The availability of detailed census data at the level of parliamentary constituencies has promoted the testing of ecological models of voting behaviour in the United Kingdom. To date, most of the testing has used the 1966 Census data, which is inferior to that for 1971 in certain respects. As a result, it has not been possible to test in detail two models – that of ‘embourgeoisement’ and that of the ‘property-owning-democracy’. But the cross-classifications available at constituency level for the 1971 Census allow the relative importance of the two to be assessed.

The first elections to the European Parliament were held on 10 June 1979 with each EEC country following its own electoral system. As in all recent elections, on-the-night forecasting was an integral part of the television coverage in Britain. This paper describes the method used by the BBC to predict the composition of the British delegation to the European Parliament. The performance of the method is discussed and some tentative conclusions are drawn about the behaviour of the British electorate.

In the section of their book Political Change in Britain entitled ‘The Sources of Uniform Swing’, David Butler and Donald Stokes presented a model to account for an apparent paradox in Britain's electoral geography – the relative uniformity of the ‘two party swing’ between elections across all parts of the country. This uniformity is paradoxical because if the national trend in changes of partisan preferences were to apply equally in all areas then the swing against a party experiencing a net loss of votes should be greatest in those areas where it won most votes at the first election: indeed, as Butler and Stokes show, it may even register a net gain of votes where it was weakest, despite the national trend being against it.

In ‘Arms Races: External Security or Domestic Pressure?’ (this Journal, x, 121–8), Willem Saris and Cees Middendorp advance a compelling argument for the preponderance of domestic forces in driving arms races. However, the weight of the argument must be borne by the well-reasoned explanation of industrial forces and saturation levels, since the empirical analysis is fundamentally flawed in three ways. Definitions are lacking, data sources are ambiguous, and the application and interpretation of statistical procedures are inappropriate.

When the results of an analysis fly in the face both of generally accepted theory and the practice of professional politicians, it is usually a good idea to reconsider whether the point is really established. Such is the case with the small literature regarding the effects of nomination strategies on the outcomes of Dáil Éireann elections that has developed in these pages since the publication of an article by Cohan, McKinlay and Mughan (‘The Used Vote and Electoral Outcomes: The Irish General Election of 1973’, this Journal, v (1975), 363–83). The original article made several valuable points, including a demonstration that patterns of transfer votes may be more important than the distribution of first preference votes in determining the result of elections under the single transferable vote (STV) system, and an extension of the wasted vote concept to STV, majority, and proportional elections. Unfortunately, the notes that have followed have all focused on the most questionable conclusion of the original article, that overnomination hurts a party's chances of electing the maximum possible number of deputies.

Although we appreciate the attention the critic has given to our paper, we are somewhat disappointed about the kind of criticism. It is said that the ‘empirical analysis is fundamentally flawed’. But if the analysis is flawed it must be very easy to show it by a reanalysis of the data. However, if one takes the time to look at the data used in this study one can see immediately that when the USSR's level of armaments is very low the USA is producing large amounts of missiles. On the other hand, when the USSR has a large number of missiles the USA's production is nil or very little. Consequently one must conclude that the USA cannot possibly be reacting to the activities of the USSR in the simple ways suggested by Richardson or Hamblin et al. This result was confirmed by our statistical analysis of the data. One can of course try other statistical procedures, as we did, but they all produce the same result: there is no reaction effect in the USA's behaviour.

Professor Katz raises a number of points, but I shall limit myself to four. Firstly, I accept his argument that the way in which parties nominate candidates, even over-nomination, may be functional for party unity. Indeed, nowhere in the previous articles has anything else been suggested.

In the July 1978 issue of this journal McKinlay and Little have written an interesting interpretation of British aid policy from 1960 to 1970, based on a statistical study of aid allocation during that period (viii (1978), 313–31). The purpose of the following Note is to question the adequacy of the support for their conclusion that the foreign policy objectives of aid have been dominant compared to humanitarian motivation. It is not intended to deny the existence of foreign policy objectives in aid giving. Aid policy in the United Kingdom, like most other policies, is not made by one ministry but is a compromise between ministries. The question at issue is the relative weight of different motives in aid allocation. There is no doubt that the authors' study has revealed clearly some interesting statistical associations and non-associations between aid allocation to different countries and, inter alia, their size, degree of poverty, colonial, trading and security links with the United Kingdom, Communist bloc associations, and various measures of their political stability and democratic character. However, the main question which must be raised is whether the statistical evidence amassed and analysed really permits the authors to draw their firm conclusions about the relative weight of different British objectives during this period.