The battle for the Solomon Islands remains one of the Second World War’s most important campaigns in the Pacific Theatre. It halted Imperial Japan’s advance and brought the first Allied victories on land. Following the devastating Japanese defeat at Midway, Allied landings on the Solomon Island of Guadalcanal on 7 August 1942 struck the next blow against the Axis forces. The ensuing island-hopping jungle fighting between Allied soldiers and outnumbered but tenacious Japanese occupiers has become infamous. Traces of this brutal episode linger as coral-encrusted shipwrecks and an annual ‘iron harvest’ of unexploded shells. For all the fame of the Solomon Islands campaign, though, parts of this story remain untold – moments never recorded by the Allies nor discussed by the defeated Japanese; a legacy of occupation known only from forgotten ruins in deep rainforest.
Island conflict archaeology
Kolombangara is conspicuous within the Solomon Islands’ Western Province. Its jungle-covered slopes rise to a cloud-shrouded summit 1,770m above sea level, leaving it towering above the surrounding atolls. In late 1943, Kolombangara became the seat of Imperial Japanese power in the Solomon Islands and was fortified to resist the Allied advance northwards until Japan’s fortunes could be restored. This ‘turning point’ was not to be, however, and by August 1943 Kolombangara was encircled by Allied forces, eventually compelling its Japanese occupiers to evacuate. In their haste to escape, they abandoned everything too heavy to carry within dugouts that had been prepared to repel an invasion that never came.
Those artefacts lay untouched for 77 years, before being rediscovered by an archaeological expedition in February 2020. It was led by Nikolaus Hochstein Cox, who had been working as a Cambridge Archaeological Unit archaeologist, and consisted of Major (retired) Andy Hawkins, MBE QGM, conflict archaeologist and Durand Group founding member, and Martin Potts, expedition cameraman.
This survey came about after Nikolaus’ friend Robert Prebble worked on Kolombangara in 1995, becoming the first non-islander to be shown a stone-carved dugout, which Prebble remembered being littered with artefacts and described as a Japanese field hospital. Despite this first-hand account of significant remains, there was little published information about how the island slotted into the larger picture of the Second World War – Kolombangara had been relegated to the footnotes of military history. But even from this sparce evidence a story eventually emerged, involving a determined Japanese general, and a mystery concerning a unit of troops among his forces.
Throughout the Solomon Islands Campaign, General Minoru (Noboru) Sasaki was renowned, even by the Allied forces whose advance he stymied, for his refusal to yield Japanese-held territory without a fight. From June 1943, this Hiroshiman cavalry officer and tank commander led Japan’s ‘Southern Detachment’, which was tasked with mounting delaying actions on New Georgia and Arundel (now Kohiqo) Islands, south of Kolombangara.
Sasaki managed to hold off four Allied divisions with a significantly smaller and under-equipped Japanese contingent for a remarkable length of time. However, on 5 August 1943, after the Americans had taken New Georgia’s Munda airfield, Sasaki fell back to Kolombangara. Alongside his command, surviving documents indicate he withdrew the 13th and 229th Infantry Regiments, 10th, 52nd, and 58th Artillery Battalions, 17th Military Police (the notorious Kempei-tai), and the 8th Combined SNLF – a detachment of the Kaigun Tokubetsu Rikusentai, or Special Naval Landing Force (SNLF), of the Imperial Japanese Navy.
This increased the Kolombangara garrison to 12,000 men. The island’s north-eastern coast became the 229th Infantry’s regimental headquarters, which monitored enemy activity in ‘The Slot’ – the wartime nickname for New Georgia Sound. Meanwhile, Kolombangara’s southern shore hosted a combined army and navy presence focused on an airfield beside the Vila River. This became the hub of Japanese control in the Western Province, hosting the 13th Infantry’s regimental headquarters, alongside the artillery battalions and Kempei-tai, the 12th Company of the 229th, and reputedly the Navy’s own 8th Combined SNLF. Such tempting targets attracted daily Allied artillery and aerial attacks.
With New Georgia and Arundel in Allied hands, Sasaki assumed that Kolombangara would be next. Previously, the Allied strategy had been to advance one island at a time without ‘leapfrogging’ Japanese-occupied territory, making Kolombangara the clear target. Sasaki’s men dug in, constructing gun emplacements and dugouts to protect the airfield and garrison. Allied intelligence – in the form of maps now available in the National Archives as well as Cambridge’s University Library (accessed by Amy Bigwood, expedition researcher) – followed this defensive build-up with interest, plotting the positions of the gun emplacements, the hospitals and living quarters, vessel moorings, and aircraft disbursement grounds. One notable absence from these Allied maps, though, is the position of the SNLF unit supposedly deployed to the island. If this formidable fighting force had participated in preparations for the defence of Kolombangara, where was it based?
