6

‘China’s Madrid’

The synonym of civil war

In mid-March 1946, after a series of setbacks in the Nationalists’ bid to take over Manchuria, the GMD armies deployed in Shenyang and its environs launched sweeping operations to take selected Communist positions. The prime target of the GMD assault was Sipingjie.¹ The city, which lay about half way between Shenyang and Changchun, was now used by NEDUA as a stronghold to stop the northern advance of the Nationalists. This should have given the US-armed GMD armies the best chance to eliminate the CCP forces in positional warfare. The GMD Manchurian force’s military capability, however, was hampered by a lack of logistic support.

The GMD’s main striking force, N1A, was shipped to Manchuria by US Navy vessels, but logistics difficulties hindered the transportation of some of the most lethal artillery capable of piercing the CCP’s defences.² The problem for the GMD army officers in Manchuria was that they had to reconcile the backwardness of their traditional logistics system with the relatively sudden arrival of military equipment from the United States. In early 1946, the N1A commander Sun Liren pleaded with Wedemeyer to supply cold weather clothing to his army, which operated in the north.³ In February, the Americans supplied 43,000 sets of US military Arctic-type clothing, blankets and sleeping bags to the GMD’s forces in Manchuria with a promise of more to come.⁴ The GMD armies were forced to carry the US winter gear with hired carrying coolies, commandeered pack animals and conscripted mule carts. Some of the soldiers were forced to wear oversized US rubber snow boots on the march in mild weather conditions. The troops’ speed and the security of the armies were therefore compromised.⁵ The GMD’s frontline logistic units at the port of Huludao were stretched out when shipments of US military winter gear continued to reach Manchuria until the summer months of 1946.⁶

The officers of the N1A could rely on the army’s pack animals to transport their US military equipment. However, to their chagrin, thousands of military animals were left in the South China city of Guangzhou, owing to the difficulty of obtaining US logistical support. Logistical assistance for the GMD government was only the fourth priority of the US Army.⁷ US Army logistics officers informed their Nationalist counterparts that only six properly converted LSTs would be allowed to move the livestock for all GMD armies departing for Manchuria. As only six vessels were available for a massive sea transport operation, it could take many months to complete. This meant that the troops and weaponry could reach Manchuria well before the horses and mules. The Americans advised the Nationalists to make use of US Army trucks as an alternative means of transportation but took no account of the difficulty of supplying fuel to truck columns.⁸

As the pack animals were unlikely to arrive any time soon, the officers of the New 38th Division of the N1A had to purchase fodder from the front line and ship the provisions back to Guangzhou to keep the livestock fed. Also included was immediate relief to a contingent of about 2,000 men who had stayed behind with the division’s livestock and were living in abominable conditions. The rescue campaign managed to save only 380 animals, and 62 per cent of the division’s transport capacity was lost.⁹ The loss of manpower and animal power was not unique to the New 38th Division. The US Army records show that the N1A needed to assign a whopping 11 per cent of its total strength (i.e. 4,781 men) to stay behind in order to take care of the army’s horses and mules.¹⁰ The commander of the 93rd Army, by contrast, gave the order to use captured enemy weapons to barter for pack animals with the civilian inhabitants.¹¹ The improvisation was intended to provide a quick fix for the army’s logistical problem, albeit it would not mitigate the risk of army corruption.

The US forces preferred to move the GMD armies to the north sooner than later. This was because the Nationalists needed the Americans’ help to ship almost everything, including green vegetables, to supply those Nationalist forces waiting to board the US ships.¹² But when the Chinese soldiers encountered the American sailors on the American vessels, both parties experienced culture shock. Some American liaison and naval personnel made disparaging remarks about the sanitary habits of the Chinese soldiers without considering that those aboard their ships were the GMD’s best troops and that many of them were English-speaking. Recalcitrant elements of the American personnel considered that the US arms and equipment in the hands of the Nationalist servicemen should be subject to confiscation and looting.¹³ A Nationalist eyewitness report paints the American servicemen accompanying Chinese soldiers aboard the US ships as pirates, who committed theft and vandalism, initiated unprovoked assaults on Chinese officers and accepted bribes.¹⁴ Diaries of the Nationalist junior officers have demonstrated that the GMD forces en route to the north-east were plagued by discipline problems and growing war-weariness among the soldiers.¹⁵

The GMD’s elite corps, not unexpectedly, did not start their Manchurian campaign auspiciously, even though many of them had brilliant fighting records against the Japanese. The stiff resistance of the CCP forces exacted a significant toll on the Nationalists during the intense and gruelling skirmishes in the lead-up to the GMD’s main offensive. After the occupation of a railway town within 60 kilometres to the south of Sipingjie on 3 April, the N1A was ready to launch a major offensive on the CCP’s stronghold.¹⁶ This chapter is about these military operations amid the collapse of the US-backed ceasefire mechanism.

‘Debating societies’

In Chongqing, Marshall’s deputy Alvan Gillem took up the facilitator’s role in the absence of the former and made every effort to broker a deal between the two warring factions for the entry of truce teams into Manchuria. Zhou, however, rejected a Chiang-approved settlement outright, particularly over the provisions allowing the Nationalists to exclusively occupy key localities along a strip 30 kilometres wide on either side of the main railways in the name of national sovereignty.¹⁷ As Marshall had drafted the original proposal, a PRC source considers it the beginning of Marshall’s betrayal of his impartiality, despite the fact that he had explained to both parties before he departed for Washington that he merely wanted to help them to find the simplest way of getting the truce teams into Manchuria immediately.¹⁸ Given that Zhou was well aware of Marshall’s biased stance in favour of the GMD, such a claim literally questioned the purpose of the American’s proposal: rather than resolve the conflict via peaceful mediation, Marshall, under conditions of bias, only wanted to create conditions for the GMD’s troops to advance.

