarop. He arrived to find his compatriots hard pressed, his own the only even comparatively fresh unit on the scene.

But despite the heavy pressure, General Matsuda did not commit this final reinforcement to the main fighting, then rolling inexorably toward Aogiri Ridge and Hill 660. Instead, on 12 January, Sato led his command off in the opposite direction on a special mission: defense of the western approaches to headquarters against Marine patrols now beginning to operate in the wide saddle behind Mt. Talawe. There, as related earlier, various elements gave Captain Stevenson's patrol a bad time in the Mt. Munulu region on the 23rd and 24th, and stopped Captain Hunt's force short of Niapaua on the 28th.

By then, although the Marines had yet to find it out, the withdrawal of the Japanese main body had been under way for nearly a week, with the advance elements already beyond range of immediate interception. This withdrawal was specifically directed by 17th Division orders dated 21 January 1944, at Gavuvu.³⁷ These instructed the entire 65th Brigade to retire along the north coast where the survivors could obtain supplies at various barge staging points and possibly be evacuated by sea. Other orders directed other elements to do everything possible to aid and expedite the movement, especially shipping engineers and medical personnel. The division commander directed reassembly of what remained of the Matsuda Butai at Iboki preparatory to establishment of a new defensive position along the line Talasea-Gasmata to check further incursions in the direction of Rabaul. Altogether, it was anything but the haphazard enterprise certain of its aspects might have led observers to believe.

The task thus imposed would have been a prodigious one under the best conditions. And the conditions faced by the Japanese could scarcely have been much worse. The troops, on short rations even before the campaign began, were now seriously weakened physically; also demoralized by defeat and riddled by disease as a result of three weeks of jungle fighting in monsoon rains. Iboki lay some 60 miles away, the Talasea-Gasmata line more than another 60 miles beyond that. And these wretched men, many with feet so infested with fungus infection that they walked with difficulty, must cover this distance afoot over jungle trails, carrying with them everything they needed to live and fight: weapons, ammunition, packs and at least enough food to last until they reached the first designated ration dump where they could replenish. They even tried to bring off their remaining machine guns, automatic cannon and artillery--at first.

The services of natives as guides and carriers would have been priceless at this time, and the Japanese impressed all such unfortunates as fell into their hands. But callous treatment had alienated the Melanesians long since, and now the spectacle of their late overlords in defeat began turning latent hatred into overt hostility. Thus, a potential resource became an added menace.

Despite the difficulties that lay ahead, Matsuda lost no time in putting behind him the scene of his military failure. In anticipation of the withdrawal order, he had already broken contact with the invaders and laid his own plans. Now he started his main body toward the north coast in successive echelons, sick and wounded first, artillery in the rear, over the unmapped inland trail to

MAIL CALL was always popular. Here it goes for some of the 5th Marines. (Army Photo.)

Kokopo which the groping Marines did not find until nearly two weeks later. To insure the movement against enemy interference, he left behind two covering forces: Colonel Sato's unit in the high ground to protect his rear, and what was designated the Hanahara Group³⁸ in the Natamo region along the southern shore of Borgen Bay, the direction from which interception could be most feared.

As things turned out, Matsuda's plans revealed excellent foresight and good judgment. Be it remembered, the general had sat out the war and the weather in a snug house, with plenty of food and medical care, and apparently he craved some more of the same. In any event, he placed himself and his headquarters group in the second echelon,³⁹ with

what remained of the 141st Infantry, thereby gaining protection of the strongest cohesive unit and enjoying the security of both advance and rear guards. Superior Private Toshio Herotsune, a medical corpsman in 1/53, encountered him at the point where the withdrawal trail crossed the headwaters of the Natamo River and later reported him in excellent health and spirits, passing along the cheering word that strong reinforcements had landed at Iboki and were marching to the brigade's relief.⁴⁰

All the cohesive echelons of the main body had cleared the area by 27 January, on which date Colonel Sato received orders to fall back in turn, still covering the rear of the brigade. However, one of his units remained engaged with the Hunt Patrol west of Nakarop during the morning of the 28th until the Marines broke contact. Thereupon it retired quickly to the designated rendezvous point: the village of Gurimati, some two miles east of Agulupella, thereby neatly sidestepping the several patrols from the Borgen Bay region that were converging on the later place, as previously related.