Wither on the vine
As it turned out, the Japanese soldiers on Kolombangara laboured in vain. American Admiral Halsey had seen his men suffer appalling casualties on New Georgia and wished to avoid further slaughter. His men were exhausted and suffering the effects of prolonged combat in gruelling conditions against a determined enemy. So Halsey decided to bypass Kolombangara and strike the lightly defended enemy positions on Vella Lavella Island, to the north- west, thereby sparing his men the ordeal of assaulting the fortified nexus of Japanese operations. This plan was executed on 12 August, when an assault on Vella Lavella saw its disorganised defenders overrun. Success allowed the Allies to consolidate their control of the surrounding islands, leaving Kolombangara’s 12,000 defenders untouched, but also isolated from their nearest comrades at Bougainville.
Halsey expected the surrounded soldiers to ‘wither on the vine’, but 12,000 troops amounted to a force the Imperial General Headquarters could not afford to abandon. During the night, from 28 September to 3 October 1943, the entire garrison was evacuated in Daihatsu landing barges and torpedo boats, which ran the gauntlet of watchful American cruisers in The Slot. The evacuation proved costly for the Japanese: 29 small boats were sunk and one destroyer was damaged, but 9,400 men were rescued. These survivors were redeployed to Bougainville, where they continued to resist the Allies until Japan’s capitulation in 1945.
Just as General Sasaki faded into obscurity after the war, so too his measures to defend Kolombangara disappeared beneath its jungle. Although the 1st Battalion, 27th Infantry (United States Army) came ashore three days after the Japanese evacuation, they moved on swiftly, finding only discarded artillery pieces, abandoned Mitsubishi Zeroes on the runway, and Japanese troops who had been too unwell to evacuate. The Americans turned the island over to a contingent of Fijian soldiers, who in turn handed it back to British colonial authorities. For the British, Kolombangara was a place of lucrative plantations; former Japanese fortifications on the swampy land between the Vila and Teme region were of no significance.
By 2020, when Nikolaus contacted Ferguson Vaghi – the director of the Kolombangara Island Biodiversity Conservation Association – about the potential for undertaking conflict archaeology, the jungle had long since obscured the derelict Zeroes, while in the early 2000s Bougainville rebels reportedly seized the littoral anti-aircraft guns. Even so, elements of Sasaki’s defensive preparations were rumoured to have endured, in the form of dugouts that were burrowed into the hills of Teme and replete with in situ artefacts. According to the Allied maps, this area to the east of the Vila was never considered an enemy strongpoint. But it appeared possible that the dugout Robert visited in 1995 was part of an extensive subterranean defensive system. To assess this, contact was made with a local landowner, Stenrick Riapitu, and it was agreed that we would undertake the first archaeological survey of his land.
Five dugouts at Teme
Between 28 February and 14 March 2020, our survey located six Japanese dugouts and we entered five. All six dugouts ran along the inland face of a coastal ridge, on a south-east to north-west trajectory. This ridge comprises resilient igneous rock, allowing most of the stone-carved passages within to survive 77 years of disrepair, even though many of the entrances – dug into softer topsoil – had collapsed. In all cases, these subterranean structures had two access points, ensuring that occupants would not become trapped if one was captured or blocked during fighting. Examining the dugouts indicated each structure served a specific wartime purpose.
We employed an identical survey process in all of the dugouts. Andy and Martin photographed the artefacts in situ, while Nikolaus recorded the minimum number of each artefact type visible. Unique items were extracted and cleaned outside. Martin took scaled shots of these artefacts, as well as photogrammetry photos, allowing them to be rendered in 3D, so that their shape could be recreated for post-excavation analysis without the original object being removed from Kolombangara. Andy and Nikolaus also planned the dugouts and determined their depth. Finally, all of the artefacts that had been extracted were returned to their original positions. It had been agreed with Stenrick that the team would not retain any objects.
Starting at the south-easternmost dugout on the ridge, investigation revealed it contains three carved alcoves where military equipment was stored. The alcove closest to the south entrance, which offers rapid access to the shore, still contains 37mm armour-piercing high-explosive shells. These munitions were stockpiled there so the defenders, who expected an amphibious Allied assault, could run the shells to coastal gun emplacements. The other two alcoves lay deeper within the dugout; these contain glass sake and Dai-Nippon beer bottles, medicine bottles, aluminium Navy water canteens, hand grenades, 6.5mm bullet cartridges, Type 93 No. 3 gasmasks and filters, belt buckles, and Bakelite toothbrushes. This assortment of small arms and military-issue personal equipment identifies the dugout as the quartermaster’s store; when Japanese servicemen needed ammunition, replacement gasmasks, or even a new toothbrush, they came here.
