Tales of the unexpected: The selection of British party leaders since 1963

Jeremy Corbyn’s election as Leader of the Labour Party in 2015 stunned observers and practitioners of British politics alike. In this article, we first outline a theoretical framework that purports to explain why political parties operating in parliamentary systems choose the leaders they do. We then examine 32 leadership successions involving five major British parties since 1963, and note that many of these were unexpected, in that they were triggered by unforeseen circumstances, such as the sudden death or resignation of the incumbent. Examining each party in turn, we briefly explain why the winners won and identify at least eight cases (a quarter of our sample) where a candidate widely expected to prevail at the outset was ultimately defeated by a ‘dark horse’, ‘second favourite’ or even ‘rank outsider’. Of these, Corbyn’s election in 2015 was the most unexpected and, consistent with the findings of studies of party leadership conventions in other parliamentary systems, namely Canada and Spain, suggests that ideological and policy concerns are sometimes more important than considerations of party unity and electability, especially when a leadership contest is dominated by party activists.


Introduction
Jeremy Corbyn's election as Leader of the Labour Party in 2015 stunned observers and practitioners of British politics (Dorey and Denham, 2016;Quinn, 2016). As Stark (1996, pp. 97-98) explains, few party leadership candidacies in Britain are unexpected, and there are many factors which 'drastically reduce' the potential field.
Selection rules limit the possibilities to MPs; nomination thresholds make it mathematically impossible for more than a certain number of MPs to be nominated for any one contest. But probably more important than these rules are the unofficial criteria for being considered a potential leader. Candidates must be sound in mind and body, have Cabinet or Shadow Cabinet experience, command confidence among colleagues and be able to communicate with the nation via the media. They must also appear capable of satisfying the party's strategic goals of remaining united, winning elections and governing competently. All of this is necessary in order to be perceived as serious ''candidate material''.
As we explain below, Corbyn clearly met the first of these 'unofficial criteria', but arguably few others, if any, besides.
In this article, following Sjoblom (1968) and Stark (1996), we first outline a theoretical framework which purports to explain why political parties in parliamentary systems choose the leaders they do. We then examine 32 leadership successions involving five major British parties since 1963, and note that many of these contests were themselves unexpected, in that they were triggered by unforeseen circumstances, such as the sudden death or resignation of the incumbent. Examining each party in turn, we briefly explain why the winners won and identify at least eight cases (a quarter of our sample) where a candidate widely expected to prevail at the outset was ultimately defeated by a 'dark horse', 'second favourite' or even 'rank outsider'. Of these, Corbyn's election in 2015 was the most unexpected and, consistent with the findings of studies of party leadership conventions in other parliamentary systems, namely Canada and Spain, suggests that ideological and policy concerns are sometimes more important than considerations of party unity and electability, especially when a leadership contest is dominated by party activists.

Who Wins and Why?
In attempting to explain party leadership successions in Britain since 1963, and specifically 'why the winners won', we seek to extend and enhance the work of Stark (1996) and Quinn (2012), who have previously addressed this question for the period from 1963 to 1995 and 2010, respectively. We also seek to expand and advance the growing scholarly literature on party leadership selection in Anglophone and European parliamentary democracies, where the same question has hitherto received little systematic investigation, and others considerably more. Examples here include Kenig's study of 143 party leadership contests and successions in 11 parliamentary democracies between 1964 and 2007 which asks demographic affinities they may share'. Voters' perceptions of the candidates' relative electability, competence to lead the party in the legislature and ability to unite the party and build a strong extra-parliamentary organization, they argue, 'have all been important considerations in some contests'. In addition to these factors, they argue, we must also consider the role of the 'ideological and policy positioning of the candidates'. These are 'not particularly important' in parliamentary contests, which are rarely characterized by profound policy disagreements. Ideology, they conclude, is often an important factor in such contests, but is typically 'far from definitive' (Cross and Blais, 2012b, pp. 147, 149).
When and where policy disputes do occur in party leadership contests, whether or not the choice of party leader is restricted to parliamentarians alone, these often reflect underlying intra-party ideological divisions or disagreements, particularly when a party is choosing a leader following a General Election defeat. On such occasions, the leadership contest is likely to be, to a considerable extent, a de facto post-mortem on why the party had lost -particularly if it had been in governmentwith some leadership contenders (or their supporters) alleging that the incumbent leader had presided over policies which were insufficiently congruent with the party's basic principles -or ideology. This has often been a characteristic of Labour leadership contests following a General Election defeat, with the left attributing such losses to the incumbent leadership's 'betrayal' in failing to pursue sufficiently 'socialist' policies.