After pausing to reorganize and assay the situation, Sato struck eastward again and reached the main escape route at Kakumo. Following this in a generally northeasterly direction over some rugged high ground, he arrived without incident at Kokopo on 2 February, the same day that the Marines finally discovered the trail many miles in his rear.

The coastal track presented an increasingly gruesome sight as the withdrawal continued. Obviously, if any cohesion were to be retained among the fleeing remnants, able-bodied troops could not be further hampered by invalids. Men unable to continue must shift for themselves. Such unfortunates were provided with hand grenades and instructed to blow themselves up at the approach of the enemy. Many did not prolong their agony to that extent. Some begged their officers to shoot them.⁴¹ Many others died of illness or sheer starvation before the Marines came up with them. The stench of death hung over New Britain's north coast like a miasma.

But there was little Sato could do for the stragglers if he were to carry out his mission. So he pressed on, to such good effect that he reached Karai-ai early on 12 February, the day that the 3d Battalion, 5th Marines, belatedly aware of the Japanese course, set up an advance base at Gorissi. More than 20 miles now separated the rearmost organized element of the Japanese from the foremost of their pursuers.

Karai-ai was still operating as an active base. Here Sato placed nine of his sick in the small hospital run by the 1st Shipping Engineers, in hopes of their eventual evacuation in one of the scarce barges that still ventured in under cover of darkness. Then he drew five days rations for the rest of his force and set out the following day for his next destination: Iboki Plantation, which he reached on the afternoon of 16 February, while the Marines were still trying to find a way across the swollen El River, 30 miles behind.

In the original withdrawal orders, the Commanding General, 17th Division, had designated Iboki as rallying point for the 65th Brigade, but at some time in the interim the plan had been changed. The advance echelon, 2d Battalion, 53d Infantry and some sick and wounded had been evacuated by sea,⁴² and such other units as remained in command control were plunging

through the jungle beyond with Cape Hoskins as their destination. At Iboki the main escape route swung widely inland to avoid a series of deep river mouths and broad swamps, and Sato had to travel some miles⁴³ south in order to report to Matsuda Butai headquarters, located at a village called Upmadung.

There he found not only the general but Colonels Katayama and Sumiya,⁴⁴ all of whom had dropped out of the Americans' ken since the beginning of the withdrawal. There, too, Sato drew a modification of his original assignment. As translated from the 17 February entry in the journal that furnished most of the material on the 51st Reconnaissance Regiment contained herein: "Received orders that this Regt will be SATO Det, occupy key positions at ARIA River mouth area, supply KOMORI Det (T. N. 1 Bn, 81 Inf Regt) by land, and gather intelligence."⁴⁵

But he was not destined to enjoy the company of these officers for long. On 23 February Matsuda embarked his headquarters and as many hospital patients as possible in such few barges as could reach the scene and departed the Aria River for Cape Hoskins where, after circling Wilaumez Peninsula, he reported in to 17th Division Headquarters on the 25th.⁴⁶ Shortly thereafter Sumiya and Katayama in turn set out for the same destination overland⁴⁷ and disappeared into the all-enveloping jungle, leaving

THE COMMANDER AND EXECUTIVE OFFICER of the Gilnit Patrol, Lieutenant Colonel Lewis B. "Chesty" Puller and Major W. J. Piper.

Sato very much on his own. With characteristic vigor he set about evacuating the stores remaining at Iboki to Upmadung and rehabilitating such stragglers as he could while awaiting further orders.

