Two referendum questions to change the Constitution were put to eligible Australian voters on 6 November 1999. The two questions were formally described in the official Australian Electoral Commission booklet as follows:

1. Constitution Alteration (Establishment of Republic) 1999-11-21
o to alter the Constitution to establish the Commonwealth of Australia as a republic with the Queen and Governor-General being replaced by a President appointed by a two-thirds majority of the members of the Commonwealth Parliament.
2. Constitution Alteration (Preamble) 1999
o to alter the Constitution to insert a preamble (AEC 1999,3).

The failure of the referendum questions - on the establishment of the republic and on the insertion of a Constitutional Preamble - was widely predicted prior to the event. Opinion polls had shown a steady decline in the support base for the republic over the preceding weeks. The Australian's Newspoll conducted on 22-24 October, for example, revealed the 'No' case leading comfortably and support for the 'Yes' case falling over the previous month. When coupled with Australia's poor record on producing affirmative referendum outcomes, such polling led many to consider a negative outcome to be inevitable. The result was, however, even worse than most had predicted. Majorities in every state rejected both questions, as had happened nine times before in referendums. The national count of just over 45% in favour put the republic question in among the lowest third of all referendum results. The national vote for the Preamble, at just over 42% was the tenth worst, out of 44 questions this century.

What many found particularly surprising (and some rather galling) was the apparently high level of support for the republican question in Liberal-held seats, most notably in the Prime Minister's own electorate of Bennelong, where close to 55% voted 'Yes'. In New South Wales, the supposed 'heartland' of the Australian Republican Movement, eight Liberal-held and seven Labor-held seats recorded a 'yes' vote. Over the whole country, however, (as the figures stood, in the week after the referendum) only seventeen Liberal-held seats supported the republic, compared to twenty-four Labor-held seats. The highest level of support was in the Labor-held inner-city electorates of Melbourne (with 71%) and Sydney (with 68%). But the party/electorate pattern was very uneven. In South Australia, only three electorates, all Liberal-hel voted 'Yes'. The single electorate to support the republic in Western Australia (Curtin) was Liberal-held, and the same applied in Queensland (Ryan). But in Tasmania, the single favourable seat (Denison) was Labor-held, and the two ACT electorates, both Labor-held, each registered over 60% support.

The Preamble question had a pattern of its own. The electorate of Melbourne supported it at just over 50%, while Sydney recorded less than 49%. The Sydney result contrasted with five other New South Wales electorates, including Bennelong, where a 'Yes' vote was recorded for both the republic and the Preamble. Nine Victorian electorates (six Labor -held and three, including Kooyong, Liberal-held) supported both questions, as did six New South Wales electorates. But not one electorate in any of the other States or in the ACT supported both questions. Significantly, although not surprisingly, not a single National Party-held electorate anywhere supported either the republic or the Preamble. No electorate anywhere voted for the Preamble while rejecting the republic, although twenty-six electorates in all voted the other way round.

These results will generate much investigation, both statistical and analytical, in the months, indeed years, to come. In the referendum's immediate aftermath, a great deal of speculative analysis circulated. The consensus appears to have narrowed in on three (not incompatible) explanations for the defeat. One hypothesis is that referendums have no chance of succeeding if they do not have wide, cross-party support and particularly the support of the Prime Minister. A second hypothesis holds that this particular referendum failed because it represented the aspirations of an 'elite' and it alienated the 'people', especially the 'battlers'. A third points at insufficient voter knowledge, with the apparent correlation between higher levels of formal education and a greater propensity to support the republic, and the complete failure of both questions in rural electorates, held up in demonstration of this explanation.

From what we know about referendum results this century, the first hypothesis is probably right, although it tells us, in fact, relatively little. No referendum indeed has ever succeeded without the support of the Prime Minister, but no referendum has succeeded without bipartisan support, and this has always included the Prime Minister. But bipartisan support is, as is widely known, not sufficient; at least three referendums have failed this century, despite having bipartisan support. In addition, the Prime Minister's support for the Preamble was in itself insufficient to ensure success (except, perhaps, in his own electorate).