However, the Conservative Party has also evinced similar ideology/policy disputes when (s)electing a new leader following electoral defeat. The most notable example was in 1975, when Margaret Thatcher was the surprising (to many commentators) choice of successor to Edward Heath. Although it has widely been claimed subsequently that Thatcher's main appeal was that she simply was not Heath (nor was she closely associated with him, unlike her main rivals in the contest), Bailey and Cowley (2015) argue that her election was partly attributable to a yearning among many Conservative MPs for a different political stance and approach. Although she was certainly not a Thatcherite at this stage, closer inspection of her rhetoric and speeches did evince a nascent change of direction, which many Conservatives hankered after: 'The full details of Thatcherism may not have been clear … but that she stood for something was. Ideology was a determining factor in how MPs voted and a clear factor in how the election was discussed' (Bailey and Cowley, 2015;Cowley and Bailey, 2000). Subsequently, following the Conservative Party's crushing General Election defeats to New Labour in 1997 and 2001, the ensuing leadership contests became a proxy for intra-Party debates about whether it had been too 'Thatcherite', or not 'Thatcherite' enough. Most recently, in the 2016 Conservative leadership contest, when Theresa May was eventually 'crowned' as leader following the elimination or withdrawal of the other contenders, her principal opponent, Andrea Leadsom, had been supported by those Conservative MPs who hankered after a clear reversion to Thatcherism, following the professed social liberalism of David Cameron, who much of the Conservative right had distrusted, even despised, since his election to the leadership in 2005 (Kerr and Hayton, 2015;Dommett, 2015).
As Cross and Blais (2012b, pp. 151-152) explain, extra-parliamentary party members and activists are now formally a part of the leadership selectorate in many parties and might be expected, for several reasons, to be influenced by different factors, and hence behave differently to their parliamentary counterparts.
For one thing, they are certainly less likely than parliamentarians to know the candidates and thus factors such as friendship, and career incentives, are likely to be less significant in their voting decision. Ideology may be a more salient factor in expanded selectorates as May's law [of curvilinear disparity] suggests that party activists (e.g. convention delegates) should be more ideologically extreme and more strongly motivated by policy concerns than are parliamentarians. Conversely, electability might be less of a factor among the activists as, for the most part, their jobs are not dependent on the electoral success of their party. While activists surely prefer their party to be in government, they may not be as willing as MPs to sacrifice ideological concerns for electability.
Cross and Blais note that there is little systematic evidence on the voting behaviour of party members and activists in leadership contests. Most of the evidence that does exist, they explain, relates to the Canadian case, where the selectorate has been expanded for nearly a century, and is 'mostly restricted to studies of delegates to leadership conventions', not electorates in the more recent 'One Member, One Vote' contests. We return to this point in our conclusion, and note that, while unexpected party leaders are rare in parliamentary systems, 'dark horse' candidates have sometimes 'emerged' as winners not only in the British context, but also in other parliamentary systems. Based on the findings of studies of party leadership conventions in two such countries, Canada and Spain, we conclude that Corbyn's election as Labour leader in 2015 appears to support Cross and Blais's contention that ideological and policy concerns can sometimes be more important selection criteria than considerations of a party's internal unity and electability, especially where such contests are dominated by party activists (Cross and Blais, 2012b, p. 153).

Key Selection Criteria in Parliamentary Systems
In 1968, Sjoblom (1968) identified three strategic goals for parties operating in parliamentary systems: to remain united, win elections and implement policies. Adopting this framework, Stark argues that unity is a party's first-order goal. 'Only when a party is functioning as a cohesive unit', he explains, 'can it devote its full Tales of the unexpected attention and resources to the second-order goal of winning elections. Electoral victory gives a party the power to begin to fulfil the third-order goal of turning promises into policies' (Stark, 1996, p. 125).
Parties in parliamentary systems, Stark argues, must consider these strategic goals when selecting their leaders. When assessing the relative merits of leadership candidates, a party's highest priority is to choose someone who will preserve party unity. 'Acceptability to all major segments of the party is the first-order criterion. However, only in extraordinary circumstances does this unity goal become an explicit consideration'. It is only likely to prove decisive on those rare occasions when a leadership contest takes place when a party is 'dangerously divided'. Otherwise, leadership candidates are assessed principally on the basis of their perceived ability to lead their party to victory in a General Election. Hence, electoral appeal, or 'electability', is the second-order criterion. In most instances, Stark argues, these first-and second-order priorities are the main criteria used to assess leadership candidates. Some consideration must also be given, however, to their perceived ability to serve in government. Hence, 'competence', or the capacity to lead a successful administration, is the 'third-order' criterion (Stark, 1996, p. 126).
Valuable though Stark's refinement of Sjoblom's model is, the criterion of 'acceptability' needs to be considered more carefully than has traditionally been the case. There are two reasons for critiquing the seemingly common-sense notion of 'acceptability'. The first concerns the subjectivity of such judgements, because it depends on who is assessing the 'acceptability' of a leadership candidate, and according to what criteria. A party's parliamentarians are highly likely to adopt a different definition of 'acceptability' when evaluating leadership candidates, in comparison to their extra-parliamentary members or affiliates. For the former, 'acceptability' will often be inextricably linked to the candidates' likelihood of achieving party unity, and of leading it to victory in the next General Election. By contrast, while a party's grass-roots supporters would ideally prefer to see their party in government, their interpretation of 'acceptability' will sometimes place a rather higher emphasis on ideological stance or purity of principles, even if these render the party less attractive to the wider electorate.
The second reason why the criterion of 'acceptability' warrants more careful consideration is that it is often relational; that is, a leadership candidate might be judged 'acceptable' not only in terms of their own personal characteristics or credibility, but also in comparison to the perceived 'unacceptability' of their opponents -as was the case, for example, with Alec Douglas-Home in the Conservative Party leadership 'contest' of 1963 (Macmillan, 1973, p. 514).
