In depth | Tabu-dong — how a regiment that stopped held the valley

2026-10-13 · 1950-08-18 – 1950-08-25

Overview

Part 1: Tabu-dong (August 1950) showed how valley and ridges held each other up; this piece looks at the decisions inside that story. The US 27th Infantry attacked on 18 August, stopped when the ROK regiments on the ridges could not keep pace, and from that spot threw back seven nights of attacks by the North Korean (NK) 13th Division [1]. It won by noticing that the enemy came down the same road every night and stacking mines, artillery, tanks and bazookas on that road. Below, terrain, weather and the scenario's phases p1–p8 (with what-ifs p2x and p5x) are taken in turn.

Black-and-white panorama of a narrow valley between steep hills, a road along its floor and the label SOI-RI at lower left — Photo 1. The 'Bowling Alley' valley (1950). The 27th Infantry halted just north of Soi-ri, lower left — Photo: US Department of Defense files (in Appleman, 1961), public domain (PD-USGov-Military), via Wikimedia Commons

Strategic and operational context

In August 1950 North Korea massed five divisions north of Taegu against ROK and US divisions that were not linked to one another; on 18 August its mortar rounds reached Taegu station [3]. ROK Army orders published by the National Archives show the intent: No. 138 (17 August) told the ROK 1st Division to secure its main line and attack north; No. 139 (18 August) expected the enemy main body down the Kunwi–Tabu-dong–Ch'ilgok axis and attached the 10th Regiment [4][5]. Lt. Gen. Walton Walker sent his reserve, the 27th Infantry, north of Taegu "without delay" on the 17th and next day ordered a limited attack to restore the ROK line near Sokchok, about 6 km north of Tabu-dong [1]. The plan assumed the ridges would advance with the valley. That assumption broke on the afternoon of the 18th.

Terrain and weather in detail (OAKOC)

Detailed sketch of the valley: NK attack axis from Sinjumak, 27th Infantry companies and Pershings north of Soi-ri, the double minefield, reserves behind, ROK regiments on both ridges, the 21 August patrol — Figure 1. The valley front in detail (sketch, partly estimated). Places from OSM [18]; company positions estimated from Appleman, ch. XIX — Own work, after sources 1, 6, 11

Observation. The Yuhak-san summit area stands at about 816–839 m and Ka-san at 880–902 m, against about 207 m at Tabu-ri and 118 m at the Ch'onp'yong fork [17][8]. The valley falls northward, so the regiment looked slightly down a straight, poplar-lined road with long fields of fire [1]. From ridges 600–700 m higher, though, the whole floor lay open (Photo 1).

Avenues of approach. Armour had two: the road and the stream just west of it [1]. Appleman puts the fork a little over a mile in front of the regiment and Sinjumak on the curve just north of it [1]; Part 1's "short of the fork" should read "north of the fork". Infantry could climb the side valleys to the ridges and work round to the rear, as the 22 August infiltration did [1].

Key terrain. The narrow neck where road and streambed run together, and above all the ridges: Yuhak-san changed hands nine times [8].

Obstacles. Slopes and stream kept tanks on two lanes. At dusk on 21 August the regiment laid antitank and antipersonnel mines with trip flares about 150 yards out across road and streambed, and a surface belt about 100 yards beyond [1]. Uzal Ent, then a platoon leader in the 27th, calls the outer belt a decoy [11].

Cover and concealment. By day the enemy hid tanks in Sinjumak; daylight belonged to US aircraft [1]. ERA5 reanalysis gives highs of 28.5–30.5°C for 17–22 August and sunset near 19:10 [16]; the memorial records heat near 37°C [6], which a coarse grid would understate.

The moon. First quarter fell at 00:35 Korean time on 21 August [15]. On the night of the 20th a half moon stood high after dusk and set near midnight, matching Appleman's bright moon silhouetting enemy tanks [1]. The 21–22 August attack began near midnight as it set, and the defenders counted the column by illumination [1]. This is the author's inference.

