Prepared by:
Mollie Dunsmuir
Law and Government Division
January 1992




   A. Authority for the Referendum
      1. Constitutional
      2. A Referendum Statute
      3. A Specific Statute

   B. Originating a Referendum
      1. Executive Branch
      2. Legislature or Political Party
      3. People or Local Government

   C. Wording the Question

   D. Effect of a Referendum
      1. Referendums and Plebiscites
      2. Binding Referendums
      3. Abrogative Referendums
      4. Plebiscites or Consultative Referendums

   E. Subject of Referendum
      1. Constitutional Amendment
      2. Territorial or Sovereignty
      3. Moral or Political Issues
      4. Bill C-9, the Canada Referendum Act (1978)

   F. Regulating the Referendum Process
      1. Grouping the Interests
      2. Financing and Media Access

   G. Counting the Vote





Referendums have a long and varied history. In modern Western democracies, they have been used in some Swiss cantons since the sixteenth century and in various states of the United States of America since the eighteenth. Switzerland is the only nation to have held a large number of nationwide referendums, however, accounting for over half of all national referendums ever held in the democratic countries. Indeed, statistical summaries of referendums often omit Switzerland because including that country skews the results. Referendums are both more and less widespread than is often thought. Almost every Western democratic country, the United States being the most notable exception, has used national referendums at least once; however, as of 1978, only four had held as many as 10: France (20), Denmark (14), Australia (39), and Switzerland (297 since 1848, when Switzerland became a federation).(1)

Discussions of referendums tend to vary in terminology and approach. There is, for example, no definitive breakdown of the various types of referendum. The second part of this paper (Context), therefore, discusses various aspects (who initiates the referendum, the effect of the referendum, the purpose of the referendum, etc.), and gives examples of how various countries deal with these.(2)

Referendums or plebiscites can cover a wide range of topics, and have greatly differing effects. In Canada, referendums have been used primarily at the provincial or municipal level to deal with specific statutory issues, such as the prohibition of liquor sales or daylight saving times.(3) This paper, however, does not deal with provincial, municipal or state experiences, but only with referendums at the national level. The one exception, understandably, is the Quebec referendum of 1980, which, although technically a provincial plebiscite, was modelled on national referendums elsewhere.

Throughout the paper, there are references to Canada’s two national referendums (on prohibition, 1898, and on conscription, 1942), to the Confederation referendums in Newfoundland in 1948, and to the various referendum criteria proposed from 1978 to 1980: in the Quebec Referendum Act used for the 1980 referendum on sovereignty-association; in the proposed resolution respecting the Constitution of Canada, 1980; and in Bill C-9, 1978, the proposed Canada Referendum Act.

Finally, it is important to remember that the experience of unitary countries with referendums may be of limited use to a federal state such as Canada. Although many American states, particularly California, are experienced in using referendum procedures, a referendum has never been held at the national level in the United States:

[Referendums] are unconstitutional at a national level in American politics. Congress cannot make a decision by referendum; it is clear that the courts will say that Congress could not put to a referendum a binding question of national policy. Second, it is inherently inconsistent in the federal system; that does not say it is not convenient, it is merely inconsistent. The full arrangement of the American political system at the federal level is a delicate balance between population and geography; it is inherently unrepresentative in many of its aspects.(4)

In both of Canada’s national referendums, the geographic or federal aspect has been a major issue. In the prohibition referendum of 1898, only 44% of the populace bothered to vote, and prohibition was approved by a margin of 51% to 49%. Quebec, however, rejected prohibition decisively, by 122,614 to 25,582. In the event, the federal government seems largely to have ignored the results of the referendum, leaving the matter to the provinces instead. Nonetheless, the incident indicates the inequity that could come about if the results of a referendum approved by less than one-quarter of the eligible electorate were imposed on a province that had rejected it by a measure of almost 5 to 1.

