Violence, Political Strategy and the Turn to Guerrilla Warfare by the Congress Movement in South Africa

The Congress movement in South Africa was transformed in the early 1960s from a movement committed to the exclusive use of non-violent means in the struggle against apartheid to one focused on rural guerrilla warfare as a free-standing and sufficient first step towards ‘all-out war’ and the armed seizure of power. But few, if any, of the participants in the Congress movement’s deliberations in 1960–61 on whether to ‘turn to violence’ had believed that this was the strategy that they were endorsing when they authorised the abandonment of exclusive reliance on non-violence. The choice facing the Congress movement after 1960 was not between mutually exclusive alternatives of ‘non-violence’ on the one hand and ‘violence’ or ‘armed struggle’ on the other. Rather, Congress leaders contemplated a range of different forms of violent action and considered their relationship to various forms of non-violent activity and to the kind of transition from apartheid that these actions could and should be intended to produce. This article analyses the range of strategic options that were canvassed within the Congress leadership in the early 1960s and the decision-making process by which those options were gradually narrowed. That process was shaped by ambiguity, unilateral action, unintended consequences and state repression, with the result that the Congress movement’s ‘turn to violence’ ultimately took a form that few Congress leaders had initially desired or anticipated.

words, 'a deliberate and a heavily contested choice'. 9 On the basis of research in a wide range of newly accessible sourcesincluding trial records, a rich body of life writing and oral histories, and scattered internal Congress movement documentationwe now know far more about the roles within that contested decision-making process of ANC president Albert Lutuli, 10 the South African Communist Party (SACP) and groupings and individuals within the SACP, including MK's first commander-in-chief, Nelson Mandela. 11 There is a danger, however, that the fierce debates over these topics have sometimes generated more heat than light. There has still been little systematic focus on the decisions and debates among Congress leaders on political strategythat is, on ways to employ the means available to them in order to achieve their desired ends.
This article's attention to 'thorny questions of strategy and tactics' brings two issues into focus. The first is what Joe Slovo, a founding member of the MK high command, once called the 'the all-important theoretical problem of the way we saw the transition to a democratic revolution'. 12 Within the older vein of critical scholarship on the Congress movement's armed struggle, Legassick in particular paid close attention to the transition scenarios that opponents of apartheid envisaged and sought to bring about. 13 But, despite the far greater volume and variety of source material now available, recent research has largely ignored this issue. When it has been addressed, this has often been in studies of Mandela, who is sometimes treated as a unique outlier within the Congress movement in his belief that the purpose of violent action was to increase pressure on the National Party government to negotiate. 14 More commonly, some recent scholars have implied that historical actors' commitment to the use of violent means, especially guerrilla warfare, necessarily entailed rejection of the possibility of negotiations and commitment to the military defeat and overthrow of the government. 15 Three distinct scenarios of transition to a post-apartheid future influenced strategic thinking within the Congress movement in this period. The first was a white realignment transition, in which a parliamentary realignment and/or a shift in the voting preferences of the white electorate would lead to the replacement of Prime Minister H.F. Verwoerd's government with an alternative white regime that would be more open to compromise or, at least, more responsive to pressure, and would thus itself ultimately give way to non-racial democracy. The second was a directly negotiated transition between the ANC and the governing National Party. The third was a takeover transition, in which there would be a 'seizure of power' following the collapse or military defeat of the government. Beliefs about which of these scenarios were desirable and realistic shifted over time. They shaped and were shaped by Congress leaders' beliefs about the various forms and combinations of nonviolent and/or violent activity that they should adopt. This is the second issue brought into focus by attention to political strategy: violent political action can take multiple forms and can relate to non-violent forms of political activity in a variety of ways. This was obscured in some of the earliest and most influential contributions to the recent vein of scholarship on the 'turn to violence' -Scott Couper's study of Lutuli's attitude and Stephen Ellis's analysis of the role of the SACPin which the choice facing the Congress movement in 1960-61 tended to be reduced to a simple, mutually exclusive binary between 'non-violence' on the one hand and 'violence' or 'armed struggle' on the other. 16 But far from viewing them as mutually exclusive, for instance, many Congress leaders initially endorsed violence as a means of enhancing the effectiveness of continued mass, predominantly non-violent campaigns. As Rusty Bernstein, a leading participant in the Congress movement's debates on violence, later explained, while there were some Congress leaders who were opposed in principle to any kind of violence, the more significant disagreements within the Congress leadership in the early 1960s were over the question 'what violent path are we going to follow?'. 17 Those disagreements have not always been apparent to historians because sometimes they were not immediately apparent to historical actors themselves. Terms such as 'violence', 'armed struggle' and 'guerrilla warfare' do not have single, fixed meanings and were understood in different ways by participants in the Congress movement's deliberations. This ambiguity is notor not onlya historiographical problem for historians to solve by attempting to adjudicate between different participants' recollections and determine what was really decided. By temporarily concealing Congress leaders' divergent understandings of what was being agreed, ambiguity was at times a crucial feature of the historical process itself.