Organisation and command relationships

Side	Echelon	Unit	Role	Source
UN	Army	Eighth Army (Walker)	Committed the 27th, then the 23d Infantry	[1]
US	Regiment	27th Infantry (Col. John H. Michaelis)	Attack, then blocking position	[1]
US	Battalion	1st Bn (Lt. Col. Gilbert Check): C, A, B	West of road	[1]
US	Battalion	2d Bn (Lt. Col. Gordon Murch): E, F, G	East of road	[1]
US	Tanks	C Co, 73d Tank Bn (M26)	2 on road, 2 in streambed, 4 in reserve (Ent: 3 in streambed)	[1][11]
US	Artillery	8th FA (direct support), 37th FA less A Btry	105 mm; 24 howitzers (Ent)	[1][11]
US	Regiment	23d Infantry (Col. Paul L. Freeman Jr.)	Guarded guns and rear from 19 Aug	[1]
ROK	Division	1st Division (Brig. Gen. Paik Sun-yup)	c. 20 km of ridge	[8][9]
ROK	Regiment	11th	East ridge over the valley (est.)	[7][12]
ROK	Regiment	12th	Hill 837, Suam-san, Yuhak-san	[6]
ROK	Regiment	15th	Hill 328	[6]
NK	Division	13th	Main effort in the valley	[1]
NK	Division	1st	East ridges; one regiment infiltrated 22 Aug	[1]
NK	Division	15th	Yuhak-san, shifted east c. 20 Aug	[3][6]

No single commander tied the US regiment to the ROK division; liaison officers and a few KMAG advisers did, and Appleman states no command relationship [1]. The 8th FA's fire control sat in one fire direction centre (FDC) [1].

Day-by-day account

18 August: attack, halt, first night

The regiment crossed the line of departure behind its Pershings at 1300 according to Appleman, 1530 after road delays according to Ent [1][11]. The enemy main position lay about 2.5 miles behind the outposts. Just north of Soi-ri, Michaelis learned neither ROK regiment had advanced and dug in [1]. That night an F Company bazooka team stopped one tank, another was abandoned after two dud rockets, and Lt. Lewis Millett kept calling artillery with a T-34 fifty yards away [1].

19–21 August: pattern and preparation

On the 19th ROK troops regained ground on the ridges and the 23d Infantry moved up behind the 27th [1]. On the 20th a moonlit attack broke at 150–200 yards, and North Korea shifted its 15th Division away from Yuhak-san [1][3]. On the 21st two platoons and three tanks went a mile up the road and wrecked five abandoned tanks; at dusk the minefield went in [1].

Night of 21–22 August: the Bowling Alley

Nine tanks and several SP guns led the attack; illumination showed nineteen vehicles. A Pershing stopped the lead tank at 125 yards, bazookas stopped an SP gun and a tank, and artillery and 90-mm fire destroyed seven more tanks and three SP guns. It lasted about five hours [1].

A wrecked T-34-85 by a road with a unit name painted in English on it, two helmeted US soldiers beside and on it — Photo 2. A T-34-85 stopped on the Taegu front; the file description places it in the Bowling Alley, 21 August 1950 — Photo: US Army, public domain (PD-USGov-Military), via Wikimedia Commons

22 August: enemy at the back

At about 1000 Lt. Col. Chong Pong-uk, the NK 13th Division's artillery commander, defected and gave away his hidden guns [1]. By noon an NK 1st Division regiment that had come down the east ridge was six miles behind the 27th, firing on five miles of road. Michaelis reported that the ROK on his left had given way; Eighth Army and KMAG checks found them in place [1]. At 1630 two hits wrecked the 8th FA's FDC and six men died; the batteries took over fire control. Walker sent the 23d Infantry to clear the rear [1].

23–26 August: withdrawal and relief

The 23d retook the heights over the guns; at 1335 on the 23rd the enemy blew the road and withdrew [1], and the ROK 12th Regiment took Yuhak-san [6]. On the 24th a recovery team towing a T-34 set off a buried mine, wounding twelve. The ROK 1st Division took over from 1800 on the 25th to 0345 on the 26th. Confirmed NK losses for 18–25 August: 13 T-34s, 5 SP guns, 23 vehicles [1].