Chiang Kai-shek, in turn, rejected Zhou’s counter-proposal and found the stipulation that the GMD troops could only move into the localities ‘now’ being evacuated by the Soviet forces most irritating. Consequently, both parties quickly deadlocked at the word ‘now’.¹⁹

Given that resentment ran high on both sides after the Nationalist CEC and the ease of NEDUA’s occupation of territory to Shenyang’s north following the Soviets’ withdrawal since mid-March, Zhou was in no mood to compromise. When Gillem began to speculate whether Zhou was using delaying tactics, Zhou simply returned to Yan’an on 21 March, citing communication difficulties with Yan’an as the reason for his departure.²⁰ Chiang felt the time pressure more acutely than Zhou, as the military picture in Manchuria was not looking up for the GMD. He succumbed to pressure on 22 March, informing Gillem that he would make concession.²¹

Zhou flew back to Chongqing on 25 March. Both parties quickly agreed to eliminate all controversial points and signed an instruction regarding the entry of field teams into Manchuria. It authorised the truce teams ‘to bring about a cessation of fighting and to make the necessary and fair readjustments’.²² The instruction certainly was general enough for both parties to subscribe to it, but it did not produce a mechanism for its practical implementation. Previous studies have found the instruction useless because it did not even pick a ceasefire date.²³

From the perspective of the mediator, it should be the goal of the mediator to secure an entirely efficient deal for the disputants, not merely an agreement that is acceptable to both.²⁴ Instructions on detailed matters could have been easier on the members of the truce teams. The problem was that the instruction also went to hundreds of army officers of both opposing factions who might have a different understanding of the local situation. As Marshall elucidated later, an inflexible instruction would have been less adaptive to changes in circumstances.²⁵ The January truce agreement had a clear-cut ceasefire date on which both parties should cease fighting, but it was not as efficient as expected. In high-stakes negotiation scenarios, how a party should behave and how other parties might behave can be entirely asymmetric.²⁶

Byroade soon discovered that the instruction did not provide enough details for his Executive Headquarters to turn it into an action plan.²⁷ The truce teams to Manchuria were either helplessly immobilised or could only accomplish little more than a fact-finding tour.²⁸ When Manchuria became a synonym for civil war in May, individual field team members of the two belligerent parties soon engaged in activities that defeated the entire purpose of sending truce teams. A CCP veteran admitted that the CCP delegates had used their privilege to radio intelligence on GMD troop dispositions to the NEDUA’s headquarters.²⁹ US documents recorded a GMD delegate suspected of being engaged in espionage inside the NEDUA occupied area.³⁰ This pattern of behaviour soon spread to China proper.³¹

These regrettable incidents also exacerbated the diminishing prestige of the Committee of Three, the Executive Headquarters and its field teams. The Executive Headquarters and its field teams had been put under pressure to resolve alleged ceasefire breaches, which made them particularly vulnerable. A report in May 1946 shows that a staggering 1,800 complaints of alleged truce violations had been lodged with the Executive Headquarters since January, and a great number of cases remained unresolved. When agreements on complaints over ceasefire violations could not be reached at the field-team level, the cases in dispute would be referred to the Executive Headquarters. The members of the Executive Headquarters reviewed the cases to no avail owing to the lack of facts. The failure to make decisions distressed all parties of the Executive Headquarters as some behaved like members of ‘debating societies’, as Byroade put it. The Executive Headquarters ended up passing the buck to the Committee of Three, but as matters often could not be decided merely by meetings at the top, the dispute became a merry-go-round that no party could get off.³²

Sipingjie: China’s Madrid

On the Manchurian battlefield, the NEDUA dragged the Nationalists into a two-front battle: amassing forces in Sipingjie and Benxi for a strong defence. While a big battle loomed at Sipingjie, the leaders of the two belligerents had different concerns about the use of force. Towards the end of March, Mao, having recovered from his illness, directed his armed forces with the aim of holding Sipingjie with large-scale positional warfare. He ordered Lin to defend Sipingjie at all costs. Both Lin and Zhou doubted their military capability for conducting trench warfare. They suggested that it was more appropriate to reduce the physical strength of the enemy rather than defending the city and, in the long run, force a peace deal. This strategy received Mao’s support, which concurred completely with the paradigm of offensive accommodation.³³

Chiang, unlike Mao, was less sanguine about the prospects of a military victory. He was more concerned about his armies’ rear area security than a swift and decisive victory over the elusive enemy, despite the fact that he asked about the possibility of taking Sipingjie by 20 April.³⁴ He was aware that his main striking forces in Manchuria, the N1A and the N6A, were inexperienced in fighting the NEDUA,³⁵ although they had been battle hardened in the China–Burma–India theatre during World War II. Chiang’s generals shared his view. The N1A was under orders to stay close to the 71st Army (71A; arriving at combat positions in late March 1946), and the two armies would push forward towards Sipingjie together in a trapezoid shape.³⁶

The caveat was that if the two large friendly combat units stayed too close, it would make it even easier for the CCP troops to break up their formation and start a dogfight: launching attacks to push the Nationalist flank guard units into their already overcrowded main body and causing enormous congestion. While the Nationalists’ formation was split up, the stampeding Nationalist troops became manoeuvring targets of the CCP’s close-range assaults. The CCP forces adopted this tactic in the final phase of the Laiwu campaign of February 1947 in the province of Shandong. An entire GMD army corps of roughly 40,000 men was annihilated in no more than two hours fighting in that battle.³⁷