There are glimpses of some of those divergent understandings in some of the most recent scholarship. 18 But these insights have not yet been drawn together into an analysis of the range of options that were considered within the Congress leadership and of the decisionmaking process by which those options were gradually narrowed. This article offers such an analysis. Elsewhere, I have argued that both the decision in 1960-61 to abandon reliance on exclusively non-violent methods and the specific form of MK's initial violent activitiesspectacular symbolic sabotagewere determined by the concern of MK's founders to discourage opponents of apartheid from supporting rival groups committed to violence or initiating uncontrolled violent action themselves. 19 But the sabotage campaign was always understood as a temporary placeholder: it was not itself a strategy for achieving South Africa's liberation from apartheid. What remains to be explained, then, is how, having abandoned exclusive non-violence, the Congress movement ultimately came to adopt a specific political strategy for ending apartheidfocused on rural guerrilla warfare as a free- standing and sufficient first step towards all-out war and armed takeoverrather than any of the alternative strategies that were canvassed within the Congress leadership in the 1950s and early 1960s.

Considering Violent Alternatives in the 1950s
Within the Congress movement, the decade before the formation of MK was one of tactical experimentation and strategic uncertainty. Congress leaders spent the decade trying out a range of non-violent activities, including demonstrations, civil disobedience, boycotts and strikes, in order to prepare for and help to realise a transition that was only vaguely conceived and that was not discussed in detail within the Congress leadership. 'We cannot tell what exact form the changes will take, how exactly or when they will come', wrote Michael Harmel, one of the SACP's leading theorists, in 1959. But the revolutionary defeat of apartheid was a 'certainty' and it need not involve violence. 20 Nevertheless, while they continued to experiment with various non-violent forms of political action, the leaders of the Congress movement frequently also debated the possibility of initiating some kind of violent activity. They felt under pressure to do so from some of their subordinates and wider constituency, who often reacted with violence to state intervention in their own lives. 21 Throughout the 1950s, Congress leaders discussed whether they should seek to transform their supporters' apparent enthusiasm for localised, reactive violence into organised, proactive initiatives that would not only impede the implementation of apartheid policies but contribute to ending white supremacy. These discussions have led recent historians to trace 'the drift towards armed struggle in South Africa' back into the 1950s and late 1940s and to conclude that the 1960 Sharpeville massacre was, in Dubow's words, 'not so much the proximate cause of the turn to sabotage as the trigger for a plan that had been discussed in small circles for some time'. 22 But no single alternative 'plan' emerged from Congress leaders' frequent considerations of abandoning exclusive nonviolence in the 1950s. Rather, Congress leaders considered a variety of different forms of violent activity against a variety of targets on a variety of scales for a variety of purposes.
Some Congress leaders advocated the adoption of targeted violence to coerce members of the communities that they regarded as their constituency. This, they believed, would enhance the effectiveness of activities that were primarily non-violent in nature, including campaigns for supporters to withdraw their labour power, through strikes, or their purchasing power, through boycotts. ANC leaders, Mandela later wrote, 'often discussed the question [of] to what exten[t] we should rely on coercive measures in organising political demonstrations'. 23 As early as 1946, when there was discussion of calling a general strike in support of striking African mine workers, Walter Sisulu and other leaders of the ANC Youth League concluded that they could guarantee success by sabotaging the railway line that transported black workers from the south-west townships into Johannesburg. Sisulu attempted to obtain explosives for this purpose but was unable to do so. 24  pickets to obstruct or assault those who failed to observe calls to stay at home from work or to boycott specific shops, products or services. Despite considerable debate, however, the national ANC leadership formally opposed such actions, declaring itself 'against the use of coer[c]ive measures as a means of mobilising the support of the people'. 25 Congress leaders also debated forms of larger-scale violence that would have represented a considerably greater departure from their existing approach to the liberation struggle. Some contemplated urban insurrection. In 1955, when the government began implementing its plan to remove the entire African population from Sophiatown, the freehold suburb west of Johannesburg, some opponents of apartheid hoped that forceful resistance to the removals could initiate an immediate mass uprising that would spread elsewhere. At a heated meeting before the removals began, some Congress leaders advocated erecting barricades to 'fight it out with the government'. Gang leader Don Mattera, who was among those preparing to defend Sophiatown, recalled awaiting a call from the ANC to begin 'an uprising which would spark national insurrection and eventually result in the overthrow of the Nationalist Government'. But the majority of Congress leaders balked, insisting that 'an insur[r]ection was not just a demonstration of anger, but a serious matter requiring careful preparation, and that an uprising based upon a local issue would be suicide'. 26 More frequently, Congress leaders contemplated some form of guerrilla warfare. They were encouraged by high-profile guerrilla struggles elsewhere in the world, from Mau Mau in Kenya to the communists under Mao Zedong in China. As early as 1949, when Mandela, Sisulu and other leaders of the ANC Youth League were drawing up the 'Programme of Action' that they wanted the ANC to undertake, some Youth Leaguers were already speculating that it would be necessary in future to resort to guerrilla war. 27 In 1953, at Mandela's suggestion, Sisulu visited China to explore whether Mao's government would provide assistanceincluding weaponryif the ANC were to adopt violent methods. Chinese officials demurred, warning Sisulu that it was not enough to 'just be impressed' by guerrilla struggles elsewhere. 