The second hypothesis seems intuitively correct, although its conclusion cannot be separated from questions about the campaign itself. The 'No' campaign, indeed, ran very strongly along populist 'anti-elite' lines, with, however, a confusion between several versions of 'elites': there were the alleged 'elites' ('Chardonnay drinkers') at the heart of the republican movement, those classed as 'elites' merely by being residents of Sydney, Melbourne and Canberra, and and another version of 'elites' meaning simply Federal politicians. Furthermore, it is unclear whether the 'battlers' wanted no change to the Constitution, or opposed the model on offer because they preferred a direct election method of choosing the head of state.

The third (related) hypothesis, concerning levels of education, has greater credibility. A positive correlation between formal education and support for a republic has existed over a number of years, as Murray Goot (1994) has demonstrated. But the value of such a finding remains to be explored, since it does not in itself reveal why people with lower levels of education reject constitutional change of the type proposed in the 1999 referendum. And it leaves unexplained the Preamble result, with, for example, both ACT electorates, where high concentrations of tertiary educated voters are to be found, rejecting the Preamble. The unevenness of voting around the country is still to be explained, and only a detailed analysis of both the campaign and the role of individual State politicians, will begin to throw light on this. We do not know at this stage whether the better result achieved for the republic in Victoria was generated by the climate created in the aftermath of the Kennett government's defeat, or whether the two were entirely independent. But, from the perspective of the pattern of referendum results this century, the campaign would appear to be the most likely single explanation.

Although Australia very nearly reached the end of the 1990s without a referendum, there have now been referendums in every single decade this century. The 1910s, with twelve separate questions, and the 1970s, with ten questions, have been the peak. The first decade of the century and the 1950s had only one question each. The 1990s now join the 1930s, with two referendum questions each. Australia has now had the experience of forty-four questions, only eight of which have succeeded. Although numerous explanations have been advanced in an attempt to explain this pattern (with some analysts, such as Brian Galligan (1995) questioning whether an explanation is necessary), there is no consensus. Some have suggested that Australians, being good federalists, reject attempts to increase Commonwealth power. However, two of the eight successful referendums (on social welfare powers, section 51 (xxiiiA), and 'the races' power, section 51 (xxvi)) expressly had that effect, and another two (sections 105 and 105A) (taking over states debts and establishing the Loan Council) ultimately did. Others have suggested that, as a corollary to the first explanation, 'technical' alterations to the Constitution alone are likely to succeed. While the first success, in 1906 (changing the date under section 13 at which newly elected Senators took their seats) was purely technical, and the 1977 alteration (setting a retirement age of 70 for Federal judges under section 72), may arguably be classified as such, the successful alteration to require casual Senate vacancies to be filled by a nominee of the outgoing Senator's party (under section 15), could scarcely be considered 'technical', nor could the successful removal of section 127 which had excluded aboriginal people from the Commonwealth census count. The double majority requirement in section 128 accounts for only five out of the 36 defeats, in the sense that only on those five occasions did the proposed alteration receive more than 50 per cent nationally, only to lose because it failed to win majorities in a majority of States as well. Of these five, two - on simultaneous elections for the Houses of Parliament - were the same question put at different times. Three of the five were split three to three on the state count. Three of the five were split 3-3 and the other two lost 2-4 in the State count.

A number of these defeated referendums have followed formal attempts to review the Constitution, either extensively or in its entirety: a Royal Commission appointed in 1927; a Constitutional Convention in 1942; a Joint Parliamentary Committee in 1956; a Constitutional Convention (which sat on average every two years) between 1973 and 1985; and a Constitutional Commission from 1985 to 1988. The Republic Advisory Committee, appointed by the Keating Government in 1993, also conducted a far-ranging inquiry into the Constitution, and - albeit indirectly - was responsible for the establishment of the Constitutional Convention which met in early 1998. Although the 1998 Convention appeared at the time much more hopeful, partly because half its members had been directly elected and this appeared to lend it a legitimacy the previous 'Conventions' (other than that of 1897-1898) lacked, and partly because of the high level of publicity it attracted, it now joins all the others in having no direct issue.