In 1942, the problem was illustrated in even starker terms. Sixty-four per cent of eligible Canadians voted to release the Mackenzie King government from its pledge not to use conscription for overseas service. However, 72% of Quebeckers voted "No." Since many in Quebec believed that the promise had been made specifically to their province rather than to the country as a whole, the results exacerbated, rather than mitigated, the divisive nature of the conscription issue.

Aside from issues of jurisdiction and geographic fairness, the dynamics of a referendum in a federal state are not always helpful:

In Australia, a federal constitution presents jurisdictional problems and her referendums have ranged from the regulation of airlines to the outlawing of the communist party. Almost all the referendums have involved an extension of federal power and 78 per cent of the propositions have been defeated, largely through the whipping up of provincial loyalties.(5)

The reasons given in favour of referendums vary widely, and depend to a considerable degree on the type of referendum involved. One of the most common type relates to constitutional change, which many argue should be legitimized by popular consent. In other cases, especially those involving issues of territoriality, sovereignty or government structure, the government may decide that a popular vote is required to confer the necessary degree of legitimization. Where a substantial moral issue is involved, such as prohibition, divorce, or abortion, the government may wish to pass responsibility back to the electorate. Finally, referendums can be used to solve a political problem, such as an issue which causes intra-party splits (the United Kingdom EEC referendum in 1975), or where a government needs to resolve a difficult political dilemma (the Canadian conscription referendum of 1942, or the Scottish and Welsh devolution referendums of 1979).

On a more theoretical basis, proponents of the referendum process argue that important constitutional changes should not be made over the heads of the people; that referendums allow more effective popular control of public authorities; and that evolving concepts of democracy mean that a government can no longer rely on a simple majority of Parliament where major initiatives are involved.

The most avid proponents of direct democracy go further, and argue that a greater use of referendums would ensure that all issues were dealt with, not merely those of interest to the party in power; that governmental decisions would be brought closer to the people, ensuring, it is hoped, a greater degree of responsibility and interest; that public decisions would be publicly arrived at; that the will of the majority would be accurately expressed; that voter apathy and alienation would be forestalled; that the long-term public interest would be better served by voters considering the next generation than by politicians considering the next election; and that citizens would be able to maximize their own potential through increasing involvement in the democratic process.

On the other hand, there is also evidence that referendums can act as a barrier to change as often as they facilitate it. It has been suggested that "the referendum device has one enormous danger; used as it is likely to be used in Britain, it places an undue power in the hands of the parliamentary majority of the moment. That majority can exploit and misuse it."(6)

The other major criticism of referendums are variants on the themes that they:

  • undermine representative democracy;
  • are divisive; and
  • can be destructive of minority rights.

Each of these arguments can be true in some situations, and untrue in others. In a representative democracy, for example, a member of the legislature is elected by the people to use his or her best judgment on behalf of the electors. The judgment and consciences of the members of the legislature are paramount. In a direct democracy, such as Switzerland, it is the judgment of the public that is paramount. A binding referendum, or one initiated by the public, clearly fetters representative democracy, but is consistent with direct democracy. However, a non-binding plebiscite, called by the government to gain approval for a specific action or piece of legislation, can be a valuable tool even in a representative democracy.

One British commentator describes such plebiscites in this way:

Of course, our Constitution exists in six unwritten words: The Queen in Parliament is supreme. Those of us who believe in that kind of constitution and do not want a written constitution must admit that there is nothing incompatible with that and referendums; if Parliament in its supreme wisdom decides that a referendum is an aid to its processes, then it is perfectly constitutional.(7)

Even with plebiscites, or consultative referendums, the issue is not always simple. On the one hand, a plebiscite may enhance representative democracy by conferring additional legitimacy on a controversial governmental initiative. On the other hand, such plebiscites, if overused, may still give elected representatives an excuse to avoid exercising their own conscience and judgment. According to one western American politician:

I come from a state where referendums and initiatives are common; they have the great vice of weakening the will of legislators to tackle difficult problems. It is so easy to let the public decide whether there should be abortion, whether there should be immigration, whether there should be a tax, as we legislators say that we will leave that to the sacred and sovereign judgment of the people.(8)

Whether referendums or plebiscites undermine representative democracy or enhance it often depends upon the intent of the referendum as well as its form. In some cases, the result might influence a legislator to vote against his or her individual conscience, as was the concern when the possibility of putting the issue of capital punishment to a referendum arose in Britain. One influential Conservative wrote to The Times of London in 1978 as follows:

Such a referendum would raise all the grave problems of the authority of Parliaments and the position of MPs in the most acute form … there can be no doubt that the majority of the public would support the restoration of capital punishment … The only purpose of a referendum would be to bring pressure on MPs to vote for a proposal they would otherwise reject … I do not believe that whatever the referendum disclosed, members would vote against their consciences and so there would be no practical effect but the attempt to induce them to do so would be profoundly misguided.(9)

Indeed, commentators cannot even agree on whether referendums are an inherently conservative mechanism or a force for change. They can be used as a delaying and/or educational tactic, when it is perceived that a vocal or extremist minority has had an undue influence on legislation, such as in the Scottish devolution referendum. With the exception of those countries with a constitutional requirement for referendums, the right to call a referendum generally lies with the government. This makes it a powerful tool for the majority.

On the whole, government-sponsored measures seem to be rarely defeated except in Australia and Switzerland. Moreover, some such defeats have been the result of political miscalculation, or have been subsequently negated.(10) Australia stands out as the only country where government-sponsored referendums have been defeated more often than not.(11)

On the other hand, in the first referendum of the 20th century, the people of Norway voted by 99.9% to separate themselves from Sweden, a country of which they had been a part for two or three centuries. This sort of result leads commentators to argue that: "the notion that referendums are barriers to change is simply in sharp contradiction to the facts."(12)

Whether or not a plebiscite is divisive can also depend largely upon the issue and circumstances. On the one hand, reducing complex issues to a simple question capable of a yes or no answer reduces the possibility of a consensus solution emerging. Moreover, unlike legislation, the results of referendums are usually impossible to ignore, even if it becomes clear that the effect will be divisive and unsatisfactory.

The Irish referendums of 1983 (to incorporate the ban on abortion into the constitution) and 1986 (to allow for divorce after five years of marriage breakdown) provide an example. The 1983 referendum resulted from a political error of judgment by Prime Minister Garret Fitzgerald, combined with a determined pro-life lobby. Critics of the referendum claimed that it was unnecessary, since abortion was already banned in Ireland except where the mother’s life was threatened; sectarian, because it was opposed by the Protestant and Jewish minorities; and potentially medically dangerous, in that the proposed amendment could put the mother’s life at risk.

The amendment passed by a two-to-one majority, but only half the populace voted. The high rate of abstention was generally considered due to disapproval over the divisive nature of the referendum. The deputy Prime Minister described it thus:

Never in our history have we seen a campaign that was so bitter and divisive as the present one. … Neighbour has been turned against neighbour. Eminent professional men have bitterly denounced each other in public. The churches could hardly be further apart. Colleagues in the same political party have launched personal attacks on each other.(13)

The Prime Minister himself issued a statement on the eve of the referendum, apologizing for having contributed to the rancour and division caused by the campaign, and accepting a share of the responsibility for the holding of the referendum.

In 1986, Prime Minister Fitzgerald initiated another constitutional referendum to allow for civil divorces, confident that the liberal urban vote, together with the youth vote, would carry the day. Instead, the referendum lost 60-40, with "city pitted against country, young against old and state against church."(14) In both cases, the results of the rancorous and emotional campaigns had extra-national implications, affecting the country’s relations with Northern Ireland and Britain.

On the other hand, divisiveness seems to have as much to do with the immediate circumstances and politics as with the referendum process itself. In Italy, a referendum can be popularly initiated by 500,000 voters or by five regional councils, and in May 1974, such a referendum approved legislation liberalizing divorce laws, with no apparent ill-effects.