Also warranting more careful consideration is Stark's notion of electoral appeal or 'electability', as this criterion is also subject to differing interpretations, depending either on ideological stance in a party, or/and whether the judgement is being made by a party's parliamentarians or grass-roots members. In many cases, MPs will choose a leader who is not considered to be 'too ideological', because this would militate against both promoting party unity and maximizing the party's popularity by appealing to the 'median' voter (who is presumed to be repelled by perceived dogmatism). Of course, this raises an epistemological issue, namely the manner in which MPs make their judgements or acquire knowledge about the electability of a leadership candidate vis-à-vis electoral or public opinion. Unless these MPs carefully peruse opinion polls conducted at the time of the leadership election, or perhaps undertake their own private polling (unlikely, due to the resource constrains in terms of time and money), judgements about the candidates' 'electability' are likely to remain intuitive and subjective, and perhaps shaped by a priori assumptions about who the median voter is, or where the hallowed centreground of British politics is located at any particular juncture.
However, some MPs, and more particularly some of a party's grass-roots members, will be convinced that their party's 'electability' would be increased by adopting a stronger or more explicitly ideological stance. Certainly, if their party has just been defeated in a General Election, there will be some MPs, and probably rather more extra-parliamentary members, who insist that the absence or dilution of its professed ideology was a major reason for its defeat, either because erstwhile supporters felt the party had betrayed them by abandoning its basic principles, or because its perceived lack of ideology caused confusion among many voters about what exactly the party stood for, and which policy objectives it would pursue in government.
In the case of the Labour Party, for example, the left has frequently attributed the Party's lack of electoral appeal to an absence or abandonment of 'socialist' ideology and policies or, at least, a failure by the erstwhile leadership to explain and promote these with sufficient commitment and conviction. Certainly, the Bennites in the early 1980s, and the Corbynites today, were and are adamant that Labour's defeats in the 1979 and 2015 General Elections, respectively, were due not to the Party being too 'left-wing', but because it was too timid and 'conservative', and had thus 'betrayed' its supporters. Similarly, since 1990, many Conservatives have attributed each and every failure to win a General Election to the leadership's refusal to adhere to 'Thatcherism' -a charge levelled against David Cameron by the Conservative right following the 2010 General Election campaign, in which he had supposedly 'muddied the message'.
Since 1963, there have been 32 leadership successions involving five major British parties: Labour, the Conservatives, the Liberal Party, the SDP and the Liberal Democrats (the latter party having been created in 1988 following a merger between the Liberal Party and the SDP). Of these, 11 involved the Labour Party, 11 the Conservatives and 10 the Liberal Party, SDP and Liberal Democrats combined. In the following three sections, we examine the leadership successions involving each of these parties and, applying the key selection criteria of 'acceptability', 'electability' and 'competence' described above, briefly explain 'why the winners won'. In the first, we consider those involving the three 'third' parties: the Liberal Party, the SDP and the Liberal Democrats.

From Thorpe to Farron: The Liberal Party, the SDP and the Liberal Democrats
Between 1967 and 2015, there were 10 Liberal, SDP and Liberal Democrat leadership successions, of which two involved the Liberal Party, three the SDP and five the Liberal Democrats (see Table 1). On every occasion, the party in question was seeking to fill a vacancy, and there were no formal challenges to the incumbent. In 1982 and 1988, two new parties, the SDP and the Liberal Democrats, were seeking to fill the post of Party Leader for the first time. On every other occasion, a leadership succession was precipitated by the incumbent's resignation, of which several, notably those of Jo Grimond in 1967, Jeremy Thorpe in 1976 and Charles Kennedy in 2006, were sudden and traumatic for the party concerned and, to most outside observers, entirely unexpected.
In 1983 and 1987, the succession for the SDP leadership was uncontested. In 1967, Jeremy Thorpe was elected by Liberal MPs. On every other occasion, the leader was chosen via a ballot of party members. On the eight occasions when the leadership was contested, the key selection criteria ('acceptability', 'electability' and 'competence') were arguably the same irrespective of the composition of the electorate. However, with the possible exception of the 1982 SDP contest, Liberal, SDP and Liberal Democrat MPs and party members did not anticipate that their new leader would ever become Prime Minister. Consequently, the key selection criteria on most occasions were arguably 'acceptability' and 'electability' -namely 'to choose someone who would not divide the party and who might also draw favourable attention to the party's efforts' (Stark, 1996, p. 127).
The SDP's leadership contest of 1982 was different in this respect because, over the previous year of its brief existence, there had been reason to believe the Party might soon help to form a government. As Stark (1996, p. 129) explains,  (Thorpe, 1999, p. 132). Liberal MPs recognized that he was the Party's most effective campaigner. He was now, after Grimond, the longest-serving Liberal MP and 'seemed to be his natural successor as leader. A highly extrovert personality, an excellent and regular performer on television and radio, he was already one of the party's best-known figures' (Wallace, 1983, p. 59). 'Politician with a zest for action' was The Guardian's verdict in a headline published the morning after his election (The Guardian, 19 January 1967).
In 1976, both candidates, David Steel and John Pardoe, were 'acceptable' to Liberal Party members. Of the two, they concluded that Steel had the greater electoral appeal. In 1988, Liberal Democrat members chose Paddy Ashdown over Alan Beith, principally for the same reason (Stark, 1996, pp. 127-128, 130-131). As Ashdown's Press Officer during and after the leadership campaign, Harriet Smith (1999, p. 21) later recalled.