The Itni Patrol

In the meanwhile, quite indifferent to the grisly game of hare and hounds taking place along the north coast, a patrol group of formidable proportions assembled at Agulupella for the trek southward across the island. Lieutenant Colonel Lewis B. Puller arrived on the scene 30 January to take charge. For the first time since the patrols started operations in this general area, they were united in one place under one command and assigned one mission.

Designated the Gilnit Group after the native village that was its destination, Colonel Puller's unit was directed to clear out the enemy to the line Borgen Bay-Itni River, thus securing all of western New Britain and completing the secondary mission assigned the division. The organization included: The Uncle Group, composed of three reinforced rifle companies--K/1 (former Hunt Patrol), G/5, and a composite company from the 7th Marines (former Parish Patrol); the Natamo Task Force composed of 1/5 less one reinforced company, and E/5 (reinforced); plus a headquarters

THE GILNIT PATROL puts out aircraft panels at Arigilupua (pre-war population 31). (Army Photo.)

detachment made up of intelligence personnel from regimental and division levels. These units mustered a combined strength of 1,398 and were augmented by three Australians: Major Mather, Lieutenant Gray and Sub-Lieutenant Wiedeman, who were accompanied by 150 native carriers.

Upon his arrival, Colonel Puller dispatched patrols south and southeast Agulupella to feel out the enemy, but it became readily apparent that any effective Japanese aggregation which might offer a fight had departed in the general eastward movement. The group commander radioed division that the only living sign of the enemy "was a police dog seen dashing through the village of Niapaua.⁴⁸

Supply for a patrol of such size posed exceedingly difficult problems. Originally it had been planned to handle this by LCM from Gloucester to Sag Sag and thence overland to Agulupella, but Colonel Puller labeled this impracticable. The trails leading into Agulupella were no different from others of Cape Gloucester, and under a combination of constant rain and heavier use than that for which they were designed, they quickly turned into mire. Native carriers were utilized to bring in supplies from the 5th Marines' perimeter, but that was at best a stopgap measure.

Combining the problem of supply with the obvious evacuation of the area by the Japanese, division decided to reduce the size of the Gilnit Group and supply the remainder with what was then the novel method of plane-drops.⁴⁹ This was ambitiously begun with B-17's from Dobodura, but it was

quickly and painfully learned that the bombers had to fly so high that their drops were scattered over a huge area, a portion of their deliveries sometimes landing within the patrol itself and causing casualties. Small cub planes were then resorted to, and they proved effective for the duration of the Gilnit patrol.⁵⁰

While the main force readied itself at Agulupella, Captain Hunt's Company K, 1st Marines, moved on to Arigilupua to protect engineers engaged in bridging a river near that village. Bored by the necessity of waiting, Hunt requested permission to patrol southward, which Puller granted.⁵¹ He thereupon departed toward the Itni with 11 men, several from each platoon. The combination of a small force lightly armed, a well-marked trail and no living signs of the enemy⁵² put the group at Nigol on the Itni at dusk and there it spent the night.

The next morning Hunt sent a runner back to Colonel Puller with information of his location and a request that rations be airdropped. Japanese craft found on the river bank were not considered adequate for crossing to Gilnit, however, so the patrol continued southward until stopped by a large swamp, whereat it turned back and spent a second night at Nigol.

Shortly after dawn a native runner from Puller met the patrol with an order that Captain Hunt place himself under arrest for exceeding his orders and that the next senior member take charge and bring back the patrol at once.

Meanwhile, construction of the bridge at Arigilupua and sufficient stockpiling of supplies, the primary hindrances to the patrol's movement to the Itni River, were accomplished by 6 February, and the Gilnit Group, much reduced in size, moved southward from Agulupella in echelons. The force now contained a composite reinforced company from the 7th Marines, I/5 (reinforced) which had replaced G/5 a few days earlier, and elements of the 11th Marines,⁵³ 17th Marines and division headquarters. In addition to the Australians and native carriers, the group now numbered 384 Marines.