28 Nevertheless, the idea of guerrilla warfare continued to hold a romantic attraction for Mandela. In his autobiography, he recalled surveying the dense forest while driving through the Eastern Cape in 1955 and dwelling on 'the fact that there were many places a guerrilla army could live and train undetected'. 29 After Sharpeville: The Decision to Abandon Exclusive Non-Violence, 1960-61 Though Congress leaders frequently discussed abandoning their exclusive reliance on nonviolence in the 1950s, their discussions produced no consensus on what form of violence they should adopt if they were to do so. To most of the Congress leadership, the potential costs of the various forms of violence that they discussed outweighed any potential benefits. But many Congress leaders began to reconsider this assessment following the Sharpeville massacre on 21 March 1960, when police fired on a crowd of peaceful protesters mobilised by the rival Pan Africanist Congress (PAC). The massacre and its aftermathwhich included the banning of the ANC and the PAC and a five-month state of emergencyfurther intensified popular impatience with the Congress movement's adherence to exclusive non-violence. Many Congress leaders feared that, if they did not respond to this increased pressure from below, their supporters might initiate uncontrolled violence on their own or defect to the PAC or another group less reluctant to embrace violent action. 30 In December 1960, the SACP's national conference adopted a resolution, proposed by Bernstein, that the Congress movement must reconsider 'its tactics of exclusive reliance on non-violent methods'. 31 In June 1961, Mandela then presented a proposal to 'turn to violence' to meetings on consecutive days, first of the ANC national executive, and then of the Joint Executives of the Congress Alliance, the umbrella body of the ANC and its formal allies. 32 Mandela's proposal was vigorously opposed by several senior participants in these meetings, including ANC president Lutuli. But eventually a compromise was agreed: Mandela was permitted to form an independent body to carry out acts of violence, while the ANC and its allies in the Congress Alliance would continue to engage in non-violent activity. Slovo was subsequently appointed by the SACP to join Mandela in constituting the 'High Command' of the new body, MK, onto which others, including Sisulu and Govan Mbeki, were subsequently recruited.
Most participants in these meetings of the SACP, the ANC and the Congress Alliance did not believe that they had authorised the launch of a guerrilla insurgency. In the wake of the Sharpeville crisis, Mandela's long-standing interest in guerrilla warfare had intensified. During 1961, he undertook an intensive course of reading on 'armed warfare with particular emphasis on guerrilla warfare', both rural and urban. His self-made syllabus included works on rural guerrilla struggles in China, Cuba, the Philippines and elsewhere, and Menachem Begin's memoir of the urban guerrilla revolt waged by the Irgun in Palestine. 33 By mid 1961, Mandela himself, together with those he recruited on to the new high command, was already anticipating and planning to wage some kind of guerrilla struggle. But it is unclear to what extent, if at all, guerrilla action was even discussed in the crucial meetings of 1960-61 that led to the formation of MK. Bob Hepple, who chaired the SACP's Johannesburg district committee, recalled that at the time of the party's December 1960 conference there was 'no suggestion' of 'a full-scale guerrilla war'. 34 According to Mbeki, those already committed to launching guerrilla operations deliberately obscured their intentions in the crucial meetings in June 1961. Others present 'didn't realise' the form and scale of the violence that they were envisaging. 35 What Congress leaders did authorise in 1960-61 was, as Bernstein later wrote, the use of violence for the purpose of 'aiding and abetting the non-violent movement'. 36 Unlike 'fullscale guerrilla war', the use of violence in this way was acceptable to a much broader coalition within the Congress leadership. Throughout the 1950s, Congress leaders had always decided against using violence to enhance the effectiveness of boycotts and strikes. But the Sharpeville crisis appeared to be a moment of both increased threat and new opportunities, if the Congress movement could take advantage of them. On the one hand, it seemed that state repression was successfully deterring popular participation in exclusively non-violent campaigns: Bernstein recalled a common worry that 'any minute now people are not going to heed our call to come out on general strike'. 37 On the other hand, the government appeared newly vulnerable. While Prime Minister Verwoerd and other hardliners remained implacable, prominent Afrikaner nationalist intellectuals, industrialists and churchmen echoed the white opposition in demanding that the government stabilise the situation by reforming its racial policies. Even some government ministers appeared to endorse such proposals, and rumours circulated that a new coalition government might be formed. 38 In a discussion document circulated to SACP units in August 1960, Harmel not only initiated formal consideration within the SACP of abandoning exclusive non-violence but also devoted considerable attention to the possibility of white realignment. Noting the 'many disagreements and criticisms' being voiced 'not only among White South Africans in general, but even inside the Nationalist Party', Harmel concluded that 'the Verwoerd Government is not as strong and stable as it pretends'. He stressed that: [i]f we want to preserve the possibility of a peaceful transition for the South African revolution, our first duty is to remove the Verwoerd government. It is still not too late for the democratic and progressive forces, under militant leadership and taking advantage of the splits within the ruling class and widespread world support for our cause, to remove this hated, minority government by the force of popular pressure on a mass scale. 39 Harmel concluded that the 'immediate task' for the SACP and its allies was to bring about the replacement of Verwoerd's government by one that 'may well fall short of what we want' but that would be 'prepared, at the very least, to negotiate with all sections of the people for a peaceful settlement of such immediate grievances [as] passes, wages and representation in Parliament. This would be a government of transition, between Verwoerd and national liberation'. 