The study of government popularity functions has been popular among political scientists since the pioneering work of the early 1970s. Most of the studies have concentrated on data from the United States or from Great Britain. The reason for this has probably been the availability of data over long periods of time in these countries, and the fact that in two-party systems the effects of economic changes on government popularity seem to be stronger than in countries with multi-party government coalitions. However, some studies have also shown that in countries with multi-party governments the effects of economic conditions are similar, although possibly slightly less strong.

The concept of cross-cutting cleavages has provided an influential explanation of why some regions with severe ethnic divisions have been able to establish and maintain political stability, while others have experienced violent conflict. It is suggested that where a deep communal cleavage is cross-cut by other politically salient divisions, a network of overlapping group memberships is created. This generates a series of divided loyalties that reduce the internal cohesiveness of any one group and encourages greater social and political stability. In the absence of such cross pressures, cumulative, reinforcing allegiances create ‘fragmented’ political cultures where all the politically relevant sources of division lie parallel, where group loyalty is paramount and where intercommunal hostility is the norm.

This Note explores the candidate-endorsement process in the Liberal Democratic Party (LDP) of Japan during its period of hegemony (1955–90). Even in parties without an enduring factional structure such as the LDP, nominations are often troublesome – witness, for example, the reselection controversy in Britain's Labour party at the end of the 1970s or the perennially damaging fights in American primary elections. Moreover, it is easy to understand why nomination politics is so consistently problematic: the gist of the problem is simply that different groups within a party may differ as to who should receive the party endorsement in a given district (or, in list systems, who should get the safe spots on the list). Group A naturally wants its candidate(s) endorsed (there may be more than one in multi-member districts), but so do groups B, C and D. The resulting interaction between groups can be what a game theorist would call a co-ordination, or Battle of the Sexes, game.

No event in recent times was more unexpected than the assassination of Sweden's prime minister, Olof Palme, shortly before midnight on 28 February 1986. The murder sent a shock wave throughout the Western world, especially, of course, among Swedish citizens. By noon the next day, virtually all Swedes had learned about his death. People in Sweden displayed their emotions publicly in an unprecedented manner. Church attendance, which had dwindled for decades, suddenly, albeit temporarily, soared.

Secession, defined as ‘formal withdrawal from a central authority by a member unit’, has been particularly rare in democracies. In fact, there has never been a single case of secession in democracies if we consider only the well-established ones, that is, those with at least ten consecutive years of universal suffrage. The cases most often mentioned happened only a few years after the introduction or significant expansion of universal suffrage: Norway and Sweden in 1905, Iceland and Denmark in 1918, Ireland and the United Kingdom in 1922. What is more, one would hesitate before calling the first two cases real secessions, since the ties between the political entities involved were very loose at the outset. Secessionists never managed to split a well-established democracy through a referendum or an electoral victory. We must conclude that it is very hard for them to achieve and maintain the magic number of 50 per cent support. My aim is to explain why this is the case.

If their treatment of power indices is anything to go by, reputable social scientists have a surprising tendency to lose touch with reality when using elementary mathematics. R. J. Johnston's article in this Journal is (unfortunately) a good illustration of this. The subsequent exchange between Johnston and Garrett, McLean and Machover also fails to get to the heart of the matter.

No one who has read either Rawls's Political Liberalism or my An Essay on Rights can fail to profit from the penetrating analysis to which Peter Jones has subjected them.

Disposed to agree with most of what he says about the former, I am also inclined – but as yet unable – to disagree with much of what he says about the latter. However, alongside his many acute criticisms on which I'll need to reflect further, Jones's report of two key moves in my argument requires some revision and may, therefore, warrant this brief early response.

The introduction of a ‘one-party participatory democracy’ in Zambia in 1973 under the United National Independence Party (UNIP) of President Kenneth Kaunda made significant changes in the nature and extent of political participation, regime structure and public policy in that country. Among a number of constitutional changes, the proscription of the opposition parties – African National Congress (ANC) and United Progressive Party (UPP) – was probably the most important. There is a relatively extensive literature describing these changes and evaluating their significance. A number of further changes which affected these political variables in varying degrees occurred during the life of the one-party system, which lasted until 1991, but much less has been written about these changes, at least in part because they have been assumed to be insignificant. This Note describes the collection of a systematic events dataset on changes in electoral and political party participation (including changes in policies towards participation and changes in party structures affecting participation), regime structure (including party–government relations, central government structure and central–local government relations), and policies affecting the economy, class structure and culture in Zambia from 1973 through 1985. It then describes the use of expert judges to scale events in the dataset and evaluate their cumulative significance for dimensions of change delineated by the investigator or themselves. Finally, it presents one substantive application of this methodology: specification of the overall directions and extent of change in electoral and party participation under the one-party system. Two contradictory directions of change not so far identified in the literature on the Zambian one-party state are uncovered. It is suggested that changes in the one-party state helped to undermine its support, even among some of those Zambians who initially believed in it.

Peter Morriss (see pp. 595–7) criticizes most applications of mathematically derived power indices, including mine and the response to my original Note. He uses uncomplimentary terms, implying that we are not only wrong but also fools for being wrong.

Peter Morriss's attack on R. J. Johnston's Note takes some side-swipes at us. We are allegedly ‘not quite up to par’, because (i) we are said to advocate the Banzhaf index; (ii) we are accused of ignoring a simpler route to our conclusion.