Profiting by earlier experiences in following New Britain trails, Coloner Puller and his second in command, Major W. J. Piper, Jr., resolved to keep the group split in two sections under their respective commands. The patrol was carried out with leapfrog tactics, one group exploring side trails while the other proceeded on the main trail.

MAJOR JOHN V. MATHER, AIF, right, served on division staff throughout the campaign.

Personnel in the two groups were not static, but were changed from day to day.

The main Puller patrol and the group headed by Captain Hunt met on the trail shortly after Colonel Puller had departed Arigilupua. Hunt then learned that his company had been ordered from Arigilupua back to Agulupella and that he and his 11 men were to accompany the larger force on its trek to the Itni. The captain spent the remaining patrol time in "confinement" under the surveillance of a lieutenant appointed for the task⁵⁴ while his men acted in the dual capacity of scouts and ammunition carriers.

Following advance patrols, the group made its way rapidly over the tortuous trail southward, climbing in and out of deep gullies and wading through numerous streams. Ammunition carriers for the 81mm mortars found it particularly difficult going and frequent reliefs had to be arranged. No relief could be found from the rain, however, which continued to pelt downward, adding slippery mud to the natural difficulties of the march and making everyone uncomfortable. Despite the hardships imposed by the narrow trail, progress met with approval of Colonel

COCONUTS were rather a novelty in the immediate Cape Gloucester area. This patrol found some in a native village.

Puller, who later said he liked the Gilnit patrol because "everything's on two feet."⁵⁵

Early in its operations the patrol found that the map of the southern area was as incorrect as that of Cape Gloucester. Native villages shown were either named or located erroneously, and it became an important task of intelligence personnel to correct the trail net and village names on the map to correspond with reality.

Following the general route of Agulupella to Arigilupua to Relmen to Turitei to Nigol to Gilnit, the group came across abandoned bivouac areas, supply dumps and occasional enemy stragglers. One bivouac area, filled with the derlict debris of a retreating force, yielded further evidence of the Japanese 141st Infantry's participation in the Philippines campaign: one discarded pack containing a large and dirty U.S. flag.

Supply drops by cub planes were planned for and effected at Arigilupua, Relmen and Turitei. Because of the nature of delivery as well as the absence of field kitchens, foodstuffs for the Marines were limited to K rations, which as a steady diet for several weeks became boring.⁵⁶ The standard menu for native carriers was rice and corned beef which was provided only for them. Not a few of the Marines watched enviously as the natives ate their hot meals, using palm fronds for plates.

Not all of the drops had to do with food, however. As one officer responsible for supply recalls a novel incident that caused some bewilderment at the time:

One of the requests for a cub airdrop from Col Puller of mosquito lotion. Since it was marked urgent, the delivery was made within a few hours although there were those in the Division CP who privately wondered why the Colonel, who had often expressed some contempt for what he considered the luxuries of campaigning, had changed his mind about mosquito lotion. A patrol member queried at a later date was somewhat amused. "Hell, the Colonel knew what he was about. We were always soaked and everything we owned was likewise, and that lotion made the best damn stuff to start a fire with that you ever saw."⁵⁷

Advance elements of the Gilnit Group reached the patrol objective 9 February and the remainder of the Puller force moved in the following day. Nigol, the last village before reaching Gilnit, they found to be one of those rare garden spots which were occasionally seen on New Britain.⁵⁸ A small coconut plantation located on a grassy hill with a gentle slope down to a bend in the Itni River, a lazy appearing (but actually deep and swift) brownish-green waterway about 100 yards across. The Japanese had utilized Nigol as a supply base and the enemy tents still standing contained blankets, numerous documents, quantities of ammunition, weapons ranging from 75mm guns to .25 caliber rifles, cases of seaweed and sacks of rice, barley and soy bean flour.⁵⁹ The documents were sent back to division, but the remainder of the enemy supplies, with the exception of the blankets and 22 sacks of rice, were either burned or dumped into the Itni.⁶⁰ The salvaged blankets and rice were turned over to ANGAU.

