¹ March, J. and Olsen, J., Rediscovering Institutions (New York: Free Press, 1989)Google Scholar; and March, J. and Olsen, J., ‘The New Institutionalism’, American Political Science Review, 78 (1984), 734–49.CrossRefGoogle Scholar

² Douglas, M., How Institutions Think (London: Routledge and Kegan Paul, 1987).Google Scholar

³ The longest and in many ways one of the most helpful treatments of ‘The Prime Minister and Parliament’ is given by Wilson, Harold, The Governance of Britain (London: Weidenfeld and Nicholson, 1976), chap. 7.Google Scholar

⁴ Griffith, J. A. G., Parliamentary Scrutiny of Government Bills (London: Allen & Unwin, 1974).Google Scholar

⁵ Norton, P., Dissension in the House of Commons: Intra-Party Dissent in the House of Commons Divisions Lobbies, 1945–74 (London: Macmillan, 1975)CrossRefGoogle Scholar; Norton, P., Dissension in the House of Commons, 1974–79 (Oxford: Clarendon Press, 1980)Google Scholar; Finer, S. E., Berrington, H. and Bartholomew, J. B., Backbench Opinion in the House of Commons, 1955–9 (London: Oxford University Press, 1961).Google Scholar

⁶ Mackintosh, J. P., The British Cabinet, 2nd edn (London: Faber, 1968), pp. 567–97Google Scholar, does provide an extended discussion of parliamentary influences on cabinet government in an earlier period. By contrast, Hennessy, P., Cabinet (Oxford: Blackwell, 1986), p. 3Google Scholar, stresses their importance but barely mentions them thereafter.

⁷ Dunleavy, P. J., Jones, G. W. and O'Leary, B., ‘Prime Ministers and the Commons: Patterns of Behaviour, 1868 to 1987’, Public Administration, 68 (1990), 123–40.CrossRefGoogle Scholar

⁸ The technique used here is ‘median smoothing’, which means that for each session we have entered not the actual adjusted score for that year, but instead the median score of that year, the previous year and the subsequent year. This means that only changes which lasted for at least two sessions will show up in Figure 1: single-session ‘blips’ are eliminated. For example, in the short 1908 session of forty-nine days Prime Minister Campbell-Bannerman became ill and answered questions on only five days, giving an adjusted score of fourteen days. However, since his adjusted score in the previous session was seventy-eight days, and that for his successor Asquith was eighty-six days, the median smoothed score for the 1908 session is still seventy-eight days.

Because of the frequent replacement of one prime minister by another, all the smoothed data included here need to be treated carefully, for example, they cannot be used for recalculating information. Smoothing also tends to ‘displace’ some kinds of changes from one session to another in a very few instances, despite the use of three iterations recommended by Tukey, J., Exploratory Data Analysis (Reading, Mass.: Addison-Wesley, 1977)Google Scholar. The smoothed graphs have a jagged edge appearance because they have not been ‘hanned’: with the frequent succession of premiers hanning would be severely misleading. For simpler treatments of median smoothing, see also Marsh, C., Data Analysis for Social Scientists (Cambridge: Polity, 1988)Google ScholarPubMed and Erickson, B. H. and Nozanchuk, T. A., Understanding Data (Milton Keynes: Open University Press, 1977).Google Scholar

⁹ Prime ministers' questions were initially listed for a short time at number 51 on the order paper. The fifth and short parliamentary day, Friday, is assigned to private members' business.

¹⁰ Rowland, P., Lloyd George (London: Barrie and Jenkins, 1975), p. 379.Google Scholar

¹¹ Riddell, Lord, Lord Riddell's Intimate Diaries of the Peace Conference and After (London: Gollancz, 1933), pp. 240–1.Google Scholar

¹² Rowland, , Lloyd George, p. 379Google Scholar. See also Rhodes-James, R., Memoirs of a Conservative: J. C. C. Davidson's Memoirs and Papers (London: Weidenfeld and Nicholson, 1969), p. 49Google Scholar; Owen, F., Tempestuous Journey: Lloyd George – His Life and Times (London: Hutchinson, 1954), p. 404Google Scholar; Kinnear, M., The Fall of Lloyd George: The Political Crisis of 1922 (London: Macmillan, 1973)CrossRefGoogle Scholar; Jones, T., Lloyd George (London: Oxford University Press, 1951), p. 195CrossRefGoogle Scholar. His son attributed the prime minister's absences to his womanizing: George, R. Lloyd, Lloyd George (London: Frederick Mueller, 1960), p. 208.Google Scholar