Decision points in depth

Decision flow from p1 to p8 with dashed what-if phases p2x and p5x, and a bar chart of rounds fired on 21–22 August, led by 1,661 from B Battery, 8th FA — Figure 2. Decision flow and fires, night of 21–22 August. Phases from the training scenario; rounds from Appleman [1] and Ent [11] — Own work

p1 The flanks aren't coming (18 Aug, 1530)

Knew: neither ROK regiment had advanced; under four hours of daylight remained. Options: press on, halt and block, fall back, or patrol. Decision: halt with refused flanks [1]. Otherwise: pushing on meant meeting the main position at dusk under fire from both ridges; falling back meant meeting the first attack on unprepared ground.

p2 The first night (18 Aug, 2100)

Knew: two tanks and an SP gun were coming into registered ground; a rumour of "dozens of tanks" was weak. Decision: hold fire to bazooka range; bazookas and artillery did the rest [1]. Otherwise: sending reserve tanks up a narrow road at night throws away the ambush.

p2x Tank in the position (what-if)

A tank stops among the foxholes with infantry behind it. Recommended: keep the company down and bring artillery to the line. Millett's fire on the first night and the abandoned lead tank show the basis [1]: a tank stripped of infantry cannot hold ground.

p3 Moonlight (20 Aug, 1730)

Knew: the same road, the same preparation each night; a bright moon coming [15]. Decision: no fire until 150–200 yards [1]. Otherwise: moonlight cut both ways; early fire would have shown the positions and let tanks stop out of range.

p4 White flags (21 Aug, 0700)

Knew: surrender rumours; disabled armour that could be towed off overnight. Decision: a small patrol wrecked it, then mines at dusk [1]. Otherwise: a battalion to Sinjumak would have re-entered a valley whose ridges the enemy held, the risk avoided at p1.

p5 The Bowling Alley (21 Aug, 2330)

Knew: nineteen vehicles coming; ridges attacked too; mines and fires ready. Decision: wait for the mines, then fire everything: in the 2d Battalion sector B Battery fired 1,661 rounds, the mortars 902, 1,200 and 385 [1]. Otherwise: tanks sent ahead would have had their own minefield behind them; counter-battery first would have pulled fire away at the column's worst moment.

p5x Through the minefield (what-if)

A tank slips through an unfinished gap. Recommended: reserve Pershings to the tank, mortars and artillery on the gap. In reality the mines were ready in time [1]; the four reserve tanks existed for this case.

p6 Enemy at your back (22 Aug, 1200)

Knew: the supply road was under fire; the front was firm; a report said the ROK on the left had gone. Decision: hold and ask for help; Eighth Army sent the 23d [1]. The collapse report went up too and was wrong [1]. Ent, who climbed up to check for himself, traces it to an ROK battalion that came down after two days without water [11]. Otherwise: holding and checking the ally should have gone together.

p7 Gun line under fire (22 Aug, 1630)

Knew: the FDC was gone; batteries were firing without central control. Decision: batteries ran their own fires and the headquarters moved; the heights were left to the 23d and aircraft [1]. Otherwise: moving the guns south would have left the valley without artillery that night, and the 37th FA's rear battery was under fire too [1].

p8 The last night (24 Aug, 2350)

Knew: the enemy had blown the road; relief was due next evening. Decision: break the attack in place, relieve at night [1]. Otherwise: a pursuit would have run onto a road the enemy had mined, as the recovery team's accident showed [1].

Doctrine re-read

Engagement areas (FM 3-90, para 8-52). The seven steps run from avenues of approach and the enemy's likely course of action, through choosing where to destroy him, siting weapons and integrating obstacles and fires, to rehearsal [13]. On 21 August the 27th did the first six; no rehearsal is recorded [1].

Obstacle effects (paras 17-20, 17-21). Effects come from obstacles combined with fires [13]. The lead tank stopped at the mines, the column jammed, and the guns hit it there.

Perimeter defence (paras 9-59 to 9-62, 9-68). Units form a perimeter when they cannot tie in with neighbours, patrol aggressively outside it, reinforce the threatened sector, and must make up for its lack of depth against armour [13]. The refused flanks, the patrol and the minefield match point for point.

Reserves (para 9-10). A defence should succeed without its reserve [13]; hence the scenario's low rating for sending the reserve tanks forward.

Mission command (ADP 6-0, 2019) decentralises execution [14]; batteries running their own fires after losing the FDC is a 1950 example.

Relief in place (paras 15-10, 15-16, 15-30). Reliefs are best done in reduced visibility, with obstacle records verified and extra liaison between allies [13]; the night relief of 25–26 August and the recovery-team accident illustrate both.