Therefore, when the 71A approached an area on the outskirts of Sipingjie on 16 April, it maintained a distance of more than 20 kilometres from the N1A. Considering that the 71A was only about 35 kilometres south-west of an enemy-held city, a 20-kilometre lateral distance between the two US-armed friendly forces was a matter of necessity rather than choice if the 71A intended to synchronise the N1A’s frontal attack with a left-hook manoeuvre around and through the enemy positions to the west of Sipingjie. However, the NEDUA was still able to exploit that small gap between the two Nationalist forces and successfully ambushed and paralysed a division of the 71A. At Benxi, the NEDUA successfully immobilised the Nationalists, including a division of the elite N6A.³⁸

Chiang called for a perfectly safe plan for troops to advance in the north-east after repeated military setbacks. He outlined a foolproof plan for his generals: they should shorten their front and avoid the rash advance of troops, thereby enabling them more easily to defend their supply lines even if it would relegate the conquest of Changchun to a lower priority.³⁹

Chiang also scrupulously limited the use of aerial warfare tactics, such as strafing, bombardments and even reconnaissance over the Sipingjie line, for fear of provoking the Soviets.⁴⁰ As a crucial portion of the NEDUA’s defences were constructed inside the closely built buildings at downtown Sipingjie,⁴¹ air bombardment could have easily brought the city to its knees. The American military personnel who visited the front line were perplexed by the Nationalists’ insufficient use of aerial bombardment against the NEDUA positions in Sipingjie.⁴² The following paragraphs will show that Chiang’s generals were overly conservative and cautious during the Sipingjie campaign, which they undertook to observe Chiang’s precept of ‘seeking security first, employing tactical change-up second’ (xianqiu wendang, ciqiu bianhua 先求穩當, 次求變化).⁴³

By contrast, the operational successes buoyed the warlike mentality of the CCP leaders. Mao changed his stance after being informed that Marshall was returning from Washington. As Mao expected that Marshall would immediately step in to halt the fighting, he was more interested in seizing more territory before the fighting stopped than seeking an equitable peace deal. He therefore pressed Lin to capture Changchun and Harbin, the capital of Heilongjiang Province, before Marshall intervened. The NEDUA took Changchun on 18 April. The capture of the old capital of Manchukuo symbolised the Communists’ control of Manchuria, and Mao now considered the city their last line of defence. ‘Turn Changchun into China’s Madrid if necessary’, Mao exhorted Lin.⁴⁴

Perhaps because Mao felt that he needed further military successes to support his Madrid plan, he ordered his commanders to eliminate the N1A. But the enemy was not a paper tiger. After a cautious three-pronged operation, the GMD armies captured some crucial positions on the outskirts of Sipingjie, and the city now came within range of direct Nationalist fire. On 20 April, the N1A began to storm the city, but the NEDUA was able to put up staunch resistance thanks to the overly cautious approach of the attackers. Until the end of April, the Nationalists had Sipingjie essentially under siege, and the NEDUA was entirely on the defensive.⁴⁵

Marshall returned to China on 18 April amid the rapid deterioration of the situation in Manchuria. He was unable to stop the fighting as Mao wished, instead receiving complaints from the two warring parties on ceasefire violations against each other.⁴⁶ Chiang contended that the CCP had not upheld its end of the deal, as evidenced by its capture of major Manchurian cities and its failure to comply with the army reorganisation agreement.⁴⁷ Chiang might have felt that negotiating with the Communists was useless; he could not rule out a ceasefire in Manchuria, as it remained a political and military quagmire for him. Politically, Chiang refused to settle for anything less than his armies’ occupation of Changchun and the large cities to its north, which he regarded as the symbolic recognition of the power of his government.⁴⁸ Such a goal could only be achieved by the destruction of the NEDUA in a major battle to reclaim Sipingjie.⁴⁹ Militarily, however, Chiang showed his willingness to recognise—in light of the military realities in Manchuria—that his armies were vulnerable, if not in danger of annihilation.⁵⁰

In the face of the GMD’s Manchurian dilemma, Chiang’s dependence on Marshall increased, although he knew that the American would bring him nothing but a temporary truce.⁵¹ As discussed, one immediate consequence of dependence on third-party intervention is that the disputants are more inclined to adopt hard-line positions during negotiation. In Chiang’s case, he allowed Marshall to assume control of the proceedings in the January truce, yet the outcome was not what he expected. He therefore outlined a truce proposal that was replete with tough conditions and, if there were an agreement, he requested Marshall’s ‘guarantee’ of CCP compliance.⁵²

One of Chiang’s preconditions for peace was that the CCP would not obstruct his government in taking over the sovereignty of Manchuria, which implied the CCP military withdrawal from all Manchurian cities currently occupied. On the other hand, Chiang insisted that the CCP must earnestly carry out the cessation of hostilities agreement of 10 January and the army reorganisation agreement of 25 February they had previously signed. Chiang also supported the most controversial US-initiated proposal that the American officers of the Executive Headquarters and ceasefire teams should have the final say in the event of a disagreement between the CCP and GMD delegates.⁵³ Chiang would not consider a ceasefire in Manchuria without the fulfilment of all these preconditions.

The CCP leadership, in contrast, insisted that peace talks would not be successful without an immediate and unconditional ceasefire.⁵⁴ In other words, the CCP was seeking a truce deal that was entirely congenial to the consolidation of its newly acquired territories in Manchuria. Zhou also put forward a new demand: the deployment of five CCP army divisions along the major Manchurian cities now under its control, making it a five-times increase against the army reorganisation agreement of February.⁵⁵ On the question of the authority to make a final decision, Zhou advised Marshall that even one small attempt to throw the issue into open discussion would press the Chinese Communists to the wall, and that could be a hindrance to the ceasefire.⁵⁶