It was ultimately Ashdown's desire to take risks with the future that ensured his success. His approach was more acceptable to the ex-SDP element (which at that time made up less than one-third of party members) and to those ex-Liberals who feared that a traditional Liberal-style leader would restrict the new party's appeal to potential new members and voters.
In 1999, Charles Kennedy and Simon Hughes were both 'acceptable' to party members. While Hughes succeeded in capturing the votes of most of the party's activists, Kennedy's support came largely from 'armchair' members, among whom he was the best-known candidate, owing to his media profile (Hurst, 2006, p. 101). As most of those who voted fell into the latter category, Kennedy won, principally because of his perceived electoral appeal. In 2007, Nick Clegg and Chris Huhne were both 'acceptable' to party members. Again, Clegg won principally because, in their opinion, he was the more 'electable' of the two. According to a YouGov poll for Sky News, 53 per cent of party members believed that, of the two, Clegg had the greater electoral appeal, whereas only 9 per cent felt this was so of Huhne (Quinn, 2010, pp. 107-108;YouGov, 2007). In 2015, Tim Farron defeated Norman Lamb, partly for 'ideological' reasons (unlike Lamb, he had strongly opposed unpopular policies, notably the trebling of university tuition fees, introduced by the Conservative-Liberal Democrat coalition government), but principally, given the defeat of all but eight of the party's 57 MPs in the 2015 General Election, on the basis of his superior electoral appeal (Francis and Denham, 2015).
Of the eight 'third' party contests, the 'favourite' and 'front-runner' invariably went on to win. In 1967, Thorpe was seen as the likeliest candidate to replace Grimond and also had 'the overwhelming support both of English Liberals in the constituencies and of the Young Liberals, who bombarded the MPs and Party Headquarters with telegrams declaring Jeremy to be their choice' (Bloch, 2014, pp. 244-245;Cole, 2011, p. 159;Ingham, 2015, p. 315;Wallace, 1983, p. 59). In 1976, following a three-week campaign, Steel won decisively, receiving almost two-thirds of the votes and securing a majority in every region, except Pardoe's 'home' territory of Devon and Cornwall (Punnett, 1992, p. 138). In 1988, the effect of the Liberal Democrat campaign, if any, was to increase Ashdown's margin of victory. Ashdown was described as the 'man to beat' from the moment Steel announced he would not be a candidate. During the campaign itself, Alan Beith's attempts to improve his chances 'tended only to backfire' and weaken his support (Stark, 1996, p. 120;The Times, 12 and 21 May 1988). In 1999, Kennedy 'started as the clear front-runner and … bookies' favourite' and 'maintained a comfortable lead' throughout (Alderman and Carter, 2000, pp. 319, 324). In 2006, despite Campbell's allegedly 'lacklustre' performance during the campaign, the 'favourite', 'front-runner' and acting Party Leader secured a comfortably victory (58 per cent to Huhne's 42 per cent) in the end (Denham and Dorey, 2007, p. 43).
The effect of the 1982 SDP and 2007 and 2015 Liberal Democrat campaigns was in the opposite direction to 1988. In the 1982 SDP contest, Jenkins (who, according to The Times, had established 'an irresistible claim to the leadership' and was referred to in the same newspaper as the Party's 'yet to be anointed leader' following his re-election to the House of Commons in March) led throughout (The Times, 26 and 27 March, 1982). Owen, however, succeeded in closing much of the gap between Jenkins and himself over the course of the campaign, as did Huhne between the 'favourite', Clegg, who was 'quickly identified as the front-runner by journalists (and bookmakers)' and attracted the support of at least 39 of the party's 63 MPs, and himself during the Liberal Democrat campaign of 2007 (Branigan, 2007;Quinn, 2010, p. 107;Stark, 1996, p. 120). Indeed, following the latter contest, it was reported that had some 1,300 ballots delayed in the Christmas post been counted, Huhne would have won (Merrick, 2008). The closeness of the result was 'partly down to Huhne's aggressive campaign and activist-friendly policies', but also 'the perceived lacklustre campaign that Clegg ran' (Quinn, 2012, p. 154). In 2015, Farron was the clear 'favourite' from the outset and, although Lamb succeeded in closing the gap to some extent during the campaign, the result, once again, was never in doubt.
In sum, all eight 'third' party contests were won by the candidate widely expected to do so at the outset. In 1967, the Liberal 'campaign' was brief and conducted entirely in private. As it turned out, Steel's role as Thorpe's unofficial campaign manager was 'critical' in seeing his preferred candidate, and that of the outgoing leader, Jo Grimond, safely home to victory (Bartram, 1981, pp. 68-71;Beaumont, 1997). The subsequent campaigns were longer, more expensive to organize and more conspicuous, but frequently described as 'bland' (Francis, 2010). Most received limited coverage in the national news media (Alderman and Carter, 2000), partly because of this perceived 'blandness', but also because whoever was elected was unlikely ever to become Prime Minister. In most cases, 'third' party contests were won and lost principally on the basis of 'electability', albeit in conjunction with voters' perceptions of the other strengths and weaknesses of rival candidates. For example, in retirement David Steel recalled of the Liberal leadership contest of 1976 that there was 'not much on policy to separate' Pardoe and himself. His personal assistant endorsed this view, noting that the contest had been principally about 'style and strategy', with Steel campaigning on his 'solid record', 'shrewd toughness', 'calm pragmatism' and 'obvious' electoral appeal (Steel, 1989, p. 114). During the campaign itself, a poll carried out by Marplan found that 'amongst the general public 64 per cent preferred Steel, compared with only 23 per cent who wanted Pardoe. Among Liberal voters, the margin had been wider, 68 per cent to 20 per cent' (Bartram, 1981, p. 128). When the result was announced, it transpired that Steel had won 12,541 (64.1 per cent) 'national' votes compared to Pardoe's 7,032 (35.9 per cent), once the ballots cast by 69,726 individual Party members had been 'weighted' to reflect regional variations in the size and strength of local parties (Bartram, 1981, p. 128;Stark, 1996, pp. 5, 73-74).