Animated map of six phases from the 18 August halt to the 26 August ROK takeover; the still frame shows the final situation — Figure 3. Manoeuvre by phase (16-second loop; reduced motion shows only the final situation). Sketch, partly estimated — Own work

Source criticism

Times. Appleman gives 1300 for the attack, Ent 1530 [1][11]: planned against actual, probably. Ent's 0137 for the nineteen-vehicle report [11] suggests a lag between sighting and reporting.

Distances. Appleman's figures (line two miles north of Tabu-dong, fork a mile beyond) exceed the 4.1 km between today's Tabu-ri and Ch'onp'yong-ri on OSM [1][18]; Ent puts the fork 800 yards ahead [11]. Figure 1 compromises at 1.7 km.

Strengths. The scenario cites about 21,500 North Koreans against 7,600 ROK from the press, but the cited 2011 Maeil Shinmun article gives about 21,000 against 8,200 for the ROK division and US 27th together [9]. Neither is an official figure.

Losses. Appleman's confirmed 13 T-34s, 5 SP guns and 23 vehicles cover 18–25 August [1]; the memorial's "about 30 tanks" covers the whole 55-day battle [6]. The 1,300 NK casualties of 21–22 August are an Eighth Army estimate [1].

"The first tank battle". The memorial uses the phrase [7], but Appleman records Marine Pershings and bazookas stopping four T-34s at Obong-ni on 17 August [2], and the Bowling Alley's first clash came on the 18th [1].

ROK regiments. Appleman names the 11th and 13th [1]; Korean sources give the 11th, 12th and 15th, down to battalion positions [6][8], and look more reliable.

The report on the ROK. Appleman and Ent both record its refutation [1][11]; Paik later wrote that Michaelis apologised and came to respect the ROK Army [12].

Several men in police uniforms rest with rifles on a grassy slope; a caption in Chinese characters dates the photo 6 September 1950 — Photo 3. Korean combat police who fought at Tabu-dong, 6 September 1950 — Photo: Office of Public Information photographer (National Archives of Korea), public domain (PD-South Korea), via Wikimedia Commons

Lessons

When the flanks stop, the attack stops. Switching to blocking the valley bought a week.
An enemy's repetition is intelligence. Stack mines, fires and tanks on his pattern.
Fire control is choosing the range. 150–200 and 125 yards separated infantry from armour.
Plan to lose the command post. Batteries that could fire on their own kept the front alive.
Check an ally's flank before reporting on it.

Unresolved questions

The formal command relationship between the 27th Infantry and the ROK 1st Division.
The ROK 11th Regiment's exact position and the exact site of the perimeter.
The meaning of the green flares, unconfirmed in primary records.
The Korean Institute for Military History's account was not consulted directly.