In 1978, Italy passed legislation permitting abortions within the first three months of pregnancy. Subsequently, two referendums were requested by the populace. The first (the "Catholic" referendum) would have restricted the reasons for which an abortion could be obtained; the second (the "radical" referendum) would have allowed abortion at any time during the pregnancy. The electorate rejected both proposals, the "radical" by 80% and the "Catholic" by 70%. This is all the more remarkable since the attempted assassination of the Pope took place towards the end of the campaign, and emotions ran high.

In Ireland, referendums on abortion and divorce became divisive, and resulted in laws that were more conservative than wished by the government. In Italy, referendums on the same issues further legitimated legislation that had considerably liberalized traditional views. There is no easy answer as to why the two countries had such different experiences, but the comparison suggests that it is dangerous to over-generalize about the nature and effects of referendums.

Perhaps the clearest defect of referendums is the possible tyranny of the majority. The Swiss referendum process allowed women to be denied the vote for much longer than in the rest of western Europe, and allowed the forced expulsion of immigrants from Switzerland to become a serious issue. Legislators, who are accustomed to seeing firsthand the specific results of their votes, often tend to be more tolerant than the public, which has less difficulty abridging the rights of a minority with whom they do not identify. A British commentator once noted:

Nobody has any doubt that if there were a referendum on capital punishment, there would be an 85 per cent majority in favour of it: there would probably be an 85 per cent majority in favour of repatriation of immigrants; there might well be a majority in favour of abolishing the closed shop and doing all sorts of terrible things to the trade unions.(15)

Arguably, the Dominion Plebiscite Act of 1942 is an example of the dangers. Mackenzie King had promised during the 1940 election that there would be no conscription for overseas military service. Subsequently faced with a critical manpower shortage, the government held a referendum asking the public to release it from the election promise.

Critics charged that during the 1940 federal election, the promises against conscription had been made to Quebec, not to Canada as a whole, an argument that was not at all unsound. After the election, all of Canada was now being asked to free the government from its pledge to Quebec.(16)

Even with respect to the interests of minority groups, however, there is another side to the issue. Sometimes a minority consists of an active political lobby group trying to push through legislation that does not have the support of the more passive majority. It has been suggested, for example, that the Scottish devolution referendum defused a political crisis by taking the initiative from a radical element, and placing it before the people. According to one British commentator:

The referendum is an important constitutional tool; we need it in this country without a written constitution. It is a delaying weapon, one that reasserts the right of the majority to have its voice listened to at a time when active, well-organized minorities too often can shift the balance of power irreversibly.(17)

Overall, the reasons for, and results of any referendum seem to have more to do with the issues, events, personalities and politics surrounding it than with a generalized "theory" of referendums. A description of General de Gaulle’s referendums during the Fifth Republic illustrates the point:

The aims of the referendum during the Fifth Republic may be summarized by an inelegant and contrived alliteration: It was an instrument of consultation, consensus, consent, consecration, confirmation, crisis, confrontation and confusion. It purported to consult the people on major decisions, to articulate a national consensus on such decisions and elicit the consent of the people for their implementation, to give official consecration to the new regime, to confirm the popular mandate of the president, to correct the "distortions" of the political system owing to excessive localism, to solve the crises of Algeria and of the state, to induce a confrontation with parliament, the parties and the traditional notables, and to confuse the opposition. It was intended to be both constructive and destructive, creative and nihilistic, consolidating and iconoclastic, infrequent in its use yet enduring in its impact. But above all, it was intended to be a powerful weapon of the president to be used whenever he thought the circumstances demanded.(18)

Even when referendums are held on similar subjects, at similar times and in similar countries, the results can be different and unpredictable. For example, the comparison between the Norwegian and Danish referendums on entry into the Common Market in 1972 is frequently referred to in referendum literature. Referendums are not generally part of the Norwegian constitutional process, but it was generally accepted that one would be held on the issue of Common Market. In the event, 78% of the popular voted and the vote was 53% against membership and 47% for it.