Proceeding southeast a short distance from Nigol and remaining on the Itni's west bank, Colonel Puller bivouacked his forces directly across the river from Gilnit and sent a small patrol over to the village in one of the numerous boats found there by his forces.⁶¹ Upon the group's return from the

THE GILNIT PATROL passed through many native villages and all of them looked the same.

objective Puller informed division that Gilnit was abandoned and presented no signs of enemy occupation for the past year. He radioed that he had found a total of 34 barges and boats in the Nigol-Gilnit area, as well as quantities of weapons and ammunition, but that all signs of the Japanese were at least a month old.⁶²

A patrol of platoon strength was then dispatched to Cape Bushing, but it returned the same day with the information that it had found indications that the enemy had been there but no evidence that they were still in occupancy.

Under orders to remain at Gilnit 48 hours and then retrace his steps, Puller reported the destruction of enemy supplies and equipment and headed back toward the perimeter.

At Turitei, however, he received instructions to return to Gilnit. In order to insure that western New Britain was secure, Sixth Army proposed sending a patrol from the Army units at Arawe to contact the Gilnit Group. Army notified division that the Arawe patrol was making its way toward Gilnit when it was held up by enemy action at Attulu Hill, a short distance east of the Itni. It became Puller's job to send a reconnaissance patrol there.⁶³

Basing half his force at Turitei and the remainder in the bivouac area across from Gilnit, Colonel Puller had a Japanese boat repaired⁶⁴ and dispatched first natives and then Marines to Attulu Hill.⁶⁵ They brought back identical reports: no indications of Japanese occupancy for at least a month.

To establish its own western contact with the Gilnit Group, as well as to augment that organization's supplies, division meanwhile ordered a rifle platoon from 1/7 to proceed by water around the western tip of New Britain and up the Itni. Two LCM's were utilized for the journey. The platoon left Cape Gloucester on 13 February, spent the night at Aisega and arrived at Puller's bivouac area the next day. It returned to Cape Gloucester two days later.

Leaving one rifle platoon in the area across from Gilnit to wait for the Army patrol, Colonel Puller on 16 February began his march back to the perimeter. He sent a message to division in which he declared that in his opinion there had been no enemy forces in the Attulu Hill section for at least two months and that no effective enemy forces existed west of the Runglo River. A few hours after his departure he was followed by the remaining platoon, the Army patrol having established contact by sending a detachment in boats up the Itni.⁶⁶

The Gilnit Group returned to the lines 18 February with the exception of 1/5 which was ordered to garrison Agulupella and protect a radar unit which was being installed. That company was relieved a few days later by L/1 and returned to its battalion, now well on its way toward Iboki.

Although it had met no effective enemy forces, the Gilnit Group had accomplished the secondary mission of the division: clearing western New Britain of the enemy from Borgen Bay to the Itni. It had also demonstrated how a large group of men could carry on effective operations supplied only by air drops. The group had also rehabilitated 1,700 natives, killed 75 Japanese, captured one prisoner and destroyed extensive quantities of enemy weapons and equipment.⁶⁷

With western New Britain secure, the division could now concentrate on the pursuit of the enemy eastward.

Rooke Island

Rooke (Umboi) Island, lying off the Western tip of New Britain, was close enough to Cape Gloucester to constitute a nuisance as long as there was a possibility that any enemy troops remained there. Division was reasonably sure that some 300 Japanese had been on the island but believed that they had evacuated prior to 1 February. On 5 February Sixth Army warned the 1st Division that it would shortly be required to take over surveillance of Rooke, preparatory to which HMAS Benalla would begin a survey of the island's coasts on 8 February.⁶⁸

THE ROOKE ISLAND landing was unopposed.