40 For many Congress leaders, the challenge after Sharpeville was to find ways to take advantage of the government's apparent weakness by maximising 'the force of popular pressure on a mass scale' despite increased repression. The use of violence was authorised in the crucial meetings of 1960-61 as a way to meet this challenge by enhancing the effectiveness of the Congress movement's mass, predominantly non-violent campaigns. The SACP conference resolved that the use of 'organised armed force' was 'a necessary complement of mass political agitation'. 41 Similarly, in the heated debates in June 1961, Lutuli and other sceptics eventually acquiesced to Mandela's proposed 'turn to violence' on the understanding that violent activities should be used to enable the continued use of nonviolent methods, especially stay-at-home strikes. Mbeki  Though Mandela probably always viewed this as a temporary concession, it was thus possible to secure a consensus on abandoning exclusive non-violence only on the understanding that, as Mandela later wrote, violent activity 'would not be the centrepiece of the movement'. 43 Crucially, moreover, it was left ambiguous what kind of violent activity had been authorised. The SACP conference resolution mandated preparation for 'forcible forms of struggle' but was silent on the specific forms being endorsed. 44 There was a similar lack of clarity in the deliberations of the ANC and the Congress Alliance six months later, perhaps because debate centred on non-violence as a principleand whether it should be abandonedrather than on violence as a tactic that could take many forms.
Some participants, including Lutuli and SACP general secretary Moses Kotane, believed the meetings of 1960-61 had endorsed limited, non-lethal violence to coerce those whom the Congress movement regarded as its own constituency to participate in strikes. Kotane reported to Soviet officials in November 1961 that the SACP's 1960 conference had resolved 'to employ some elements of violence during our mass struggles, such as picketing and disruption of communications'. 45 Aware that Mandela was already contemplating much more extensive forms of guerrilla violence, Kotane had initially opposed Mandela's proposal that the ANC should 'turn to violence'. He subsequently played a crucial role in brokering the compromise in the June 1961 meetings. Mbeki remembered Kotane telling the Joint Congress Executives that '[w]hat we are asking for is that when a stay-away takes place, we should be allowed to put barriers, so that buses can't cross taking people to work'. 46 One of the Congress of Democrats representatives in the same discussion similarly recalled that the meeting had endorsed strike-related sabotage, though he believed that this would be directed at influencing the behaviour of employers rather than workers. Rowley Arenstein remembered that sabotage was discussed as a means of punishing and deterring retaliation by employers against strikers: if a strike was broken, the employer 'could be taught a lesson by having [their] factory burnt down'. 47 Others believed that the meetings of 1960-61 had authorised not just confrontational picketing or punitive sabotage but armed self-defence to protect participants in the Congress movement's non-violent activities. According to Hepple, some delegates at the SACP conference understood the resolution to be endorsing 'the setting up of armed units in townships and rural areas to protect people from police attacks' as a 'necessary … secondary activity' in support of the 'priority' of 'organising the urban working class'. 48 Along somewhat similar lines, witnesses in later trials in the Eastern Cape claimed that, when the ANC national executive's decision to abandon exclusive non-violence was communicated to their ANC cells, the emphasis was on armed self-defence against police interference in Congress activities. One former member of MK testified that the local ANC chief steward had informed members of his ANC zone in Port Elizabeth that they 'must now arm themselves when they come to meetings so that if a policeman should enter and disturb us we shall kill him then and there. Even when distributing leaflets if we should come across a policeman who wants to arrest us or disturb us, we must kill him'. 49 43  Lutuli's objections were overruled, and MK's campaign of symbolic sabotage continued. Though unexpected, the actions of MK's commanders were calibrated to maintain a majority in support of their actions within the Congress leadership. 'Since the ANC had been reluctant to embrace violence at all', Mandela later wrote, 'it made sense to start with the form of violence that inflicted the least harm against individuals: sabotage'. 52 Although the crucial meetings of 1960-61 had not authorised guerrilla insurgency, the high command had immediately begun preparing for guerrilla war. In October 1961, they arranged for six leading recruits to be sent to China for guerrilla training. In May 1962, just five months after MK's first sabotage actions, Mandela co-authored a memorandum to the government of Ghana that described them as 'the first phase of a comprehensive plan for the waging of guerrilla operations'. 53 Members of the high command viewed symbolic sabotage as a stopgap. It was an activity that stretched many participants' understandings of what had been agreed in the crucial meetings of 1960-61 but, though surprising, would not alienate a majority of the Congress leadership. And it could be undertaken until MK had the manpower and the firepowerand the political authorisationnecessary to mount guerrilla operations. In the meantime, spectacular sabotage would advertise the Congress movement's abandonment of exclusive non-violence and thus discourage impatient opponents of apartheid from initiating violent action themselves or supporting rival groups. 54 MK's sabotage campaign and the state's unexpectedly vicious response to it had a militarising effect on the active leadership of the Congress movement. Verwoerd's government massively increased spending on the police and military and expanded police powers of 'banning', house arrest and detention without charge. Police use of torture became widespread. As MK continued to launch sabotage bombings amid escalating repression, attitudes within the Congress leadership were rapidly transformed regarding the kinds of violence that MK should undertake, the relationship between violent and non-violent action and the kinds of transition from apartheid that Congress activities were intended to achieve.