Marshall was aware that the civil war in Manchuria had dealt a serious blow to his credibility as a mediator.⁵⁷ One of the warning signs was that Marshall could not find a favourable juncture to call the Committee of Three meetings since he returned from Washington simply because there was no point of mutual agreement between both parties. Instead, Marshall could only settle issues verbally via meeting individually with one party and then the other.⁵⁸ The Committee of Three was in a state of flux since the falling out between the two parties in March and the deterioration of the military situation in April in Manchuria. The committee changed four GMD members in only a month. It began with Chiang’s decision to send his military protégé Chen Cheng to replace Zhang Zhizhong regardless of protests from Gillem. Chen was quickly succeeded by Yu Dawei, the minister of communication, and then Xu Yongchang, the director of military operation. Some of these officials were just there to make up the numbers. Yu told Gillem openly that he had no desire to sit on the committee.⁵⁹

Marshall nonetheless focused on what he could do, not what he had lost. Marshall told Chiang that the Nationalists were militarily powerless to achieve the total destruction of the NEDUA in Manchuria. If Du’s troops continued to push northward, they would be in imminent danger of being defeated. A military fiasco would seriously compromise Chiang’s position. The GMD should, according to Marshall, concentrate its best troops in between Shenyang and Huludao rather than pushing them northward into a calamitous civil war.⁶⁰ Marshall then made representations to Zhou, telling him that the Chinese Communists’ occupation of Changchun ‘would boomerang to their great disadvantage’,⁶¹ as the fall of Changchun aggravated Chiang’s thirst for revenge. The CCP might be entirely unfazed by Chiang’s crack troops, but Marshall acquainted Zhou with the fact that further escalation of the military conflict would only be to the detriment of China.⁶²

Marshall spared no effort to urge both Zhou and Chiang to endorse the petition submitted by the Democratic League in late April to settle the Manchurian problem.⁶³ The petition sought to sanction all GMD troop movements on the railways in Manchuria. The CCP forces would pull out from the rail lines (i.e. the big cities) in support of the sanction, and the Nationalists would be allowed to send representatives into key cities in northern Manchuria, meaning that Lin Biao needed to withdraw his forces from Changchun. It was also averred in the petition that the Nationalist bureaucrats sacked from the chairmanship of the Political Council of the NEHQ were to be replaced by three non-partisan members.⁶⁴

The two belligerent parties failed to appreciate the bipartisan interests of the proposal from a minor party. Zhou suggested that he agreed to send a joint commission to Manchuria after a general ceasefire, but made no mention of the CCP withdrawal from the big cities. Chiang was less diplomatic and rejected the proposal outright.⁶⁵ Struggling against the odds to keep the hope of peace alive, Marshall envisaged a major embarrassment for Washington if he continued to be involved in China’s internal strife without finding a circuit breaker to stop the military conflict.⁶⁶ He revealed to Zhou later on that he would not be placed in the position of a mediator when there was almost certainty that a stalemate would develop.⁶⁷

Mao’s offensive accommodation plan was thus effectively thwarted: first, hopes for a workable peace agreement in the foreseeable future were dashed, and second, it was unprofitable to continue a war of attrition, as Sipingjie was under siege by the enemy’s elite. The military picture was grim on the southern front, too. The gradually improved logistics made it possible for the Nationalists to bring more troops and artillery pieces to the front in their second-phase assault against Benxi. On 22 April, the NEDUA’s captains from Benxi reported that it looked unlikely that they could defend their positions should the Nationalists launch a large-scale offensive.⁶⁸

But Mao was unmoved and insisted that they press ahead. He ordered the commanders in Benxi to fight tenaciously at least to delay the enemy until there was finally a peace agreement. The consequence of Mao’s order was effectively to turn the CCP’s operations in Manchuria into a defensive accommodation mode of warfare, which aimed to use the war of attrition to compel a favourable peace. There was no favourable intelligence to support Mao’s plan, but he conjured a justification from thin air for his alteration of the CCP’s military strategy, which was in line with his Madrid motto. On 27 April, Mao asked Lin Biao to augment the numbers of the defence forces in Sipingjie, and their last line of defence was therefore shifted from Changchun to the besieged city. ‘For peace and democracy … turn Sipingjie into China’s Madrid’, he wrote, demanding that his officers conjure up a military miracle.⁶⁹

Meanwhile, the GMD’s victorious offensive on 3 May finally forced the NEDUA to retreat from Benxi. The fall of Benxi radically compromised Mao’s strategic position, allowing the enemy a freer hand to strike Sipingjie and making it all but impossible for the NEDUA to maintain numerical superiority. Mao was well informed about the change in the force-to-force ratio, but he preferred that Lin Biao conduct one more strike before seeking to terminate the operation, in case his Sipingjie defenders were able to prevail against the attackers and turn the tide. Lin gave Mao a simple and direct reply: it was not possible to dislodge—much less rout—the battle-hardened enemy within such a short time.⁷⁰

Mao ignored Lin’s report, as he decided to stake his armed forces on maintaining the status quo in the major cities of northern Manchuria. Hopes for peace vanished on 29 April. Chiang bluntly rejected a proposal to settle the Manchurian problem that was endorsed by both Marshall and Zhou. Mao, however, remained adamant on the unyielding defence of Sipingjie. He urged Lin to defend the city for the longest possible time and to avert enemy breakthroughs. As peace was unimaginable at that stage, Mao’s decision in effect shifted the strategy of defence accommodation into a plan for defending the status quo that relied on gradually diluting the enemy’s strength without realistic hope for a peace settlement.⁷¹

The defenders at Sipingjie came under greater threat when the GMD’s N6A arrived, fresh from its victory in Benxi. The enemy’s reinforcements robbed Lin’s defenders, 70,000 to 100,000 in number, of their clear numerical superiority, because they were now facing more than ten GMD divisions (more than 120,000 men). The Communists at Sipingjie were left with only two options: either to heed Mao—that is, continue their courageous positional battle—or to disengage in a planned retreat at a moment of their choosing. Although Mao’s recalcitrant general Huang Kecheng had urged him to call off the defence of Sipingjie and shift the troops to consolidate their rural rear, he asserted that the longer the defence was maintained, the greater the chance of achieving a peace settlement and thus averting an open general war. Mao therefore directed Lin to retain Sipingjie, to ensure that he had enough bargaining power in hand when in due course he played the trump card (Changchun) to revitalise the negotiations.⁷²