Company B, 1st Marines was selected as the landing force for Rooke Island and departed Gloucester at 0900 on 12 February in six LCM's. Its mission was to reconnoiter the island, set up a defense for a proposed radar installation and recruit native labor crews. Landing was without incident, as was, in fact, the entire eight-day vigil the company spent on Rooke. A camp was established at Gizarum Plantation on the west coast, but the patrol's inability to maintain radio contact with division caused a transfer to Aupwell on the east coast.

Personnel of Company B enjoyed fresh fruit that abounded on the island and engaged in extensive patrolling to the northwest and southwest, visiting native villages that bore such names as Gom, Goxom, Aropot, Arot, Maramu and Kabib. Many scattered machine-gun emplacements, abandoned bivouac areas and some remnants of medical supplies provided obvious signs that the Japanese had once occupied Rooke, but patrols found no live trace of the enemy. Off the northwest coast was a Japanese plane that had crashed on a reef prior to the company's landing, and a unit from the Fifth Air Force journeyed over to salvage what it could.

It was not the intention of higher echelons to establish a permanent garrison on Rooke. Therefore, on 17 February the company

was notified that it would be returned to Gloucester on the 20th. After guiding a group of Army signal officers to Gizarum Plantation, the Marine patrol embarked in LCM's and rejoined its parent unit, bringing 40 native labor recruits along.

Footnotes

1. An active and important exception was Maj Guy Richards, Asst R-2, 5th Marines, a former newspaper man who had worked previously in Melanesia and was well versed in pidgin English and native psychology. Australians included in addition to Maj Mather and SubLt Wiedeman several ANGAU officers formerly stationed in the region.

2. See Chap II. It may be noted that the division's orders did not specifically stipulate destruction of the enemy force, but this was obviously desirable if not essential to completion of the mission.

3. This junction lay at coordinates 62.2-84.2, by which terms it is referred to in reports and orders occurring during the next several days.

4. Two rifle platoons, 1 LMG section and miscellaneous personnel, including 2 representatives of ANGAU, 2 war dog teams and a radio team from 11th Marines. Dawes.

5. A few natives of this region had allied themselves with the Japaneses, making the loyalty of all suspect. This gave rise to an unusual incident when Capt Parish was obliged to convene a military court consisting of himself, Maj Mather and SubLt Wiedeman to try a native named Karol, formerly employed by the enemy as a boss boy, whom some of his compatriots accused of leading a patrol into an ambush. Karol was acquitted largely on testimony of two Marine members of the patrol in question. (D-2 RofE, 27Jan44, Addendum,) Capt Parish continued to employ Karol as a scout and found him wholly trustworthy. (Maj P. J. Parish, comments on prelim script, hereinafter cited as Parish.)

6. Dawes. 1st MarDiv SAR, IV, 1, states merely that the patrol received sniper fire in the vicinity of Mt. Langla and automatic weapons fire from the ridges of Mt. Munlulu, a short distance beyond it.

7. Dawes. This source estimated enemy strength at 2-3 platoons, with 3-4 machine guns. The patrol suffered no casualties in this action.

8. Forty native bearers carrying rations accompanied the patrol, but these rations proved to be more than 50% spoiled. Ibid.

9. Co K, less Wpns Plt and rear echelon units, plus arty liaison party and Regt'l Scouts; total, 118. 3/1 Hist, App A.

10. Owing to communications failure, Parish did not learn that Stevenson had broken contact with the Japanese. Parish.

11. D-2 Journal.

12. The wounded had to be evacuated over the long and difficult supply route that proved a hampering factor throughout these patrol operations: via east-west trail to Sag Sag, thence by water to the airdrome area. Two died of their wounds in the process.

13. Tactically successful but strategically disastrous. See below.

14. Patrol led by Lt J. S. Stankus of R-2 Section and consisted of 6 R-2 scouts, an Army sergeant handling a scout dog, an artillery spotting team, a demolitions man, 2 officer observers, 2 rifle platoons and the machine gun and 60mm mortar sections of Co A.

15. J. S. Stankus.

16. These units now comprised the ADC Group, Gen Shepherd continuing to exercise command in the Borgen Bay area until 9 Feb.