In October 1962, the prospect of guerrilla warfare was explicitly raised at the national conferences first of the SACP and then, a week later, of the ANC. Held over the border in Lobatse, Bechuanaland, and in the absence of Lutuli, who was immobilised by a banning order, the ANC conference now endorsed the position that Mandela  formation of MKnow acknowledged as the ANC's 'military wing'but that had previously been a minority position within the ANC leadership. The ANC national executive's conference report characterised the sabotage campaign as an 'elementary phase' that would be followed by 'the advanced stage of guerrilla warfare'. 55 The launch of MK also altered the relationship between violent and non-violent activity that had been agreed in the meetings that sanctioned its formation. When Kotane had visited Moscow in 1961, his Soviet interlocutors had stressed 'the Marxist-Leninist doctrine on the combination of all forms of struggle': '[i]t is necessary not to counterpoise one form of struggle to the others but to combine skilfully all these forms', Soviet officials advised. 'Without consistent political and organisational work among the masses victory is impossible. The winning of the masses to your side and preparation for the armed struggle are two sides of the same question'. 56 In line with such advice, and despite the increasingly repressive conditions in which they were operating, Congress leaders continued in their public pronouncements to stress the importance of non-violent political work. Though the government was closing 'every channel of legal protest and normal political activity', the new party programme adopted at the SACP's 1962 conference rejected 'theories that all non-violent methods of struggle are useless or impossible'. While endorsing the need 'to meet violence with violence', the programme committed the SACP to 'work for the use of all forms of struggle by the people, including non-collaboration, strikes, boycotts and demonstrations'. 57 The ANC's conference report similarly stressed that 'OUR EMPHASIS STILL REMAINS MASS POLITICAL ACTION'. But the report indicated an important shift in how such action was conceived. 'Political agitation', it explained, 'is the only way of creating the atmosphere in which military action can most effectively operate'. 58 In the crucial meetings in 1960-61, participants had authorised violent action in order to facilitate the prosecution of primarily non-violent campaigns. This was now inverted: the ANC's official position was that the purpose of non-violent campaigns was to facilitate the prosecution of violent action.
In practice, moreover, MK's launch paralysed the Congress movement's efforts to organise non-violent activities. Congress leaders continued to publicise plans for new mass campaigns. In April 1963, for instance, the ANC leadership notified local branches that it had 'decided to embark on a campaign of active struggle that will embrace the masses of the people' and that would culminate in pass burnings and a general strike. 59 But the campaign was never implemented. Within the active Congress leadership, MK's activities monopolised attention and resources. Simultaneously, the MK high command redeployed much of the movement's most valuable manpower: in 1962-63, MK recruited hundreds of activists from the ANC, the South African Congress of Trade Unions and other Congress bodies to be sent for military training abroad. The effect of this, Mandela later reflected, was that MK 'drained the political organisations of their enthusiastic and experienced men'. 60 Far from facilitating non-violent activity in increasingly difficult conditions, as the Congress leadership had originally agreed MK should do, the formation of an armed wing impeded it.
The launch of MK also led Congress leaders to reconsider the transition scenarios that seemed desirable and realistic. When the Congress leadership had agreed in 1960-61 to authorise the use of violence to enhance the effectiveness of non-violent campaigns, their deliberations had been shaped by the enhanced prospects of white realignment and the consequent need to find ways to exert mass popular pressure despite escalating repression. In the wake of the Sharpeville crisis, however, a white realignment began to seem increasingly unlikely. Verwoerd brushed aside calls for reform, reasserted control over his own party and then further increased the National Party's dominance of white politics in the October 1961 general election.