¹³ Harris, K., Attlee (London: Weidenfeld and Nicholson, 1982), p. 180.Google Scholar

¹⁴ Harris, , Attlee, pp. 180, 192Google Scholar. Churchill at first offered the role to Chamberlain, but Labour MPs objected, so formally the prime minister retained the title. Attlee was for a time described as Deputy Leader of the House of Commons, but in fact carried out all the tasks involved. Cripps became Leader of the House in 1942.

¹⁵ Wilson, , Governance of Britain, p. 145.Google Scholar

¹⁶ Attlee was Minister for Defence as well as prime minister in 1945–46, when he answered questions on seventy-seven and sixty-four days respectively (on an adjusted basis). Churchill also took the Defence portfolio on gaining power in 1951 until the end of the Korean war, and also answered questions on seventy days per session at first.

¹⁷ ‘Every Opposition suspected that, even if ministers were not filibustering to ensure that Prime Minister's Questions were not reached, government backbenchers were rising in such numbers as to fill the time’: Wilson, , Governance of Britain, p. 132.Google Scholar

¹⁸ The exceptions outside this range are very minor. Harold Wilson answered questions on only forty-one days in the 1974–75 session, but then overcompensated by scoring fifty-six days in the 1975–76 session leading up to his resignation. Margaret Thatcher achieved scores below forty-seven days in three sessions, 1984–85, 1986–87 and 1989–90.

¹⁹ Irwin, H., ‘Opportunities for Backbenchers’, in Ryle, M. and Richards, P. G., eds, The Commons Under Scrutiny (London: Routledge, 1988), p. 81.Google Scholar

²⁰ Dalyell, T., Misrule: How Mrs Thatcher Has Misled Parliament from the Sinking of the Belgrano to the Wright Affair (London: Hamilton, 1987), p. xiii.Google Scholar

²¹ Griffith, J. A. G. and Ryle, M., Parliament (London: Sweet & Maxwell, 1989), p. 260.Google Scholar

²² Bromhead, P., Britain's Developing Constitution (London: Allen & Unwin, 1974), p. 141.Google Scholar

²³ Sedgemore, B., The Secret Constitution (London: Hodder & Stoughton, 1980), pp. 190–1.Google Scholar

²⁴ Griffith, and Ryle, , Parliament, p. 177Google Scholar. The core procedural rule of the Commons is that the House is always considering a specific motion, or an amendment to a motion, on which speeches are made and a vote can be taken, for or against. When ministers make a statement, there is no such motion.

²⁵ The prime minister is also directly responsible for the security and intelligence services, where scandals have involved a few statements since the 1960s, and for the civil service, where major reforms can entail occasional statements.

²⁶ Griffith, and Ryle, , Parliament, pp. 334–6Google Scholar. See also, Irwin, , ‘Opportunities for Backbenchers’, pp. 77–82.Google Scholar

²⁷ Thatcher made some use of written answers to ‘planted’ questions to lend her personal weight to smaller-scale domestic policy initiatives without making statements.

²⁸ Quoted in Egremont, M., Balfour: A Life of Arthur James Balfour (London: Collins, 1980), p. 189.Google Scholar

²⁹ In the same period, however, Campbell-Bannerman answered questions on 182 days and intervened in debates on eighty-two days (actual scores). His reluctance as an orator was partly founded on his indifferent performance as opposition leader to Balfour and partly on his dislike of the Commons following his ‘ordeal’ in becoming Liberal leader. See: Spender, J. A., The Life of Sir Henry Campbell-Bannerman: Volume II (London: Hodder & Stoughton, 1923), pp. 52, 356.Google Scholar

³⁰ These were: 1911–12 under Asquith and 1916–18 under Lloyd George.