17. Indications to this effect had already appeared in captured documents and prisoner of war interrogations, but the Marines did not learn the full Japanese plan until discovery of Gen Matsuda's headquarters many days later than events related here.

18. "The enemy's route of withdrawal . . . was first discovered on 26 January, and thereafter the ADC Gp was eager to pursue and intercept the enemy along the North coast. Division had other plans however, was preoccupied with the junket of Puller's armed forces (the 'Itni Patrol'), and only later undertook the leapfrogging operations in the direction of Rein Bay, when the ADC Group was disbanded." Ltr Maj Frederic Peachy to CMC, 25Mar52, hereinafter cited as Peachy.

19. Company B, which remained in occupancy of the Natamo positions.

20. "During this day, 1-5 received an air drop of communication wire from a Piper-Cub plane. Rations were delivered by a carrying party from B-1-5 from Natamo Point. Several small ammunition dumps were discovered and destroyed." (Col W. H. Barba, comments on prelim script, hereinafter cited as Barba.)

21. As shown on Map 15. The Government Trail and so-called Corduroyed Trail (actually corduroyed only a short distance inland) join a little north of Nakarop.

22. D-2, 31Jan44. This building evidently served as a communications center and staff headquarters. Matsuda's personal quarters were located off the main trail roughly midway between Magairapua and Nakarop, its approaches so skillfully concealed that patrols did not find it until several days later. Well camouflaged bivouac areas infested this entire region.

23. D-2 Journal, 1Feb44.

24. For some peculiar reason, Matsuda's staff did not destroy its papers but merely threw them in a slit trench and buried them. The Marines had little difficulty in finding them and digging them up, thus acquiring virtually the complete 65th Brigade office files, including copies of a citation the brigade had received for its action on Bataan.

25. 1st MarDiv SAR, IV, 9.

26. D-2 Journal, 3Sep44.

27. Specially trained to handle landing craft in Army amphibious operations, units of this brigade served with high efficiency throughout the Southwest Pacific Theater and Philippines. Personnel who served in the Cape Gloucester campaign came to regard themselves more as Marines than soldiers; after the war one reserve officer of field rank attempted to effect his actual transfer to the Corps. (Statement of BrigGen J. T. Selden, 5May49.)

28. D-2 Journal, 10Feb44.

29. 1st MarDiv SAR APPEASE, 1. This source describes the technique developed as follows--"The plan employed was that of pushing a succession of fresh platoons along the coastal track and then landing a company from LCM's at strategic points en route after the platoons had secured them."

30. Barba.

31. Ibid. For change in Japanese plans see sub chapter following, also Chap XI.

32. 1st MarDiv SAR, IV, 10.

33. Ltr MajGen J. T. Selden to author, 7Mar52.

34. "Katabasis. Literally, a going down; specif., the return march to the sea of the Greek auxiliaries of the Anabasis; hence, any similar retreat." Webster's New International Dictionary, 2d Edition.

35. Except as otherwise noted, detail concerning Sato's operations derives from a diary captured subsequently. The writer was never identified, but internal evidence indicates that he must have been an officer very close to the Colonel, or possibly Sato himself. Translation used as reference appears in ATIS 939, 1.

36. POW 151.

37. Specifically TSUKI Op Or-A-No. 84, 21Jan44. modified and expanded by several supplementary orders. Translated in ATIS 883, 3.

38. Lt Hanahara had commanded the Machine Gun Co of 2/141, but this suicidal group that held off the 5th Marines for so long appears to have been a composite of men from many units, as were most Japanese groups by this time.

39. Monograph No. 128, (Army) Southeast Area Operations Record Part IV, Supplement 1. This work was prepared under U.S. auspices in July, 1949 by ex-LtCol Isamu Murayama, IJA, 17th Div opns officer on New Britain; hereinafter cited as Murayama. Because its author remained far from the fighting, he is highly untrustworthy on operational matters, merely perpetuating the errors and outright fiction contained in reports reaching 17th Div Hq. However, he appears to have been an excellent staff officer, and his comments on matters known to him at first hand shed much light on Japanese activities. Probably they are as reliable as his memory and the scanty sources available to him permit.