Following the ridge north-west leads us to another dugout. Although much of its tunnel has collapsed, a dome-roofed chamber opening off the passage remains intact. Within, benches had been carved out of the stone, allowing troops somewhere to sit, suggesting that this chamber served as an air-raid shelter. A short tunnel lay at its far end, providing storage space for weapons or food. No artefacts were found there; evidently, the air-raid shelter’s supplies were taken during the evacuation.
The next dugout contains two storage alcoves reminiscent of those in the quartermaster’s store, and seemingly acted as an ammunition store. There, as in all five dugouts, sheets of corrugated iron roofing still hang in places, while vertical slots were carved into the walls where wooden posts were once inserted to support this roof. Although mud had built up over the original floor, we spotted eight hand grenades – with more presumably lying invisible within the earth – and two cylinders that probably contain TNT: an explosive. We suspect this played a part in Japanese preparations to boobytrap the dugouts during the evacuation – preparations that were thankfully never finalised.
Both of the entrances to the following dugout had collapsed. Although a day of excavation proved sufficient to clear other clogged access points, in this case heavy seasonal rains frustrated our efforts, meaning this dugout remains sealed.
Next is the field dressing station, which was shown to Robert Prebble in 1995. Though locally called the ‘hospital’, the space is too small to cater for a garrison of 12,000 men. Indeed, Allied records indicate a larger field hospital inland from the Vila served that purpose. The dressing station was instead presumably intended to provide immediate medical assistance to the men defending the Teme rainforest perimeter. The ground outside the dugout is littered with beer and sake bottles, as well as the remains of stretcher beds. This distribution could be explained by patients normally resting outside, and only being taken underground for urgent treatment. A narrow corridor at the back of the dugout seems to have served as a surgery, where a surgical stand, leg brace, and alcohol and medicine bottles were discovered. The higher proportion of sake to medicine bottles suggests that towards the end of the dugout’s period of use alcohol was being utilised in operations, possibly as anaesthetic. Another interesting feature is that the dressing station seemingly had two distinct periods of use, as the eastern wall of the ‘surgery’ had niches in the shape of pickaxes and shovels carved into it. Although the in situ artefacts suggest a medical role, creating dedicated spaces to hang digging equipment hints at an earlier use associated with tunnelling.
We named the northernmost dugout ‘Snake Tunnel’, which seemed appropriate from both its serpentine layout and a current reptilian resident, as no clear wartime function could be determined. Unlike the other dugouts, Snake Tunnel has no chambers budding off its unusually winding primary passage. Furthermore, a 10m length in the middle is both oddly narrow and low-roofed. Although this tunnel was occupied, as evidenced by finds including a jika-tabi’s split-toed rubber sole typical of an SNLF tropical uniform, and an Imperial Navy Type II/III helmet, it seems likely that the dugout was still incomplete when the evacuation occurred.
Another find from Snake Tunnel was a porcelain sake bowl. Although broken, its base remains intact and carries a maker’s mark: a painting of a waterfall flowing below a stratovolcano resembling Mount Fuji in Japan. Finding this bowl amidst the military equipment was humbling. While archaeologists become intimately acquainted with the lives of the peoples whose sites they excavate, a degree of facelessness can be a feature of conflict archaeology dealing with mass-produced military equipment. But discovering someone’s personal possession, brought with him from his homeland, reminds us that the ranks of even the most uniform armies are filled by individuals.
Who were the SNLF?
Several artefacts – the Type 93 No. 3 gasmasks, Type II/III Navy helmet, aluminium water canteens, and jika-tabi – are of particular significance, as they were exclusively issued to the SNLF. This discovery resolves the historical discrepancy between the written records indicating an SNLF unit was present on Kolombangara, and the Allied military maps, which show no trace of these soldiers’ quarters. Apparently, Allied intelligence’s focus on the Vila airfield meant they never looked far enough east to detect the SNLF digging positions in the rainforest at Teme.
The SNLF were effectively marines, that is members of the Imperial Navy who received special infantry training, allowing them to act as the Navy’s personal army. The Japanese Navy and Army had a deep-seated mistrust of each other, which can be traced back to the feuding rivalry of samurai clans, making joint operations uncommon. In the Pacific War’s early stages, special naval units launched amphibious assaults on the Philippines, Indonesia, Papua New Guinea, and even the US Aleutian Islands. Reinforced by naval parachutist units, the SNLF seized Allied airfields and overran garrisons wherever they landed, earning a fearsome reputation.