The Norwegian referendum was to have been consultative, requiring a subsequent three-quarter majority in Parliament for final approval. Prior to the referendum, about two-thirds of Parliament favoured entry, but it was expected that had the referendum approved entry, the necessary parliamentary majority would have been obtained. Furthermore, had the referendum not been held, many people had expected the government ultimately to obtain the requisite three-quarter majority.

The result had a profound impact on the political system, giving new prominence to extra parliamentary pressure groups. It diminished the authority and freedom of party political leaders and, ever since the referendum, it imported into the political parties a lack of decisiveness through fear of public reaction. There is, furthermore, a much greater sense of alienation between the political parties and the public today.(19)

On 24 and 25 September 1972, Norwegians rejected membership in the EEC by referendum; on 2 October 1972, Danes approved it. Referendums are more common in Denmark because they are provided for in the constitution. A draft law put to referendum is deemed rejected if a majority of the voters are against it, and if those negative votes comprise at least 30% of the qualified electorate. In Denmark, 90% of the population voted and EEC membership was approved by 63% to 37%.

The reason for the difference between two countries so closely related is difficult to explain. The most commonly cited factor is the differing attitudes of each country’s farmers and fishermen. In Norway, though farming and fisheries account for about 12% of the population, the producer co-operatives were extremely powerful and producer prices in Norway were between 20% and 30% higher than within the EEC. On the other hand, Danish farmers, who numbered some 10% of the work force, were confident of their ability to compete and expected benefits from EEC membership. Similarly, Danish fishermen, being deep-sea rather than coastal, were more confident that the Common Market would guarantee them a market.

Whatever the reasons for the difference in the vote, and regardless of speculation as to what would have happened had the results been reversed, the post-referendum results were similar. "In both countries, the established party system was smashed in the election of 1972 held shortly after the referendum."(20) Indeed, it has been noted that "although the Danish establishment won the referendum of October 1972, the following year saw a voter rebellion stronger and more lasting than the one in Norway."(21)

Overall, referendums are held to consult the people; to legitimize a course of action, especially in areas such as constitutional or territorial matters that go beyond the realm of daily government; or to resolve a moral or political problem that for some reason does not lend itself to normal party politics. They appear to work best when used sparingly, and reserved for major issues:

[The United Kingdom Common Market Referendum] was one of those rare matters dividing people so deeply that some sort of appeal is needed that is accepted as more legitimate than a simple vote in the House of Commons. To get the matter settled, to resolve the issue in favour of going in, one had to find a mechanism for legitimacy. Consequently, I would defend in a pragmatic way the occasional use of the referendum on major constitutional issues where one can guarantee a reasonably high level of interest among the people and where, because of the nature of the party system or the electoral system or a combination thereof, the choice between parties cannot easily settle the matter.(22)


   A. Authority for the Referendum

      1. Constitutional

In some instances, the authority or requirement for a referendum is found in a country’s constitution. The Australian constitution, for example, requires that amendments be approved by an absolute majority of both Houses of Parliament and then by a referendum. In the referendum, the amendment must be approved by a majority of voters nationwide, as well as by a majority of voters in a majority of states (or in four out of six states).

The French constitution requires constitutional amendments to be approved first by Parliament, and then by either a referendum or a three-fifths majority of the two legislative Houses together. In addition, the President, on the proposal of the government, can call a referendum on any bill dealing with government organization or a treaty affecting government institutions.

The Irish constitution provides that a bill amending the constitution must, after passing both Houses of Parliament, receive a majority of votes in a referendum.

In Italy, the constitution has several provisions with respect to referendums, but the only one that has seen significant use is Article 75. This provides that a referendum to repeal a law, other than one dealing with fiscal matters, treaties or amnesties, must be called on the demand of 500,000 voters or of five regional councils. A simple majority is sufficient for the law to be repealed.