Mao’s contingency plan was soon shattered by the N6A, which on 14 May launched an assault from the east flank of Sipingjie and gained ascendancy. From NEDUA defenders’ views, the N6A’s use of fast-moving mechanised infantry was like a German Blitzkrieg operation of World War II. Testimonies of the NEDUA veterans and a study based on these accounts emphasised the rapidity and intensity of the Nationalists’ motorised infantry—particularly an account detailing the swiftness of 600 GMD trucks pushing through the NEDUA line in a late stage of the campaign.⁷³ However, recollections of the N6A commander Liao Yaoxiang have revealed that the Nationalists’ advance in the final stage of the campaign was not at all similar to a lightening war. Liao remembered that one of the N6A’s so-called acts of gallantry was the sending of a small number of troops at night scouting ahead to the location they intended to occupy. These troops were under order to attain one linear objective only: making sure the NEDUA combatants were not there. The army then halted for its commander’s fresh instruction until the next day, before advancing large groups of troop-laden trucks to the same spot their reconnaissance squads had already declared a safety zone the previous night.⁷⁴

Nevertheless, conservative tactics had the advantage of secrecy. While the NEDUA defenders were caught unaware, elements of the N6A bypassed the enemy line from a few undefended spots further east to Sipingjie. When the N6A was in a strong position to curl into the enemy’s rear and cut the NEDUA off from retreat, the rest of the GMD armies conducted further large-scale operations on 16 May, making their offensive multipronged. On 18 May, detachments of the N1A, supported by armoured vehicles, occupied the position of most advantage immediately to the east of the city. A total collapse of the NEDUA defence seemed imminent, and Lin Biao and his forces fled the city at night.⁷⁵

The fifteen-day truce

When the battle of Sipingjie was nearing its end and an NEDUA defeat seemed inevitable, Chiang sent Defence Minister-in-waiting Bai Chongxi to Shenyang on 17 May, ordering Du to halt troop advance with the capture of Sipingjie. The Bai–Du meeting, however, reached a consensus that Chiang should be advised otherwise.⁷⁶ When his forces had advanced to within 70 kilometres of Changchun on 22 May,⁷⁷ Du received a telegram from Chiang instructing him to stall the chase immediately.⁷⁸ Du defied the order in the belief that Bai would persuade Chiang to change his mind in no time.⁷⁹

After he was briefed by Bai, Chiang flew to Shenyang on 23 May.⁸⁰ Marshall was told that Chiang had to be there for fear that his generals might do ‘the wrong thing’ and launched an assault on Changchun. In other words, Chiang was in agreement with Marshall that the GMD’s occupation of Changchun before a peace settlement with the CCP was inadvisable.⁸¹

It is harder to end a war than start one. When Chiang’s plane touched down at Shenyang, Du’s forces had already defeated the NEDUA defenders in Changchun and regained control of the city.⁸² The Nationalists’ military victory was achieved amid the leaders of the Democratic League’s last-minute peace call to Mao and Chiang on 22 May, briefly reiterating their petition of April to settle the Manchurian issues peacefully. Mao accepted the petition in a telegram on 23 May, the day his forces fled Changchun.⁸³ Democratic League leader Liang Shuming rued the day that their pleas were not acted upon soon enough.⁸⁴

While Mao won the propaganda war, Chiang’s generals wanted to exploit the enemy’s military vulnerabilities. Chiang therefore ordered Du to continue the northern troop advance at a quicker pace towards Harbin at the northern branch of the Sungari River, although he indicated that he was searching for a peaceful solution to end the conflict.⁸⁵ Chiang invited his victorious generals to banquets in Shenyang before heading to Changchun for an official review and inspection of the troops. Chiang ended his inspection tour to Manchuria on 30 May and departed for Beiping without returning to the pre-war capital Nanjing, where his government had relocated earlier. He stayed in Beiping for another three days, doing almost the same things as he did in Shenyang, particularly in holding military meetings with his senior commanders in the North China theatre.⁸⁶ Chiang sent a short note to Marshall from Beiping on 1 June to explain the delay in his return to duty in Nanjing, refusing the latter’s call for an immediate ceasefire.⁸⁷

Nevertheless, Chiang’s thoughts on Manchurian issues during this period were a strange mixture of hopes of a comprehensive military victory, fears of full diplomatic fallout with the Americans and the Russians,⁸⁸ and angst about the economic consequences of the civil war on the day-to-day life of the people.⁸⁹ Chiang’s rhetoric was also a mixed bag. While he acknowledged Marshall’s advice that reclaiming Changchun by force was imprudent, he was simultaneously spreading the news, via unofficial contacts, to American semi-official personnel that his armies would not stop until they captured Harbin. Chiang’s intentions were therefore open to interpretation. A recent study concludes that Chiang would not announce a ceasefire until he had exploited his military gains against the enemy as much as he could.⁹⁰

The problem was that Chiang went to Shenyang followed by a prolonged sojourn in the war zone while his troops sped up the advance in Manchuria. Given that Chiang’s inspection tour had virtually suspended the peace talks, it might be argued that the CCP leaders would naturally consider his intentions on the basis of worst-case assumptions.