The Navy dreamed of using this amphibious assault force to cross the Pacific and deliver Japanese victory, but the Battle of Midway ended those ambitions. Losing at Midway crippled the Navy, limiting their inter-island manoeuvring capability. As Japan pivoted to the defensive and Allied air and sea superiority grew, a naval landing force became irrelevant – without a strong navy capable of offensive actions, the SNLF had little ‘special’ purpose. By the later months of 1943, the SNLF were being assigned defensive roles protecting island installations. It was this unglamorous task that the 8th Combined received upon arrival in Teme.
Given the fierce inter-service rivalry, the presence of SNLF artefacts at Teme rules out the possibility of the dugouts being inhabited by the Army. Although no regimental insignia have been identified, by a process of elimination we can determine these SNLF must have come from the Kure Naval Base in Hiroshima and the Yokosuka Naval Base in Kanagawa. From 5 November 1942, two units from these bases – the 6th Kure and 7th Yokosuka – served as a single unit, the 8th Combined, under Rear Admiral Minoru Ohta. As the war progressed, the consolidation of SNLF units became increasingly common, due to catastrophic losses inflicted by tropical disease, malnutrition, and combat. By August 1943, only the 8th Combined were still active in the New Georgia area – meaning they are the only SNLF who could have excavated the Teme dugouts.
The one-upmanship between the Imperial Navy and Army, combined with Imperial Japan’s nationalistic militarism, inspired competitive ardour for the samurai code of honour – bushido. The question of which force best embodied the samurai spirit persisted throughout the Second World War, with the SNLF ethos reflected in a 1943 intelligence report by the 2nd US Marine Division, which attributed ‘greater tenacity and fighting spirit’ to these troops, when compared to regular Imperial Japanese Army soldiers. As this honour code viewed capture as worse than death, it also coloured attitudes towards and treatment of enemy soldiers. Allied accounts of a foe treating surrendering soldiers as sub-human are well known. As such, the disturbing consequences of this code provoked anti-Japanese sentiment throughout the war and sustained stereotypes long afterwards. A mentality to ‘fight to the very last’ remains a source of morbid fascination for Western historians. But focusing on this aspect alone can result in the Imperial Japanese being portrayed as simply an inscrutable ‘other’ in historical accounts.
Seeing how the SNLF lived, slept, tended their wounded, and even brushed their teeth, brings an extra dimension to their story. The Teme tunnels and artefacts are a tangible reminder that alongside cultural practices their adversaries found hard to fathom, members of the SNLF experienced more recognisable impulses. The archaeology suggests soldiers who carried keepsakes from home, sought solace in alcohol even as supplies ran low, and took refuge underground during bombardments. Such awareness is invaluable for ensuring that the horrors and atrocities of war are properly examined. In this way, conflict archaeology can reveal far more than just the nature of conflict. While our survey has allowed us to better understand the movements of Sasaki’s Southern Detachment in 1943 and recognise the previously unknown extent of Kolombangara’s defensive preparations, we have also gained an insight into the lives of the men of the 8th Combined SNLF who, for a few months, called these jungles home.
Looking back to see forwards
Finding well-preserved and previously unrecorded ruins rich in diagnostic in situ artefacts is an archaeologist’s dream. Our findings provide a high-resolution snapshot of wartime life for Imperial Japanese soldiers. Archaeology working in tandem with history has also filled a blank in the wartime record by determining where Sasaki’s naval forces were quartered.
We feel the Japanese military possessions are an intrinsic part of the dugouts, providing a window into the lives of the men who were stationed here. It was a conscious decision that none of these artefacts would be taken, except to photograph them in daylight before returning them to their place of discovery, ensuring that their story remains preserved. Removing them would only add to the roster of wartime memorabilia found in collections worldwide.
Coupled with this is the desire to work in accordance with the wishes of local communities. Stenrick hopes to open his land and dugouts to tourists, and we collaborated closely with him, ensuring that he was kept abreast of discoveries and also advised on the procedures for the removal of munitions. These artefacts belong to him and his family. It will be their decision whether any are made available to international museum collections or remain in situ as an example of a preserved wartime site.
The discovery of personal possessions alongside military artefacts reminds us that fighting was only one part of these soldiers’ lives. While historical accounts often focus on attitudes that set the SNLF apart from their adversaries, the archaeology can bring greater depth to their story. Through an examination of the Teme ruins we can not only better understand the Japanese wartime experience, but perhaps come to comprehend and improve the world we find ourselves in today.
ACKNOWLEDGEMENTS Sincerest thanks and ngā mihi to Dr Manfred Hochstein, Ron and Diana Scott, Dr Gino Caspari, James Jordan Winstanley, Karen and Johan Morreau, John Hooker, mum and dad, and Kirsten Scott for making this expedition possible. Email: kolombangaraforgottenfortress @gmail.com