From Home to May: The Conservative Party
Between 1963 and 2016, there were 11 Conservative Party leadership successions (see Table 2). Of these, the Party was seeking to fill a vacancy on seven occasions, following the incumbent's resignation. The remainder were triggered by a formal challenge to the incumbent, of which one (to John Major in 1995) followed the incumbent's own prior decision to resign as Party Leader, thereby forcing his opponents within the parliamentary Party to 'put up or shut up' (Alderman, 1996;Heppell, 2008, pp. 98-114). Major's resignation was sudden and unexpected, as were those of Harold Macmillan in 1963 (Denham andO'Hara, 2008, p. 16;Heppell, 2008, pp. 16-18) and David Cameron in 2016. Decisions to challenge the incumbent in a formal contest for the Party leadership, notably those of Margaret Thatcher in 1975 and Sir Anthony Meyer in 1989, were similarly unexpected (Cowley and Bailey, 2000;Heppell, 2008, pp. 59, 74-75;Norton, 1990;Wickham-Jones, 1997). In 2003, the succession was uncontested. In 1963, following Macmillan's resignation, Sir Alec Douglas-Home was informally selected by the 'magic circle' (Gilmour and Garnett, 1997, pp. 186-203;Heppell, 2008, pp. 18-31). Between 1965 and 1997, four new leaders were elected, and two incumbents re-elected, by MPs alone. In 2001 and 2005, following a series of eliminative ballots among MPs, a new leader was elected via a ballot of Party members. On all 10 occasions when the leadership was contested, the key selection criteria ('acceptability', 'electability' and 'competence') were the same, irrespective of the composition of the selectorate.
In 1989 and 1995, two Prime Ministers, Thatcher and Major, defeated their respective challengers (the backbench MP and 'stalking horse' candidate, Meyer, and erstwhile Secretary of State for Wales, John Redwood) with predictable ease and were re-elected on the basis of all three of the 'standard' criteria (Alderman, 1996;Norton, 1990;Stark 1996, pp. 130, 169). Conversely, in 1965 and 2005, there was little to choose between the two leading contenders on the first criterion of 'acceptability', but Heath and Cameron were seen as superior to Reginald Maudling and David Davis, respectively, on the second, and won principally on the basis of 'electability' (Denham and Dorey, 2006, p. 41;Heppell and Hill, 2009;Kerr, 2007;Quinn, 2010, p. 114;Stark, 1996, pp. 127, 132).
Of the other six cases listed in Table 2, the leadership contest of 1975 is arguably unique, in that the winner, Thatcher, was not seen as the strongest candidate on any of the three 'standard' criteria. Indeed, there was a widespread expectation that Heath would be re-elected, in spite of having lost -albeit very narrowly -two general elections in 1974. Certainly, Heath was advised, by the Conservatives' Chief Whip, that 'according to soundings by the whips, [he] would be re-elected as leader of the party', an outcome which was also suggested by a consultation exercise conducted by the Party's 1922 Committee, which surveyed the preferences of Conservative peers and the extra-parliamentary Party. Heath was further encouraged by a Harris Poll published in the Daily Express on 3rd February (the day before the ballot) that his continued leadership was supported by 70 per cent of Conservative voters (Heath, 1998, pp. 532-533).
Yet, when the result of the ballot was announced, it was evident to Heath, and everyone else, that: 'We had all got it so badly wrong. Mrs Thatcher, against all expectations, had won the first ballot with 130 votes to my 119', whereupon Heath immediately withdrew from the contest (Heath, 1998, p. 334). The shock of this result reverberated far and wide, with an official in the United States' White House observing that 'Heath was unexpectedly defeated as Opposition leader', although the same official deemed it 'doubtful that Mrs Thatcher will win on the second ballot. A more likely victor is [the] popular Willie Whitelaw' (Ford Library, 1975a). Whitelaw himself, as Party Chairman, was shocked at Heath's defeat, having not realized the scale of 'the strong feeling against Ted Heath' in the parliamentary Party (Whitelaw, 1989, p. 141).
Having defeated Heath on the first ballot, Thatcher's principal opponent on the second, William Whitelaw, was generally presumed to be the more 'acceptable', 'electable' and 'competent' of the two. Yet Thatcher again performed much better than expected, attracting 146 votes compared to Whitelaw's 79, whereupon the aforementioned White House official remarked that this victory was 'as surprising as Whitelaw's poor showing', and indicated a considerable 'degree of unity within the Conservative Party', although he added that in the longer-term, 'she will have to move an appreciable distance from her position on the right wing of her party' (Ford Library, 1975b).