Zhou was among the first to condemn Chiang’s tour to the front line, suggesting that his real intention was to settle the conflict by force.⁹¹ Nonetheless, the recollections of Chiang’s Manchurian field commander Du seemed to suggest otherwise. As discussed, Chiang gave the order for his troops to continue the pursuit of the fleeing enemy after the occupation of Changchun, but his order did not match his actions. While the GMD’s elite corps attacked further north, a large contingent of the NEDUA’s surviving force after its recent defeat in Benxi launched a major offensive against the Nationalists’ weak garrison force in the Anshan–Haicheng zone on 25 May.⁹² As the attack took place in an area less than 100 kilometres to the south-west of Shenyang, it posted a big threat to the GMD’s rear. Chiang and Du hurriedly recalled two attacking divisional units of the N1A from the Changchun front to Shenyang on 25 May to form a task force. Considering that the Nationalists could only deploy three army divisions to spearhead their northward pursuit, the withdrawal of two combat-capable divisions drastically affected the conduct of the operation.⁹³

Just when the task force was expected to be sent at once as a relief column to the besieged garrison, Chiang made a baffling decision that the relief assignment would be put on hold until the task force finished a three-day break. The decision was a disaster to the besieged Nationalists, as it prompted their commander Pan Shuoduan to defect with a substantial number of troops to the CCP on 30 May. When the Nationalists’ relief column reached the trouble spot five days later, the NEDUA combatants had already retreated. Chiang’s decision left Du dumbfounded because it looked as if Chiang had just rubbed salt into his own wounds. In his memoirs, Du contended that Chiang’s judgement was distorted by the insubordinate N1A commanding general Sun.⁹⁴ Du’s simplistic conclusion seems plausible. However, Chiang’s decision raises the question of whether the GMD’s elite corps in Manchuria was more vulnerable than Du might have thought because the only notable achievement of Chiang’s dubious decision was that it essentially reduced the tactical combat casualties of his two elite army divisions.

The Nationalists emerged victorious in the battle of Sipingjie. They succeeded in routing the enemy forces but failed to annihilate its main body. The Nationalists claimed that they killed 40,000 enemy fighters, but as the captured NEDUA prisoners of war reportedly numbered around 400, the Nationalists’ claim is questionable at best.⁹⁵ If Du’s US-armed armies went non-stop in pursuit of the routing enemies for a considerable distance, their far-flung battle lines could be laid completely open to attack by Lin’s surviving field forces. The GMD commanders attempted to rout Lin’s NEDUA from the battlefield by using their mechanised infantry in the pursuit, and managed to concentrate 150 motor vehicles in a critical flank on one occasion.⁹⁶ A CCP source has confirmed that the formation of the NEDUA was broken up but not exterminated by the enemy’s motorised infantry.⁹⁷

Using more than a hundred military vehicles in one operation was a lot in China in the 1940s, but it was a drop in the ocean compared to the vast space of northern Manchuria. The heavily loaded GMD foot soldiers lacked the foot speed necessary to outpace the CCP combatants.⁹⁸ In the end, the GMD’s northern pursuit forced the enemies to break up into small groups and flee into the adjacent territory of the northern branch of the Sungari River, leaving behind twenty-eight cities for the Nationalists to occupy and defend.⁹⁹

Now the GMD forces in Manchuria had to defend their expanded territories. Their supply lines therefore relied even more on US logistics support. Chiang, who conferred with Marshall on a daily basis before he left for Shenyang,¹⁰⁰ knew better than anybody else within his party about the transport capabilities of the US forces in China. At the beginning of June, the operational logistics capabilities of US Navy forces in China were dwindling. The US Seventh Fleet, which had been playing a crucial role in the Nationalists’ troop movement into the north-east, was multitasked with the transport of food for the United Nations Relief and Rehabilitation Administration (UNRRA) in China and the US atomic tests in the central Pacific. The fleet, however, needed to carry out these urgent tasks under force demobilisation.¹⁰¹ Other elements of the US naval units in the Pacific, most of which suffered from depleted crew strengths, were unlikely to reinforce the fleet effectively.¹⁰² The US Navy turned over a number of vessels to the GMD navy in the hope that it could use these ships to improve the GMD’s logistical capabilities, but the program was marred by the lack of Nationalist trained crews.¹⁰³

In sum, Chiang’s elite corps in Manchuria and China proper could be only partially effective without US logistical support.¹⁰⁴ It was therefore not surprising that the CCP remained cocky even after it lost Changchun. Zhou told Marshall on 3 June that they would continue fighting in Manchuria and China proper unless the enemy halted the offensive.¹⁰⁵ Chiang and his advisers knew that Zhou’s assertion was not just a game of political brinkmanship.¹⁰⁶ If the fighting in Manchuria showed no sign of abating, it would engulf the entire nation. As Marshall had observed, the Nationalists were incapable of winning a full-scale civil war against the Chinese Communists even with US logistics and the presence of US Marines in North China.¹⁰⁷

Chiang returned to Nanjing on 3 June, ending his twelve-day tour of inspection. In a meeting with Marshall the very next day, Chiang agreed that his armies in Manchuria would cease all aggressive action for a period of ten days. He stated in the subsequent truce statement that the temporary truce gave the CCP a chance to carry out the agreements they had previously signed in good faith, to negotiate a permanent ceasefire in Manchuria and the restoration of transportation in North China, and to carry out the army reorganisation agreement.¹⁰⁸ Despite his provocative remarks, Chiang’s statement effectively abandoned all of the preconditions he previously set for a ceasefire (e.g. Marshall’s guarantee) and agreed to a pause of hostilities in Manchuria to allow time for peace talks. Zhou quickly accepted Chiang’s offer. He even successfully argued for a change of the truce length from ten to fifteen days.¹⁰⁹ On 6 June, by the time the NEDUA was preparing to give up its northern Manchurian headquarters in Harbin,¹¹⁰ the two warring parties issued press releases independently to endorse a fifteen-day temporary truce in the north-east.¹¹¹ While Lin’s troops got the fighting lull they were seeking, Du’s overextended armies were under orders to stop and fortify defensive positions near the southern branch of the Sungari River, no more than 130 kilometres away from the enemy they pursued.¹¹²

In Yan’an, Mao and Liu were busy in sending directives to their field commanders not to initiate new offensives in China proper in the hope of appeasing the Nationalists, but they also authorised their generals to launch localised attacks in the hope of recovering lost ground and retaliation. Notably, undertaking localised military actions to prevent enemy reinforcements were also considered as acts of reprisal.¹¹³ Hence, whether the war had moved into a withdrawal appeasement mode or shifted into a state of ‘localised fighting’ in the interim was just a figure of speech.