At the same time, for many Congress leaders, the creation of MK as an incipient guerrilla force seemed to open up alternative possibilities for ending apartheid. During his intensive course of reading on warfare in 1961, Mandela read Carl von Clausewitz's discussion of 'limited' war fought to secure favourable peace terms, which Clausewitz distinguished from 'absolute' war fought to 'overthrow' or 'disarm' the enemy. Mandela's gloss on Clausewitz's analysis of limited war was that 'war may be ended without destroying an enemy's army': he took careful notes on Clausewitz's discussion of 'the wearing out of the enemy' as the most common means of victory when 'the weaker resists the stronger'. 61 Mandela's views were further reinforced by his talks with representatives of the Algerian Front de Lib eration Nationale (FLN) during the tour of Africa that he undertook in 1962. One faction of the FLN opposed any negotiation with France over the future of Algeria: this included the influential propagandist Frantz Fanon, who wrote scornfully of attempts to 'settle the colonial problem around a green baize table'. Mandela was influenced not by the FLN's rejectionist hardliners, however, but by Chawki Mostefaï, a member of the FLN negotiating team then in final talks with the French to end the war and secure Algeria's independence. 62 Mostefaï told Mandela that '[t]he original objective of the Alg[erian] revolution was the defeat of the French by Military action … Settlement by negotiation was not visualised'. But, Mostefaï stressed, 'in the course of the war they realised that a pure military victory over the French would be well nigh impossible'. 63 In gaol in 1964, Mandela explained to a fellow prisoner that he had concluded from his discussions in Algeria that 'there was no point in trying to overthrow the South African apartheid regime: we had to force them to the negotiation table'. 64 In his belief that the purpose of MK's planned guerrilla insurgency was not the total destruction of the state's military capacity and a takeover by armed force but the wearing down of the National Party government until it agreed to participate in a negotiated transition, Mandela was not an exceptional, lone outlier within the Congress leadership. Weeks after Mandela was arrested and imprisoned in August 1962, the new party programme adopted by the SACP conference stressed that '[t]he Party does not dismiss all prospects of non-violent transition to the democratic revolution', and suggested that the development of 'revolutionary and militant people's [armed] forces' might lead to 'a peaceful and negotiated transfer of power'. 65 But the formation of MK led others in the Congress leadership to begin to imagine the alternative possibility of a takeover transition. 'The very existence of an armed force', Bernstein later reflected, had an 'unintended and unexpected consequence': it raised for the first time the prospect of 'the acquisition of state power by force'. In its Lobatse conference report, the ANC national executive publicly defined its objective, for the first time, as 'the seizure of political power'. What seizing power meant, Bernstein wrote, was 'never precisely defined'. But leading participants later recalled a widespread assumption that the seizure of power would involve the collapse or military defeat of the government. 'We were going to march into Pretoria and take over the government and set up a new system', remembered Denis Goldberg, a member of MK's regional command in Cape Town. 66 'Were we aiming simply to put pressure on the governmentto force it to changeor to overthrow it? If so, how?', wondered Ronnie Kasrils in his memoir three decades later. A member of the MK regional command in Durban, Kasrils explained that he had 'perceived these questions only dimly at the time. In retrospect … I came to realise that the strategy had not been clearly worked out'. 67 In 1962, this lack of strategic clarity made possible the emerging consensus within the Congress leadership on the need to initiate guerrilla warfare; concealed beneath that apparent consensus were widely divergent assumptions about the form and purpose of guerrilla operations. In the immediate term, these divergent assumptions could coexist within the active Congress leadership because all agreed that the immediate next step was the development of a guerrilla cadre. But, as Congress leaders began to look beyond that step to consider how to deploy the movement's swelling ranks of trained guerrillas, divisions began to open up. 'While there was an agreement in MK on the general idea of an armed struggle', Goldberg explained, 'how to move forward was less clear'. 68 Operation Mayibuye and the Debate over MK's 'Next Phase', 1962-63 It was in the debate provoked by 'Operation Mayibuye' that, as Bernstein put it, previously submerged disagreements about the form and purpose of MK's next phase 'began to surface very forcefully'. 69 Operation Mayibuye was a plan drafted in 1963 by Mbeki and Slovo and adopted by the MK high command. The plan was never presented to the full ANC national executive, headed by the immobilised and increasingly marginalised Lutuli, who would almost certainly have opposed it. But it was considered by the SACP central committee, where it encountered opposition from leading figures including Bernstein, Ruth First and party chairman Bram Fischer. 70 'If ever … there was ever a plan which a Marxist could not approve in the then prevailing conditions', Fischer later explained, 'this was such a plan'. He himself had been 'totally opposed to the whole idea', which he considered 'an entirely unrealistic brainchild of some youthful and adventurous imagination'. 71 The authors of Operation Mayibuye argued that a takeover transition was both feasible and necessary to end apartheid. Drafted amid ongoing ruthless repression, the plan opened with the declaration that 'very little, if any, scope exists for the smashing of white supremacy' other than 'victory by military means'. 72 In contrast, critics like Bernstein and Fischer doubted that military victory was feasible and feared the 'disastrous' consequences of the 'large-scale violence' necessary to attempt it. Although the prospects of negotiation seemed 'desperately remote', they placed their hope in the idea that the combination of limited guerrilla activity with non-violent campaigns and international pressure could bring about a 'serious political crisis' leading to a negotiated transition. 