The Danish constitution allows one-third of the members of the Folketing to petition for a referendum on a bill within three weekdays of its final passage. There are also provisions for referendums on constitutional amendments or changes to the voting age. In addition, legislation that delegates power to an international authority must be approved by either a five-sixths majority of the Folketing or by a simple majority together with a majority of voters in a referendum. Danish law also allows the Folketing to adopt a law providing for a non-binding referendum on any topic as a way of consulting people.

The Spanish constitution provides for referendums on constitutional amendments if requested by one-tenth of the members of either chamber of the Cortes Generales (Parliament). In addition, a consultative referendum can be called by the government on political decisions of special importance. Referendums were held in 1976, when the province of Andalusia was dissatisfied with the pace of its talks with the central government, and in 1986 on whether Spain should remain a member of NATO.

The Swiss constitution, based on the concept of direct democracy, relies heavily on the referendum process. All alterations to the federal constitution must be approved by referendum. Constitutional amendments can be initiated by 50,000 citizens, but historically it has been difficult to induce the electorate to accept a popular initiative. Constitutional amendments, however initiated, must be approved by a majority of the voters, and by a majority of cantons (or eight cantons). Referendums can also take place on an ordinary law when 30,000 citizens or eight canton governments demand that a law be subject to a referendum within 90 days of publication. In this case, only a simple majority of voters is required to abrogate the law.

The Proposed Resolution respecting the Canadian Constitution, issued by the Trudeau government in 1980 after the unsuccessful federal-provincial talks held in September of that year, included a provision authorizing amendments by referendum. The amending formula contemplated a referendum as an alternative to resolutions by provincial legislatures, and would have required an absolute majority as well as a majority in each of the provinces.(23)

      2. A Referendum Statute

Referendums can also be held under the authority of a general authorizing statute. The Quebec Referendum Act of 1978 allowed referendums on any question approved by the National Assembly or on any bill adopted by the National Assembly. Bill C-9, the proposed Canada Referendum Act introduced in 1978, provided for a referendum on any question relating to the Constitution of Canada. The Constitution was defined to include customs, conventions and enactments, as well as national political and judicial institutions or processes. However, to avoid permanently establishing referendums as part of the constitutional process, the bill had a "sunset" provision of five years.

      3. A Specific Statute

Referendums can also be held pursuant to a statute authorizing a single referendum on a specific subject. In countries which do not have constitutional authorizations or requirements for referendums, this seems to be the most common procedure. It was used, for example, to authorize the British referendum on the Common Market in 1975. Later, when the Labour government of Britain devised a scheme for devolution of powers to the Scottish and Welsh assemblies, it failed in the first attempt to get the legislation through Parliament. Consequently, when an amended version was brought in, the government sought to mollify the critics by providing that the bill would not be implemented until approved by referendums in Scotland and Wales.

Prior to the 1970s, Canada had passed specific statutes to authorize each of its referendums: the Prohibition Plebiscite Act, 1898, and the Dominion Plebiscite Act, 1942. When a referendum is authorized by a specific statute, the question arises as to whether the actual wording of the question is included in the statute, or is left to the executive. In both the Prohibition Plebiscite Act and the Dominion Plebiscite Act, the exact wording of the question and form of the ballot were set out. Section 8 of the Prohibition Plebiscite Act, for example, stated:

  1. The ballot papers for the purpose of voting under this Act shall be in the following form:

Are you in favour of the passing of an Act prohibiting the importation, manufacture or sale of spirits, wine, ale, beer, cider and all other alcoholic liquors for use as beverages?


The Newfoundland Referendum Act set forth both the form and the order of the questions, itself a potentially contentious issue. Since three choices were offered (commission government, confederation with Canada, and responsible government as it existed in 1933), provision was made for a second referendum should no choice gain an absolute majority on the first ballot. The Act also provided that the option receiving the smallest number of votes would be dropped from the ballot on the second referendum.