Much to Marshall’s vexation, the fifteen-day short truce in Manchuria was marred by new truce breaches. If the ceasefire violations during the January truce were committed by both sides to control key positions before a political settlement was reached, clashes within the temporary ceasefire indicated that honouring the ceasefire was optional for the two belligerent parties. On 7 June, the day that the truce came into effect, Lin Biao’s surviving field forces attacked the GMD outpost at the village of Lafa, some 200 kilometres east of Changchun.¹¹⁴ The GMD regiment that undertook garrison duties from 6 June at that newly occupied but isolated spot was trounced and its commander killed in the battle.¹¹⁵ While Zhou flippantly dismissed the Nationalists’ claims of CCP ceasefire breaches in his memorandum to Marshall,¹¹⁶ Lin’s blustering captains made a mockery of their enemy by sending the dead body of the GMD brigadier, which they captured at Lafa, back to the Nationalists in a coffin.¹¹⁷

Byroade proposed sending more field teams to Manchuria with the aim of pushing for a pause in fighting. His plan did not receive tripartite approval until towards the end of the truce period, on 15 June.¹¹⁸ Both the GMD and the CCP members in the Executive Headquarters were verging on defeatism. They advised Byroade not to go to Changchun because they saw no hope for peace. For the Chinese Communists, the truce statement of the Nationalists looked like an ultimatum; they believed that Chiang’s true intention in stopping the troops’ advance was to buy time in preparation for a full-scale civil war.¹¹⁹

Towards the end of the ceasefire period, the two belligerent parties and their American mediators seemed to accept that sporadic fighting and mutual accusations of truce violations were a new norm in Manchuria.¹²⁰ The American peacemakers knew that the temporary truce would not invoke miracles for them. All they could do was to avert a full-scale confrontation, but as Marshall revealed to Zhou, they were getting tired of doing it.¹²¹

While the brittle truce in Manchuria had been temporarily maintained, the province of Shandong became the post-ceasefire fighting hotspot. The CCP forces launched massive coordinated attacks, once again at the start of the Manchurian truce on 7 June, against the GMD-held cities, towns and garrison points along the two main land transportation corridors of that coastal province in China proper, the Qingdao–Jinan and the Tianjin–Pukou railways. When those week-long devastating post-truce attacks finally ended, the GMD forces in Shandong could defend only a few isolated spots, including the provincial capital of Jinan and the port of Qingdao.¹²²

From their position as mediators, the Americans considered the attacks as the CCP’s move to counterbalance its defeat in Manchuria.¹²³ The relative de-escalation on the Manchurian front line against the rapid escalation of conflict in China proper within the truce period could be considered as post-ceasefire conflicts triggered by the ‘non-winner’ in the ceasefire negotiations.

In early July 1946, an NEB meeting held in Harbin passed a resolution reconfirming the correctness of Mao’s rural strategy and mobile warfare principles. According to Mao, the Sipingjie campaign was an exceptional case, not to be invoked when considering future strategy. Rather, he advanced the idea that there were important trade-offs between the defence of big cities and mobile warfare, with the latter providing greater rewards in reducing the enemy’s effective strength. In the second half of 1946, approximately 12,000 cadres were dispatched to mobilise the rural population of Manchuria in order to carry out the party’s bid for a more aggressive land reform program, bolstered by military operations that eradicated the bandits. The development of Communist base areas in Manchuria later proved to be indispensable for the CCP’s ultimate victory on the mainland.¹²⁴

One day

Mao’s assessment of the 1946 Sipingjie campaign has become a standard view of the assessment of the NEDUA’s debacle in that battle; for decades, studies have given credit to Mao for his ability to make tough decisions in accordance with actual political needs, and have accepted that the defence of cities in Manchuria was dictated by international politics and the negotiations.¹²⁵

Mao’s original plan for the defence of Sipingjie was designed to achieve a clear objective: to destroy the GMD’s crack troops in mobile warfare and to ensure that a peace deal would be made on Communist terms. But the offensive accommodation mode of combat was short-lived, as the combat conditions made this policy all but impossible to implement. Mao then changed to a defensive accommodation plan, committing his forces to a positional war in the belief that the Nationalists would halt the fighting and reopen negotiations as soon as Marshall returned. This approach miscalculated the revenge-seeking enemy’s determination to reclaim Changchun, however, and therefore it too was shelved. Mao refused to withdraw his troops, still confident that his negotiators could bring about a ceasefire. Nevertheless, his attempt to sustain the defence of Sipingjie became less rewarding in the later stages of the battle, as his objective became increasingly difficult to achieve and hence his commitment to prosecuting the battle became ever more difficult to justify.¹²⁶

With hopes for a favourable peace deal dashed and in the face of the enemy’s ever-growing military superiority, Mao called on the spirit of his Madrid concept and compelled the NEDUA to realise it through a devastating defence. In this desperate late stage, however, the paramount leader could only imagine that he still had a good grip on what would turn out to be the optimal strategy. In imagining China’s Madrid in Manchuria, Mao had a vision of the defence of Sipingjie that was outside his normal tactical discourse: using mobile warfare rather than a static war of attrition to defeat the enemy.¹²⁷