73 Operation Mayibuye also proposed shifting the primary arena of conflict from the cities to South Africa's rural areas. Even more than its parent bodies, the ANC and the SACP, MK was an urban organisation, in terms of both the social background of its cadre and the geographical location of its activities. Mbeki, however, had been a long-standing critic of the Congress movement's inattention to the rural struggles of the peasantry. His close observation of the rural revolt that took place in Mpondoland in 1960-61 further convinced him that, whereas 'urban-based struggles are more difficult to sustain for much longer than a few days [,] [t]he struggles of the peasants start from smaller beginnings, build up to a crescendo over a much longer time, and are maintained at comparatively lower cost'. 74 Mbeki's advocacy of a 'turn to the peasants' by MK was reinforced by what he later characterised as the two 'most important books on guerrilla warfare that were available at the time'by Mao and Che Guevaraboth of whom 'emphasised the importance of enlisting the support of the peasantry if a revolutionary war was to succeed'. 75 Famously, Guevara stressed that one of the lessons of the Cuban revolution was that 'in underdeveloped America the countryside is the basic area for armed fighting'. In Operation Mayibuye, Mbeki and Slovo argued that in South Africa too, rural areas would be 'the main theatre of guerrilla operations in the initial phase'. 76 Other Congress leaders were unconvinced by Operation Mayibuye's focus on fighting in the countryside. Within the SACP, some had always been sceptical of the Maoist emphasis on 'the contribution of the peasantry [to revolutionary struggle] at the expense of the role of the working class'. 77 There was no disagreement with Mbeki and Slovo's assessment that immediate mass insurrection -'a general uprising leading to direct military struggle'was 'unlikely': in 1963, no one in the Congress leadership was advocating erecting barricades in urban areas, as some had wanted to do in Sophiatown in 1955. 78 But some Congress leaders envisaged urban guerrilla warfare, using the bomb and the bullet, rather than the barricade. Given their fears of intensifying racial division, as well as the influence of Guevara's critique of 'terrorism' against civilians, it is unlikely that many Congress leaders seriously considered the kind of indiscriminate bomb attacks on urban civilian targets used by the FLN during the Battle of Algiers in 1956-57. 79 But some were inspired by the targeted attacks on British police and military targets in Palestine in the 1940s by the urban guerrillas of the Irgun, who, Mandela observed, offered a successful model of 'a movement in a country which had no mountains'. 80 According to Goldberg, some in MK envisaged a next phase in which continued sabotage attacks would be used to lure the security forces into sniper ambushes. 81 Above all, the debate over Operation Mayibuye revolved around the desirable balance and relationship between violent and non-violent action. Drawing inspiration from Guevara's account of the Cuban revolution, Operation Mayibuye envisaged the landing of four groups of guerrillas in the South African countryside. 'As in Cuba', the plan stated, 'the general uprising must be sparked off by organised and well prepared guerrilla operations during the course of which the masses of the people will be drawn in and armed'. Operation Mayibuye envisaged that the infiltration of MK's guerrillas would be accompanied by 'a general call for unprecedented mass struggle throughout the land, both violent and nonviolent'. But, beyond this suggested exhortation, the plan made no specific provision for how non-violent political activity would be organised. 82 Operation Mayibuye thus formalised the lack of interest in organising mass non-violent action that had characterised the practice of the high command since MK's formation. Putting this in writing now aroused a storm of controversy, however. Critics of Operation Mayibuye derided its advocates as 'militarists' who mistakenly believed that 'the answer lay in military action alone'. 83 In response to their 'exclusively military plan', Bernstein developed an alternative proposal, a 'social and political programme which also encompassed armed force' and which stressed the need to combine 'limited guerilla activities, in the form of quick attacks on border outposts, [and] retreat[s] into bases in the British protectorates', with 'national strikes'. 84 These wide-ranging debates that raged within the Congress movement's remaining internal leadership over the purpose and form of guerrilla action were abruptly interrupted and never resolved. On 11 July 1963, police raided MK's de facto headquarters in the Johannesburg suburb of Rivonia, where the secretariat of the SACP central committee was meeting to continue debating Operation Mayibuye. Over the subsequent two years, the police were able to destroy the remaining MK, ANC and SACP structures within South Africa. The ANC's External Mission, comprised of Congress leaders who had left the country, took over the leadership of the Congress movement.

Aftermath: Into Exile
Despite the strong opposition that Operation Mayibuye had encountered among Congress leaders inside the country, core aspects of the plan and the assumptions that underpinned it continued to shape the strategy of the exiled Congress leadership well into the 1970s. Before the Rivonia arrests, Bernstein had already begun to fear that the Congress movement was 'being bounced into a fait accompli' by the advocates of Operation Mayibuye. 85 By interrupting the debate over the plan, the police raid accelerated this process. In exile, strategic assumptions that had provoked furious critique among the internal Congress leadership came to be accepted unquestioningly.
For more than a decade after Rivonia, the Congress leadership thus focused its attention, energy and resources on efforts to launch guerrilla warfare in rural South Africa as the first stage of an escalating conflict that would culminate in 'all-out war' and an armed takeover. As Slovo later explained, the 'old thinking' that guerrilla action could lead to a negotiated transition came to be viewed as 'highly questionable': the ANC's 'strategic objective', the ANC national executive affirmed in 1973, was the 'seizure of power and not reforms o[r] a negotiated transfer of power'. 86 And though this was denied in formal strategy pronouncements, the launch of rural guerrilla operations was treated in practice as a free-standing and sufficient first step towards that objective. 87 In the Wankie and Sipolilo campaigns of 1967-68, MK guerrillas joined infiltrations by members of the Zimbabwe African People's Union into white-ruled Rhodesia in an unsuccessful attempt to establish a 'Ho Chi Minh trail to South Africa'. A plan to infiltrate 50 guerrillas by a sea landing on the South African coast, modelled on Fidel Castro's landing in Cuba, was discussed for several years and implemented in 1972, only to be aborted when the ship broke down.
