Government action presupposes extraction. Historically, four different sources of public revenue may be distinguished. Tributes, which are common in peimitive and belligerent states, are variable levies exacted at irregular intervals. Tariffs raised on the basis of the physical control of strategic passageways are important sources of income for mercantile states in the early stage of their development. Taxes play some role in semi modern economies but evolve more fully as the financial mainstay of government in advanced industrial societies. In addition, many states draw large receipts from the profits of trade. The planned economies in the socialist bloc are trading states rather than tax states. The different logistic properties of these revenue sources affect the prospects for government growth at various stages of economic development. The expansion of the tax state is strongly linked to the proliferation of monetary flows in modern society. At the same time, some non tax receipts such as user charges and other incomes from trade retain their importance, and the present ‘crisis of the tax state’ might possibly lead to a more composite structure of government revenue.

Public-choice models argue that large interest groups are less likely to overcome free-rider problems because of the irrelevance of individual's participation to the supply of non-excludable group benefits. But these accounts are constructed in terms of ‘objective’ variables, and hence rely on perfect information assumptions. Paying attention instead to how people learn that interest groups are relevant for them indicates a key role for group identities, i.e. subjective perceptions of interests shared with others. Recasting the decision to form or join groups in terms of subjective variables highlights the imporlance of perceived group viability. In a liberal democratic context, increasing group size has ambiguous effects; it somewhat accentuates the irrelevance of individual participation to supply, and yet (ceteris paribus) also increases (he group's viability. Applying the group identity approach sheds light on a problem which public-choice theory cannot adequately explain: the reasons why (apart from group size) social interests are differentially difficult or easy to organize.

This article outlines a cognitive-affective model of the role of social groups in political thinking. The model is based on the assumptions that people have stored information and emotional reactions to social groups, and that people are purposive in their thinking about social groups in the sense that they are interested in understanding what various groups have obtained and whether it is deserved. The process through which social groups influence political thinking varies significantly depending upon whether an individual identifies with the group in question. Generally, people are more inclined to feel sympathetic towards the groups to which they belong. These ideas are illustrated with an empirical analysis that focuses on women's issues and makes use of data collected in the 1984 National Election Study Pilot Study.

With the decline in popular attachment to the two major parties in the United States since the mid-1960s, collective political independence has risen. Using new survey questions introduced in 1980, this article employs alternative measures of independence to reassess the phenomenon of independence in America. These new measures give us fresh insights beyond what we had using only the traditional measures. One casualty of this new approach is the portrait of the Independent given by The American Voter. This portrait appears seriously misleading, given that it is those who deny being either partisan or Independent who fit that portrait – not Independents per se. And the most politically involved voters turn out to be Independent Partisan Supporters; not simple partisans.

By the standard of most European parliaments, levels of party voting in the United States Congress are relatively low. Nevertheless, party voting does occur in the House of Representatives and the Senate. In the American context, a party vote occurs when majorities of the two congressional parties, the Democrats and the Republicans, oppose one another. The authors construct measurements of levels of party voting in Congress in the years after the Second World War. They then develop a model to test the effects of a number of independent variables that influence fluctuations in party voting levels over time. The study models the time series for party voting and demonstrates striking differences between the House and Senate in the correlates of partisan cleavage.

I have argued two points. Firstly, Skocpol has confused peasant autonomy and peasantstate alliances. The relationship between autonomy and revolt is spurious. Secondly, peasant solidarity is neither a necessary nor a sufficient cause of peasant revolt. It is not necessary because revolt concentrated in a specific niche in a peasant community cannot be attributed to peasant solidarity. Nor is solidarity sufficient, because it has effects only in communities with relatively high skill levels. When solidarity does have independent social effects, it is as a community norm and tends to be associated with a sort of revolt that is not intended to change the lord—peasant relationship.

This Comment proposes a new procedure for selecting a single winner in a multicandidate election. This procedure may be viewed as an offspeing of two other procedures which have been recently analysed in this journal.

In a recent Note in this Journal by Gallagher and Unwin, it was stated that the element of randomness due to sampling surplus votes in single transferable vote (STV) elections ‘has long been recognized, but no previous attempt has been made to assess its impact’. This is incorrect. Work done in Australia (and reported in the leading Australian journals in their respective fields) has comprehensively dealt with this problem both in theory and practice. Since STV is practised in national elections in only three countries (Australia, Ireland and Malta, the most populous being Australia) it is surprising that contributors to, and referees of, this Journal should be unaware of such material. It was largely because of the evidence contained in these articles that the Joint Committee on Electoral Reform recommended to the Australian Parliament an amendment to the Electoral Act, to avoid the problems caused by sampling votes, by using the Gregory method of counting them. The Act was amended in this way in 1983.

We are grateful to Fischer for pointing out his work on STV random sampling, which precedes ours. While acknowledging our oversight in not referring to his articles, we cannot accept the validity of either of the substantive criticisms he makes in his Comment.