One of the problems a future attempt to achieve a republic will face is that the number of necessary alterations to sections of the Constitution is substantial even under the 'minimalist' model where the effect of the change is intended to be small. The 1998 Convention did little to address the range of alterations needed, with its ten days'duration (by far the shortest of any of the review processes) devoted principally to debating the means of choosing a republican head of state. Ironically, this matter need not depend on constitutional alteration. It would be quite possible constitutionally to have a parliamentary choice, even a direct popular election, for the Governor-General, leaving the Constitution undisturbed, with the name of the chosen candidate going forth as the Prime Minister's nominee to the Queen, under the current practice applied to section 2. The means of selection is really a matter of convention, and of the manner in which the largely-undefined role of the Governor-General is interpreted by the incumbent. This principle was, in fact, strenuously tested in 1930 when the Scullin government nominated an Australian, Isaac Isaacs, as Governor-General. Opponents maintained that the position would be compromised because a local was bound to be caught up in party politics and could not be appropriately detached and impartial. Their predictions of suste, collapse (which had a very similar character to arguments used today by opponents of the direct-election model for the identification of head of state) were confounded because Isaacs performed the role of Governor-General just as his predecessors had done. What is really at issue, and what the Convention avoided addressing, are the current powers of the Governor-General, and whether these should remain uncodified regardless of how this position is filled.

Other constitutional alterations which followed from the proposed change of wording required to accomplish a republic also needed to be discussed. Some were picked up by the Senate Committee which looked into the republican alteration bill prior to its passage through the Parliament. The proposed alteration to section 117 - which currently prohibits discrimination against a 'subject of the Queen' on the grounds of out-of-State residence - raised concerns, because the replacement wording was to be 'citizen', but this would have had the effect of limiting the application of this section to only those with legal citizenship. The Senate Committee recommended an amendment to the proposed wording to avoid this result, but the recommendation was not adopted in the final bill. It is arguable that the intention behind this section would have been departed from significantly and that the view (held by a High Court minority in, for example, the 1992 case of Leeth v the Commonwealth ) of the section as providing a guarantee of equality within the Commonwealth would no longer be available.

Only a reconstituted, lengthy Convention where real constitutional work (rather than political negotiation) is done could tackle such issues. Such a Convention might even take the opportunity to 'clean up' the Constitution, excising the many spent, transitional provisions, such as those found in Chapter IV which tell us what is to happen before the 'imposition of uniform duties of customs', which itself was to occur within two years of the inauguration of the Commonwealth. But the likelihood of a referendum success for an effectively new Constitution, would appear low. On the other hand, it may emerge from analysis of the 1999 referendum campaign that the fear of appearing grand and 'visionary', and the resulting decision to promote the minimalist model, itself contributed significantly to failure. Why change a Constitution when there appears to be no compelling reason?

In my view, the best single explanation for the success of the original referendums between 1898 and 1900, which adopted the Constitution, was the idealism and inspiration which built up around the goal of federation. This too may explain the unprecedented success of the 1967 referendum on Commonwealth powers to make special laws for the Aboriginal people. The fact that the actual Constitutional changes involved were widely misunderstood, both at the time and since, only reinforces this interpretation. Recent debates about the scope for a negative application of the Commonwealth's power over Aboriginal affairs acquired in 1967 (not to mention the question of whether it has enabled the States to pass on to the Commonwelth their responsibility to provide services to Aboriginal communities) have not diminished the almost mythological status of the 1967 referendum result.

Republicans will need to ask themselves now how ardently they desire a republic and just how many processes they are prepared to endure along the way. The alternative is to wait until Britain itself takes the constitutional steps that will render the relevant sections of Australia's Constitution either redundant or inapplicable. A referendum in the wake of such a move would most likely succeed.

AEC [Australian Electoral Commission] 1999, Yes/No Referendum '99: Your Official Referendum Pamplet, Canberra.
Galligan, B (1995),A Federal Republic , Cambridge University Press.
Goot, M. (1994), 'Contingent Inevitability: Reflections on the prognosis for republicanism', in George Winterton, We, the People, Allen & Unwin, Sydney.
Kelly, P (1999), 'Republic Running Distant Second', The Australian.