Whether or not Thatcher won because she was viewed by Conservative MPs as a right-wing candidate who would imbue the Party with more ideological vigour in challenging Labour has been the subject of considerable academic debate. Some scholars have contended that there were 'clear ideological forces at work' and that the Party's MPs 'wanted a shift. They wanted new policies and they wanted a turn to the right' (Bailey and Cowley, 2015;Cowley and Bailey, 2000, p. 599;Wickham-Jones, 1997, p. 89). However, Thatcher's right-wing stance at this time should not be exaggerated, for she was not yet a Thatcherite -she became more radical over time, partly spurred by the experience of being in office (just as Tony Benn moved decisively to the left during the 1970s, having been a 'technocratic' Minister in the late 1960s). Whatever expectation of ideological change existed among some Conservative MPs in 1975, it is still widely believed that the 'one principal reason' for Thatcher's victory was that she was 'not Heath' (Stark, 1996, p. 127). Certainly, even some of those Conservative MPs who did not vote for her acknowledged that her support was drawn from all sections of the parliamentary Conservative Party, not just the right (Prior, 1996, p. 99).
The remaining five winners were chosen principally on the grounds of 'acceptability'. In 1963, Douglas-Home 'emerged' as the candidate arguably best-placed to maintain party unity -or as one of his defeated rivals expressed it, 'the product of the tendency of the Conservative Party to play safe instead of taking a calculated risk' (Hailsham, 1975, p. 224). Douglas-Home himself recalled that he had not intended to seek the leadership, but was exhorted to do so by several senior Conservative parliamentarians who 'insisted that I could command the support of a solid majority in the Commons' (Home, 1976, p. 183).
In 1990, following Thatcher's resignation after the first ballot, Major was seen by the Party's MPs as superior in this respect to the instinctively pro-European Michael Heseltine. The importance of Europe in this contest is particularly instructive, as the problems Major subsequently encountered as leader were caused by the same issue (Cowley and Garry, 1998). According to one study, ideology was the single most important factor in the voting behaviour of MPs on the second ballot. The majority of 'pure Thatcherites' voted for Major, while most 'critics of Thatcherism' supported Heseltine or Douglas Hurd. The ultimate reason for Major's success, however, was that he won most support among the 'Party faithful', who constituted the majority of Conservative MPs (Cowley, 1999). For similar reasons, William Hague was preferred by the Party's MPs to the instinctively pro-European Kenneth Clarke in 1997 (Heppell and Hill, 2008), and Iain Duncan Smith preferred to the same opponent by its wider membership in 2001 (Alderman, 1998;Carter, 1991, 2002;Heppell and Hill, 2010;Stark, 1996, pp. 127, 130, 132). In 2016, Theresa May was seen by most of the Party's MPs as the candidate most likely to restore party unity following the British electorate's decision in a referendum to leave the European Union and Cameron's subsequent resignation. According to the opinion polls, May was also seen by the general public (43 per cent) and Conservative voters (59 per cent) alike as the candidate best-equipped to serve as Prime Minister (Dunford and Kirk, 2016). As a long-serving Home Secretary, she was also seen as the candidate most 'competent' to serve as Prime Minister, and so lead a successful administration.
Of the ten Conservative contests since 1963, only four have been won by the 'front-runner'. In 1989, Thatcher was the clear favourite at the outset and maintained a large lead throughout, as did Major in 1995 (Alderman, 1996;Norton, 1990;Stark, 1996, pp. 117-118, 168-169). In 1997, Hague was 'widely tipped as the eventual winner' at the close of nominations (Alderman, 1998, p. 10), as was May in 2016. On the other six occasions, the campaign profoundly affected, and ultimately changed, the outcome. In 1963, three candidates (R.A. Butler, Lord Hailsham and Maudling) conducted themselves in a manner that proved to be selfdefeating, while Douglas-Home prevailed by 'simply letting his name go forward and doing nothing to alienate any section of the Party' (Stark, 1996, pp. 106-107). In 1965In , 1975In and 1990, the 'favourites' were Maudling, Heath and Thatcher, respectively, but all were guilty of complacency and their principal opponents on the first ballot (Heath, Thatcher and Heseltine) conducted far more active and wellorganized campaigns (Stark, 1996, pp. 107-111). On each occasion, the campaign resulted in victory for a 'dark horse' candidate who was not initially thought to be in the running (Douglas-Home) or did not contest the first ballot of MPs (Major), the 'second favourite' (Heath in 1965) (Baston, 2004, pp. 254-255, 260;Daily Express, 27 July 1965, pp. 1-2) or an 'outsider' (Thatcher in 1975).
In 2001 and 2005, the 'front-runners' were Portillo and Davis, respectively, but like Butler and Hailsham (1963), Maudling (1963and 1965, Heath (1975) and Thatcher (1990) each conducted himself in a manner that proved to be selfdefeating, and ran an inferior campaign to that of his principal opponent(s) (Clarke and Duncan Smith in 2001, Cameron in 2005. As a result, both lost 'momentum' and were either eliminated (Portillo) or 'went backwards' (Davis) during the first 'phase' of the contest, which consisted of an initial 'screening' of candidates via a series of eliminative ballots among MPs. In both cases, expanding the leadership selectorate to give Party members the final decision between the top two candidates selected by MPs had resulted in longer, more expensive and more conspicuous campaigns. This, in turn, resulted in Portillo and Davis making costly mistakes and gave two 'outsiders', Duncan Smith and Cameron, the chance to become more widely known within the Party, build 'momentum' and ultimately win against the odds (Alderman and Carter, 2002;Heppell, 2008, pp. 136-153, 174-193).