³¹ Marquand, D., Ramsay MacDonald (London: Cape, 1977), p. 402Google Scholar. See also, Wertheimer, E., Portrait of the Labour Party (London: Putnam, 1929), pp. 174–5.Google Scholar

³² Marquand, , Ramsay MacDonald, p. 698Google Scholar; see also pp. 602, 672, 680, 698–9.

³³ Dunleavy, P. J. and Husbands, C. T., British Democracy at the Crossroads: Voting and Party Competition in the 1980s (London: Allen & Unwin, 1985), p. 86.Google Scholar

³⁴ Thatcher, M., quoted in the Daily Mail, 4 05 1989, pp. 22–3.Google Scholar

³⁵ See Birch, A. H., Representative and Responsible Government (London: Allen & Unwin, 1964), pp. 131–70Google Scholar, where the parliamentary attendance of ministers is an unquestioned presumption underlying the discussion of collective and ministerial responsibility.

³⁶ In Korea the prime minister speaks in the legislature but is not an elected member of it. In France, the prime minister accounts to, but cannot sit in, parliament.

³⁷ Many cell entries in Table 1 are necessarily judgemental since systematic data is unavailable, but we have used explicit criteria for our estimates. Formalization is defined as ‘high’ when the activities involved are subject to written rules formulated in public documents; as ‘medium’ if practices are governed by strong conventions; and otherwise as ‘low’. Publicness is defined as ‘high’ if an activity takes place in a public arena directly monitored by the mass media; as ‘medium’ if the context can be indirectly monitored by the media with considerable reliability; as ‘low’ if the context is a private one; and as ‘very low’ if private activity is also protected by strong secrecy norms. Parliamentary focus is defined as ‘high’ when an activity occurs in the Commons or concerns the reactions of MPs generally; as ‘medium’ when the activity involves or concerns many MPs but is limited to those of one party; and as ‘low’ when the activity involves few MPs or concerns other reactions. Frequency is defined as ‘high’ when the activity occurs many times in a parliamentary session; as ‘medium’ when it occurs several times; as ‘low’ when it occurs once or twice per session or less; and as ‘very low’ if it occurs only once in several sessions.

³⁸ As cabinet secretary Sir Robin Butler agreed with Margaret Thatcher that he would not accompany her on overseas visits, so that he could use the time involved to carry out his other role as head of the home civil service. While his predecessor, Sir Robert Armstrong, split his time between these duties in a 75/25 per cent ratio, Butler's time allocation is more like 50/50, the increase graphically demonstrating the contemporary importance of overseas visits in the prime minister's timetable.

³⁹ Donoghue, B., Prime Minister: The Conduct of Policy Under Harold Wilson and James Callaghan (London: Cape, 1987)Google Scholar; Blackstone, T. and Plowden, W., Inside the Think Tank: Advising the Cabinet, 1971–1983 (London: Heinemann, 1988).Google Scholar

⁴⁰ Janis, J., Victims of Groupthink (Boston, Mass.: Houghton Mifflin, 1972).Google Scholar

⁴¹ See Hennessy, P., Whitehall (London: Fontana, 1990), pp. 384–5 and 364–7.Google Scholar

⁴² Cockerill, M., Hennessy, P. and Walker, D., Sources Close to the Prime Minister: Inside the Hidden World of the News Manipulators (London: Macmillan 1985)Google Scholar; Margach, J., The Abuse of Power: The War Between Downing Street and the Media from Lloyd George to James Callaghan (London: W. H. Allen, 1978)Google Scholar; May, A. and Rowan, K., Inside Information: British Government and the Media (London: Constable, 1982).Google Scholar

⁴³ Linklater, M. and Leigh, D., Not With Honour: The Inside Story of the Westland Scandal (London: Sphere Books, 1986)Google Scholar; Dunleavy, P., ‘Reinterpreting the Westland Affair: Theories of the State and Core Executive Decision-Making’, Public Administration, 68 (1990), 29–60.CrossRefGoogle Scholar

⁴⁴ These covered Lord Beaconsfield (1876–80), Lord Salisbury (1885–86,1886–92 and 1895–1902) and Lord Roseberry (1894–95).

⁴⁵ de Smith, S. A., Constitutional and Administrative Law, 3rd edn (Harmondsworth: Penguin, 1977), pp. 226–7.Google Scholar