The Congress movement's exclusive focus on ending apartheid through an escalating guerrilla insurgency that would culminate in armed takeover was in part a reflection of the perceived inadequacy of the independence secured through negotiated transitions elsewhere in Africa. Exiled Congress leaders were unimpressed by the domestic achievements of postcolonial African regimes and frustrated by what they perceived as their inadequate assistance to the anti-apartheid struggle. In a speech in London in 1969, Ruth First directly echoed Fanon's critique of the neocolonial settlements that followed negotiated transitions, insisting that southern Africa would break 'the pattern of sham African independence'. 'The South must be liberated through struggle', First declared, 'not constitutional conferences of hand picked leaders. There will be no green baize negotiations and search for pliant leaders for neo-colonial solutions'. Commitment to guerrilla warfare was the 'best guarantee' of this, for the 'very nature of guerilla warfare' produced 'far reaching and revolutionary goals'. 88 While they were repelled by the outcomes of negotiated transitions, Congress leaders were attracted by the successes of guerrilla struggles elsewhere. In addition to the Cuban model, which had so influenced the authors of Operation Mayibuye, the guerrilla mystique was enhanced by the war in Vietnam and, closer to home, by the escalating conflicts in Angola and Mozambique. Operating in the militarised world of exiled liberation movements and largely cut off from the diverse ties to multiple constituencies within South Africa that had fed into the Congress movement's earlier strategic deliberations, the exiled Congress leadership was especially susceptible to what the analyst J. Bowyer Bell diagnosed in 1971 as 'the myth of the guerrilla'. 89 Officials from the Soviet Union, now the ANC's biggest financial backer, continued to warn Congress leaders against the seductions of Guevarist militarism, cautioning that Castro and Guevara's landing in Cuba had been an 'absolute disaster' and their subsequent success a 'fluke'. 90 Exiled Congress leaders paid lip service to such warnings, but, in practice, as Kasrils later reflected, they spent the 'the heady days of the successes of guerilla war' in the 1960s and early 1970s 'looking for that magic formula of the guerilla in the bush'. 91 What Kasrils characterised as Congress leaders' 'militarist deviation' was reinforced by the transformed composition of the movement they led. 92 This transformation had begun as soon as MK started draining cadres out of the movement's political and labour organisations but was further accelerated by the intense state repression that followed MK's launch. The destruction of Congress structures inside South Africa meant that, by 1965, most of the organised and politically active members of the Congress movement were located outside the country. Of them, the vast majority were the 500-800 MK recruits who had been sent abroad for military training and then been unable to return home. (Those recruits were, moreover, almost entirely male: though women had played a prominent role in popular resistance in the 1950s, and some participated in MK's initial sabotage campaign, only a handful were among those sent abroad). 'The balance of forces within the movement', Bernstein observed, 'shifted steadily towards the military sector'. 93 The nature of the manpower available to Congress leaders shaped their strategic choices. It also created further pressures on them, both from restless MK guerrillas stuck in camps in exile, who demanded to be sent home to fight, and from the governments of host countries, who were concerned about the concentration on their territories of armed men outside their own command.
From the Rivonia raid until the Soweto uprising of 1976, Congress strategy was focused almost exclusively on unsuccessful efforts to launch rural guerrilla warfare with the ultimate objective of ending apartheid by armed takeover. Few, if any, of the participants in the Congress movement's crucial meetings in 1960-61 had believed that this was the strategy they were endorsing when they authorised the abandonment of the Congress movement's reliance on exclusively non-violent methods. That authorisation was a response to changed conditions after Sharpeville, especially the increase in grassroots pressure to respond to the state with violence, but it was only possible to achieve consensus on it within the Congress leadership for two reasons. First, the minority of Congress leaders already committed to a guerrilla-based strategy obscured the extent of their ambitions and indicated their acceptance of the less controversial proposition that limited violence would be used to complement continued non-violent mass campaigns. Second, beneath the surface of the compromise that violence would be limited and complementary, Congress leaders had divergent ideas about the forms of violence they were authorising.
The process that led the Congress movement from this initial decision to abandon exclusive reliance on non-violence to its exclusiveand unsuccessfulembrace of rural guerrillaism was shaped by ambiguity, unilateral action, unintended consequences and state repression. The MK high command stretched the ambiguity of the 1960-61 decisions to its limit and presented the rest of the Congress leadership with a fait accompli when it launched its campaign of spectacular symbolic sabotage in December 1961. The sabotage campaignand the state's reaction to itthen had a rapid militarising effect on the remaining Congress leadership, which, in October 1962, endorsed the idea that MK should escalate to guerrilla warfare. Again, however, beneath the surface of this apparent consensus, Congress leaders had different understandings of the form and purpose of guerrilla action. Among the internal Congress leadership, these differences came to be debated increasingly explicitly following the drafting of Operation Mayibuye in 1963. But that debate was interrupted by the police raid on Rivonia. As leadership of the Congress movement was transferred to the very 91 BL, MSS.Afr.s.2151, 1/2, interview with R. Kasrils by H. Barrell different context of exile, the strategic prescriptions of Operation Mayibuye came to underpin Congress strategy for more than a decade.