In sum, of the ten Conservative leadership contests since 1963, six have been won by a candidate who was not expected to do so at the outset. Of the six unexpected winners, however, most were seen by the Party's MPs as 'serious candidate material', in that they appeared to meet at least some of the 'unofficial' criteria for being considered a potential leader. Many of the Party's MPs, however, recalled his perceived disloyalty to the previous Conservative government, and few considered him 'leadership material'. As a result, only 39 of them voted for him on the first ballot, 42 on the second and 54 on the third. In 2005, Cameron had only seven months of Shadow Cabinet experience prior to his election as Party Leader, but was seen by most of the Party's MPs as the candidate best able to communicate with the nation via the media (and hence the most 'electable' as Prime Minister). Taking this 'unofficial' criterion into account, 90 of them voted for Cameron on the second ballot, almost as many as did so for Davis (57) and Liam Fox (51) combined.

From Wilson to Corbyn: The Labour Party
Since 1963, there have been 11 Labour Party leadership successions (see Table 3). Of these, the Party was seeking to fill a vacancy on nine occasions, and two were precipitated by a formal challenge to the incumbent. Of the nine contests for the vacant position of Party Leader, two were triggered by the sudden and unexpected death of the incumbent (Hugh Gaitskell in 1963, John Smith in 1994. Of the other nine leadership successions, some (including the resignation of Harold Wilson as Prime Minister in 1976 and the formal challenge to Neil Kinnock by Tony Benn in 1988) were also sudden and, to most outside observers at least, almost entirely unexpected.
When Blair stood down in 2007, the succession by Gordon Brown was uncontested. In 1963In and 1980, three new leaders were elected by the Party's MPs. Between 1983 and 2010, four new leaders were elected and one incumbent re-elected via an Electoral College, consisting of three party sections: the Parliamentary Labour Party (PLP), members of Constituency Labour Parties (CLPs) and affiliated organizations, principally trade unions. In 2015 and 2016, a new leader was first elected, and then re-elected, via a new 'One Member, One Vote' system, discussed below. On all 10 occasions when the leadership was contested, the key selection criteria 'acceptability', 'electability' and 'competence' were the same, irrespective of the composition of the electorate.
Harold Wilson was seen as superior to George Brown and James Callaghan on all three criteria in 1963 -Callaghan himself subsequently conceded that 'Harold Wilson had the stability and unifying influence that the Party needed' at that time (Callaghan, 1988, p. 150) -as was Callaghan to the five other candidates (Benn, Tony Crosland, Michael Foot, Denis Healey and Roy Jenkins) who stood in 1976. The 1980 contest was decided principally on the basis of 'acceptability'. Considerations of party unity were paramount, and Foot was considered the best candidate around whom the Party could unite, despite Healey's superior claims to both 'electability' and 'competence'. The 1983 contest was, again, decided principally on the basis of 'acceptability'. On this occasion, Neil Kinnock's 'softleft' background and refusal to serve in the Wilson and Callaghan governments  (Stark, 1996, pp. 127-131). With the possible exception of Wilson in 1963 (Heppell, 2010a(Heppell, , 2010b, the 'expected' winner of all seven of these contests at the close of nominations (Healey had been the clear 'favourite' in 1980, but subsequently lost that status following Foot's belated decision to enter the race) invariably went on to win Carter, 1993, 1995;Drucker, 1976Drucker, , 1981Drucker, , 1984Heppell, 2010b;Heppell and Crines, 2011;Stark, 1996, pp. 118-120). However, the 'trend' established by the previous seven contests was halted in 2010, and then, in 2015, dramatically reversed. In 2010, there were five candidates, but the contest was effectively a two-horse race between the 'favourite', David Miliband, and his younger brother, Ed. Of the two, David Miliband was the more experienced, but both were seen by their parliamentary colleagues as 'serious candidate material'. Of the first preferences indicated on their ballot papers, David Miliband received 111 votes, while Ed secured 84. After four months of campaigning, Ed Miliband prevailed by the narrowest of margins. David Miliband had remained throughout the first choice of more Labour MPs and Party members than any other candidate, but Ed's far superior support in the affiliates' section of the Electoral College and the larger number of second preference votes he received in both the PLP and CLP sections proved decisive. According to the opinion polls, David Miliband was seen as the strongest candidate in terms of both 'electability' and 'competence', but he was clearly 'unacceptable' on ideological grounds to the leaders of the three largest trade unions affiliated to the Party (the GMB, UNISON and UNITE). Of the two, Ed was more 'acceptable' to the Electoral College as a whole (Dorey and Denham, 2011;Jobson and Wickham-Jones, 2011;Pemberton and Wickham-Jones, 2013;Quinn, 2012, pp. 64-82).
Ideology again prevailed over electoral and pragmatic considerations in 2015 (Dorey and Denham, 2016;Quinn, 2016), albeit even more so than it had in 2010. The 'front-runner' was Andy Burnham and, based on the number of nominations received from Labour MPs, either Burnham (with 68) or Yvette Cooper (59) would have won had the choice of Party Leader been theirs alone. In 2014, however, a major change had been introduced to the Party's leadership election process, which was to have an instant, and dramatic, effect on the outcome. In order to reduce the influence of trade union leaders and officials in future leadership contests, the Electoral College had been replaced by a new system, in which there would be three categories of voters: full Party members, 'affiliated supporters' (mostly trade union members) and 'registered supporters', who could register to vote in any future leadership contest, having declared their support for Labour's 'aim and values' and paid a modest fee of three pounds. Henceforth, the Party Leader would be elected not by the PLP, or an Electoral College, but on the basis of 'One Member One Vote'.