Sources

Roy E. Appleman, South to the Naktong, North to the Yalu, US Army Center of Military History, 1961, ch. XIX "The Taegu Front" (Korean War Online transcription). https://www.koreanwaronline.com/history/Appleman/Chapter19.htm (accessed 2026-10-13)
Roy E. Appleman, same volume, ch. XVII "The First Battle of the Naktong Bulge" (Korean War Online transcription). https://www.koreanwaronline.com/history/Appleman/Chapter17.htm (accessed 2026-10-13)
Yang Young-jo, "Tabu-dong battle" (Dabudong jeontu), Encyclopedia of Korean Culture, Academy of Korean Studies, 2013 (revised 2023). https://encykorea.aks.ac.kr/Article/E0013458 (accessed 2026-10-13)
National Archives of Korea, the Battle of Tabu-dong in the records (Operation Orders No. 138 and 139), newsletter, 2016-06. https://theme.archives.go.kr/next/pages/new_newsletter/2016/06/sub03_1.html (accessed 2026-10-13)
National Archives of Korea, restoration and release of records of ROK Army actions in the Korean War, newsletter no. 102, 2019. https://theme.archives.go.kr/next/pages/new_newsletter/2017/html/vol_102/sub02_2.html (accessed 2026-10-13)
Tabu-dong War Memorial Museum, the Battle of Tabu-dong. https://www.dabu.or.kr/ver2/sub2/sub2_1.php (accessed 2026-10-13)
Tabu-dong War Memorial Museum, the Battle of the Bowling Alley. https://www.dabu.or.kr/ver2/sub2/sub2_3.php (accessed 2026-10-13)
Kyongbuk Maeil, "A fierce 55-day battle that made a sacred ground of national defence" (70th anniversary series: battlefields of North Gyeongsang, 3), 2020-06-18. https://kbmaeil.com/article/202006170281187 (accessed 2026-10-13)
Maeil Shinmun, "Spirit of Gyeongbuk, part 1, love of country 2: at Tabu-dong, Ch'ilgok", 2011-01-14. https://www.imaeil.com/page/view/2011011407033052843 (accessed 2026-10-13)
Korea Defense TV (KFN), "The Tabu-dong battle story we did not know" (YouTube short). https://www.youtube.com/watch?v=LnWR-ld7n78 (accessed 2026-10-13)
Uzal W. Ent, "Korean War: Battles of the Bowling Alley", Military History, August 2000 (HistoryNet, 2006). https://historynet.com/korean-war-battles-of-the-bowling-alley/ (accessed 2026-10-13)
Paik Sun-yup, "A strategy begins to form: When battle doesn't go by the book" (series "60 years with the military"), Korea JoongAng Daily, 2010-08-09. https://www.koreajoongangdaily.com/lifestyle/a-strategy-begins-to-formwhen-battle-doesnt-go-by-the-book/10294478 (accessed 2026-10-13)
Headquarters, Department of the Army, FM 3-90 Tactics, 2023-05-01 (para 8-52; paras 9-10, 9-59 to 9-62, 9-68; paras 15-10, 15-16, 15-30; paras 17-20, 17-21). https://armypubs.army.mil/epubs/DR_pubs/DR_a/ARN38160-FM_3-90-000-WEB-1.pdf (accessed 2026-10-13)
Mission Command Center of Excellence and Combined Arms Doctrine Directorate, "Combined Arms Center launches new mission command doctrine" (ADP 6-0 revision), army.mil, 2019-08-23. https://www.army.mil/article/225414/ (accessed 2026-10-13)
NASA Goddard Space Flight Center, Phases of the Moon: 1901 to 2000 (August 1950). https://eclipse.gsfc.nasa.gov/phase/phases1901.html (accessed 2026-10-13)
Open-Meteo, Historical Weather API (ERA5 reanalysis), Tabu-dong grid cell, 17–26 Aug 1950. https://archive-api.open-meteo.com/v1/archive?latitude=36.06&longitude=128.53&start_date=1950-08-17&end_date=1950-08-26&daily=temperature_2m_max,temperature_2m_min,precipitation_sum,sunrise,sunset&timezone=Asia%2FSeoul (accessed 2026-10-13)
Open-Meteo, Elevation API, Tabu-ri, Yuhak-san, Ka-san, Ch'onp'yong-ri and others. https://api.open-meteo.com/v1/elevation?latitude=36.0451,36.0658,36.0375,36.0804,36.0600,36.0282&longitude=128.5226,128.5041,128.5832,128.5371,128.5300,128.5572 (accessed 2026-10-13)
OpenStreetMap data via the Photon geocoder, place search for Tabu-ri and nearby places. https://photon.komoot.io/api/?limit=4&lat=36.03&lon=128.55&q=%EB%8B%A4%EB%B6%80%EB%A6%AC (accessed 2026-10-13)
Wikimedia Commons, File:The Bowling Alley.jpg (PD-USGov-Military). https://commons.wikimedia.org/wiki/File:The_Bowling_Alley.jpg (accessed 2026-10-13)
Wikimedia Commons, File:T3427inf.jpg (PD-USGov-Military). https://commons.wikimedia.org/wiki/File:T3427inf.jpg (accessed 2026-10-13)
Wikimedia Commons, File:다부동전투참가한전경대원들.jpg (PD-South Korea). https://commons.wikimedia.org/wiki/File:%EB%8B%A4%EB%B6%80%EB%8F%99%EC%A0%84%ED%88%AC%EC%B0%B8%EA%B0%80%ED%95%9C%EC%A0%84%EA%B2%BD%EB%8C%80%EC%9B%90%EB%93%A4.jpg (accessed 2026-10-13)