40. POW 121.

41. POW 131.

42. A Japanese diary captured at Talasea describes the passage of this unit through that place on 21Feb 44. ATIS 906,5. A map in Murayama indicates evacuation of 2/53 from Aria River to Volupai on 7Feb, but this appears improbable in view of date of departure from Borgen Bay and arrival at Talasea.

43. This portion of New Britain remained unexplored inland of the coast. Villages and trails known to exist were located mainly by guesswork on the maps prepared for the opposition and often appear in different places on different maps, in one case in different locations on the same map. Those shown on Map 20 represent the best approximations possible by comparison of existing maps with Japanese descriptions and U.S. patrol reports. That the two often used different names for identical places does not help.

44. POW 1000.

45. ATIS 939, 1. At this time Maj Komori was still operating in front of Arawe on the island's southern coast. See Chaps IV, IX.

46. Murayama.

47. POW 1000.

48. D-3 Journal.

49. Ibid.

50. Actually cub planes were tried before the B-17's but proved inadequate to supply the big Gilnit Group as originally set up. (Ltr Col W. J. Piper to CMC, 19Feb52.) An account of the many uses the small cub planes were put to and their effectiveness during the Gloucester operation appears in Appendix V.

51. Interview with Maj George P. Hunt, 21Feb52.

52. The patrol encountered one sick Japanese soldier who was sent back to Puller at Agulupella.

53. Although the group was much reduced, it still contained some personnel who thought that they were extraneous to the operation and were not enthusiastic over their presence. One of these, a chunky artillery lieutenant, was ordered along by Puller because the patrol commander thought the walk would do the artilleryman's girth some good. The patrol was well beyond any assistance that could be rendered by artillery.

54. By the time the Gilnit Group returned to the perimeter Col Puller dropped the arrest charges against Capt Hunt, but the latter was admonished to watch his aggressiveness in the future. In the Peleliu operation later in 1944 Puller was Hunt's regimental commander and recommended him for a Navy Cross for his actions there, which was duly forthcoming.

55. McMillan, op. cit., 214.

56. The bulk of the K rations received contained off-brand cigarettes which the Marines tried to utilize as barter with the natives. After a few puffs the natives, who clamored for better-known brands, refused the smokes, even as gifts. This confirmed the Marines' already dim view of the off-brands.

57. LtCol J. S. Day, comments on prelim script.

58. Col Katayama's CP had been located at Nigol prior to his movement northward, and 3/141 had been activated there. See Chap IV.

59. Report by LtCol Puller to MajGen Rupertus, 1st MarDiv SAR, IV, Annex H, hereinafter cited as Puller Report.

60. Some of the Marines took the opportunity to sample the seaweed before it was destroyed inasmuch as they had often heard it was a Japanese staple. It did not, however, meet with enthusiasm. Interview with Col W. J. Piper, Jr., 18Jan52.

61. Ibid.

62. Puller Report; D-3 Journal.

63. D-3 Journal.

64. Puller argued with division that he had no way to cross the Itni because he had destroyed the 34 boats and barges. A survey of the "destroyed" craft, however, produced one which was easily repairable and was utilized for the crossing. That tends to raise some question as to the extent of destruction of the remaining 33.

65. This hill, a short distance inland of Cape Bushing, had evidently been location of 1/141 headquarters prior to that unit's departure for Arawe. See Chaps IV & IX. As described by one visitor: "Attulu Hill had apparently accommodated a sizable force at one time. Permanent huts, a hospital of sorts, and considerable evidence of captured U.S. supplies (from the Philippines) were found. But no sign of recent activity." Parish.

66. Puller Report.

67. Ibid.

68. D-3 Journal.