In practice, the new system led to a dramatic surge in the number of both 'affiliate supporters' and 'registered supporters', most of whom voted for the veteran left-wing backbench MP, Jeremy Corbyn. Whereas Corbyn received just under 50 per cent of the votes of full Party members, he was the first choice of 57.6 per cent of 'affiliate supporters' and no fewer than 83.7 per cent of 'registered supporters'. Apart from the sheer scale of Corbyn's victory, the most notable feature of the contest was that most of those who voted for him were strongly motivated by ideological considerations. Only 5 per cent of his supporters believed he would maintain party unity, and only 5 per cent believed he would lead Labour to victory in the next General Election. As these figures suggest, Corbyn was seen as by far the weakest candidate on all three of the 'standard' criteria: 'acceptability', 'electability' and, as the only one with no Cabinet or Shadow Cabinet experience, 'competence' (Dorey and Denham, 2016;Quinn, 2016). In 2016, following a vote of no confidence by the Party's MPs, Corbyn was formally challenged by Owen Smith. Unlike the previous contest, Corbyn was expected to win, and did so, achieving an increased majority (62 per cent to Smith's 38 per cent), and for essentially the same (principally ideological) reasons that he had secured the leadership almost exactly a year before.

Conclusion
In this article, we first outlined a theoretical framework which purports to explain why political parties in parliamentary systems choose the leaders they do. We then examined the 32 leadership successions involving five major British parties since 1963, in order to explain 'why the winners won'. While most winning candidates were expected to prevail at the outset, we have identified several cases from Douglas-Home in 1963 to Corbyn in 2015 where the expected winner was ultimately defeated by a 'dark horse' candidate, 'second favourite' or even 'rank outsider'.
The decision by major British parties to expand their leadership selectorates beyond the parliamentary caucus has resulted in longer and more conspicuous campaigns. On four occasions since 2001, this has given a candidate not expected to win at the outset of a leadership campaign (Duncan Smith in 2001, Cameron in 2005, Ed Miliband in 2010, Corbyn in 2015 the time and opportunity to build 'momentum' and upset the odds. On four previous occasions between 1963 and 1990, an 'unexpected' winner was informally selected by a 'magic circle' of party notables (Douglas-Home in 1963) or formally elected by MPs alone (Heath in 1965, Thatcher in 1975and Major in 1990. Of these eight examples, Corbyn's election as Labour leader in 2015 was the most 'unexpected'. Previous 'unexpected' winners were seen by their parliamentary colleagues as 'serious candidate material', in that they appeared to meet at least some of the 'unofficial' criteria for being considered a potential leader. As we noted in our introduction, unexpected party leaders are rare in parliamentary democracies. However, 'dark horse' candidates have sometimes 'emerged' as winners of formal, competitive contests not only in Britain, but also in other Anglophone and European democracies. As Cross (2014, pp. 184-185) explains, Leadership elections in Canada are typically competitive events … Many leaders have won on the fourth or fifth ballot including Liberals Trudeau and Dion, Conservatives Stanfield, Clark and Mulroney, and New Democrats Broadbent, McLaughlin and Mulcair … dark-horse candidates who trail on first ballots are occasionally able to receive majority support and win several ballots later. Stephane Dion and Joe Clark are both examples of candidates finishing third on the first ballot who ultimately won on a fourth ballot.
In the case of Spain, Barbera et al. (2014) note that Zapatero (Spanish Socialist Workers Party) in 2000, Hernandez-Mancha (People's Alliance) in 1987, Llamazares (United Left) in 2000 and Imaz (Basque National Party) in 2004 were not seen as front-runners at the outset of their respective contests, yet all were elected. In all four cases, they argue, these candidates 'challenged the party establishment' and 'sought to replace the incumbent generation' (Barbera et al., 2014, p. 120). In the British cont1ext, it can be argued that, for their part, Corbyn and his supporters similarly 'challenged the Party establishment' and sought to 'replace the incumbent generation' of apparatchiks and 'career' politicians in contesting the Labour leadership election of 2015.
As Cross and Blais (2012b, p. 152) note, Johnston's (1988) study of leadership conventions in the Canadian Conservative and Liberal parties in 1983 and 1984 asks whether ideology and policy were important factors in delegates' voting decision. It finds that 'The decisive factors in both Liberal and Conservative delegates' behaviour seemed to be ideas about policy' and that 'Most important in each party was left-right self-designation' (Johnston, 1988, p. 215). Cross and Blais (2012b, p. 153) conclude that the limited evidence we have on the role of ideology in parliamentary and extra-parliamentary leadership contests suggests that 'Policy concerns may be most important in contests dominated by party activists … while more casual partisans, who participate in membership votes, and MPs, voting in the party room, may be equally or more concerned with questions of party unity and electability'. While most major British party leaders since 1963 have been selected by MPs and party members on the basis of their perceived 'acceptability', 'electability' and 'competence', Corbyn's election in 2015 and subsequent reelection in 2016 also suggests that ideological and policy concerns can sometimes be more important than considerations of a party's internal unity and electability, especially in contests 'dominated by party activists'.