In their private talk in 1959, Mao was confronted once again by the insubordinate Huang about his decision-making in the defence of Sipingjie. Huang’s query pointed exactly to the problem: why did Mao insist on sustaining the defence when all chances of achieving the original goal had evaporated?¹²⁸ In conceptual parlance, the ‘certainty effect’ is likely to reduce policy-makers’ tolerance for accepting risk but, as this chapter shows, Mao’s act was outside the postulations of modern theories of risk-taking. Huang’s disagreement with Mao underscored the latter’s style: his timing of a switch from a strategy of risk acceptance to a conservative but cost-effective approach occurred far later than one would have expected in a prudent policy-maker.¹²⁹

Chiang, on the other hand, did not have a general who was courageous enough to remind him that the political and military realities simply did not support his great gamble in Manchuria. Chiang noted his wishes to use the air force in the fight against the NEDUA, but it was difficult for him to turn wishes into reality.¹³⁰ As mentioned, Chiang restricted the use of aerial warfare over Sipingjie in an attempt to avoid ruffling Soviet feathers. It deprived the Nationalists of using air raids to eliminate the enemy, although the decision must have inadvertently reduced the scale of civilian casualties. Chiang’s generals followed his directive and carried out a war plan that was designed to avoid enemy ambush more than to execute frontal attacks. However, no war plan was foolproof in a war against the CCP’s best tacticians. While forces spread out wide were vulnerable to CCP attacks, a close formation was not much safer. The Nationalists’ counterinsurgency operation in Manchuria in 1946 was a case of damned if you do, damned if you don’t.

Chiang noted, in the immediate aftermath of his armies’ recapture of Changchun, that the civil servants and school teachers went on strike in the big cities as a result of continuing food insecurity and severe inflation. Although he believed that the recapture of Sipingjie and Changchun had forestalled the escalation of unrest, he acknowledged that his government needed to clamp down on the illegal stockpiling of food.¹³¹ The middle-class professional city dwellers should have been the Nationalists’ key support base. No matter how much political capital Chiang could earn from the battlefield, the economic consequences of the war could destroy it quickly. Chiang once used ‘crossing a broken bridge on a snowy night’ to describe his ambivalent attitude towards a number of destabilising forces, such as talking peace while waging war against the CCP.¹³² Chiang used many similes and metaphors in his writings to convey his thoughts so he must have known that most broken bridges are impassable on snowy nights.

Chiang informed Marshall that he wanted to occupy critical areas of the north-east like Changchun only as symbols of the power and authority of his government, but Marshall warned him that it was unwise, even from a symbolic standpoint, to deploy his best troops to defend these areas without a political settlement with the CCP.¹³³ Chiang stopped the fighting in the end, but his elite corps did not annihilate Lin’s forces and none of his preconditions for a ceasefire were fulfilled. If every war is as ironic as Paul Fussell sees it,¹³⁴ Chiang’s temporary truce ironically added another controversial case of casus belli against the GMD for the CCP. It gave one more justification for the Chinese Communists to flout the truce and fanned the fire of the civil war from Manchuria back to China proper, as Chiang’s ceasefire statement read like an ultimatum to them. The GMD was incapable of winning a full-blown civil war against the CCP, as Chiang had been advised by the Americans. Chiang surely knew what he wanted, but he might not have known how much he was capable of getting.

Hence the two warring parties virtually collaborated to create a vicious circle of attack and retaliation. The Nationalists launched offensives against the NEDUA in retaliation for the capture of Sipingjie and Changchun. The Chinese Communists then retaliated for the retaliation. A cycle of vengeance and retaliation can only spiral towards calamity. A recent study has raised the importance of setting limits or compensating the other party’s loss before seeking revenge in future negotiations.¹³⁵ In the case of the Chinese Civil War, however, retaliatory military actions had become something similar to a crutch that both parties could not walk without.

The escalation of military actions during the truce period furnished considerable scope for the two belligerent parties to toughen their negotiation stance. Both parties began to put forward proposals that not only caused offence to the other party but also removed Marshall’s power to influence matters. Disagreements ran particularly high between the two parties on two critical fronts. On the arrangements of a permanent ceasefire in Manchuria, Zhou claimed that the ceasefire of 10 January had lost its effect as there was no need for the GMD forces to restore sovereignty in Manchuria after the Soviets’ withdrawal.¹³⁶ On carrying out the army reorganisation agreement of 25 February, Chiang maintained that if the CCP desired to amend the existing agreement by increasing its troop quota to Manchuria, the CCP forces must withdraw from the provinces of Rehe and Chahar before September 1946, and evacuate from all localities they occupied after 7 June 1946 in the province of Shandong before 1 July 1946. Chiang also requested that the CCP fulfil extra conditions and terms, according to which the Chinese Communists would be most offended, by demanding that the two CCP-held seaports and logistics centres of Yantai (formerly Chefoo) and Weihaiwei in north-eastern Shandong be turned over to the GMD.¹³⁷

Although the two parties could not find sufficient common grounds for holding a Committee of Three meeting, they kept Marshall busy writing one party’s verbal demands or turning general statements into draft proposals, having the draft documents translated into Chinese and returning them for approval before passing them to the other party.¹³⁸ This chapter shows that Marshall pushed a course that had little traction. His efforts to send field teams into Manchuria has been regarded with scepticism. Although these efforts seemed fruitless, he continued to pass subtle messages to both parties that he was ready to steer them to the best potential solution, akin to salespersons of today starting to fill out the retail invoice before the buyers agree to the purchase.¹³⁹

When the stakes were stacked against him, however, staying in the game was going to be a struggle for Marshall. In a meeting with Yu, Marshall was informed that the two parties did not share any common ground.¹⁴⁰ While the truce deadline appeared likely to slip without any hope of reaching agreement, Marshall asked Yu what might be a proper extension of the fifteen-day ceasefire. Yu’s answer was enough to make the most seasoned mediator cringe: ‘1 day’.¹⁴¹