René Lévesque later referred to the Newfoundland referendums as a precedent for multiple referendums on the same issue:

So we had to prepare [a referendum law] after studying what had been done elsewhere, notably in Newfoundland, where the people were consulted on two occasions prior to the entry of Newfoundland into Confederation. We also studied the case of England when it entered the Common Market, the Swiss cantons, and France…

I formally exclude a second referendum during the same mandate. Furthermore, our law forbids this. For the rest, we cannot be sure … There is a precedent for this, here in Canada … the province of Newfoundland was the only one to be democratically consulted over its entry into Confederation … The inhabitants of Newfoundland were consulted twice … The precedent exists.(24)

However, the specific provisions in the Newfoundland Referendum Act for reducing three options to two make it a dubious authority for multiple plebiscites on a yes/no issue.

   B. Originating a Referendum

      1. Executive Branch

In most cases, a referendum proposal originates with the executive branch, either with or without the formal consent of the legislature. Often, a referendum proposal is approved by the legislature, but the impact of this check varies widely depending upon the nature of the government concerned.

Section 11 of the French Constitution of the Fifth Republic, for example, allows the president to call a referendum on any government bill dealing with the organization of the public authorities "on the proposal of the government during parliamentary sessions." If the result of the referendum is favourable, the president can promulgate the bill. Amendments to the constitution, however, must be passed by both assemblies before being submitted to a referendum (section 89). General de Gaulle, distrustful of parties and parliamentary representatives, interpreted the referendum provisions in a controversial manner, and used them to bypass Parliament.

The initiative for holding all the referendums of the Fifth Republic lay squarely and manifestly with the president of the republic. The referendum of January 1961 on Algerian self-determination was announced by General de Gaulle on 16 November 1960, and the "proposal of the government" was made on 8 December. The referendum of October 1962 on direct election to the presidency was announced by de Gaulle as early as 29 August — three weeks before an unsuccessful attempt on his life — when he declared his intention of modifying the constitution "to ensure the continuity of the state." He made his intention officially known to the Council of Ministers on 12 September, and the "proposal of the government" was made on 2 October.(25)

Aside from the highhandedness with which General de Gaulle proposed referendums prior to a "proposal of the government," at least two of the referendums (1962 and 1969) involved constitutional amendments, and arguably should not have been proposed under section 11 at all.

The Quebec Referendum Act provides for questions to be moved either by the Premier or to be contained in a bill. In the former situation, there is to be a 35-hour privileged debate, allowing the wording to be amended or sub-amended, prior to the final approval of the National Assembly. In the latter case, the question would be adopted in the same manner as any other clause of the bill.

Under Bill C-9, the Canada Referendum Bill, the question would have been proposed by the Governor in Council, subject to the approval of the House of Commons and the Senate after not more than 40 hours of debate. Only a member of the government would have been entitled to propose a motion authorizing a referendum. The referendum could have been held in one or more provinces, and have involved more than one question so long as each question was capable of a "yes/no" answer.

Under the federal constitutional proposals of 1980, a referendum to amend the constitution would have been held when directed by the executive and authorized by resolutions of the Senate and House of Commons.

      2. Legislature or Political Party

In Australia, there is no technical requirement that amendments to the constitution be put forward by the government of the day, and motions urging constitutional change, and even bills with the same end, have been sponsored by the opposition or private members, although never successfully. In Denmark, one-third of the members of the Folketing can petition for a referendum on a bill within three weekdays after its passage. In Spain, one-tenth of the members of either House can call for a referendum on constitutional amendments.

      3. People or Local Government

In some countries, referendums can be originated by a given number of people or local governments, and this is often called the "initiative," especially where the results are binding.

In Italy, for example, a referendum on constitutional amendments can be requested by one-fifth of the members of either chamber, or 500,000 voters or five regional councils. An ordinary law, except those relating to fiscal or treaty matters, can be repealed by a referendum, if one is demanded by 500,000 voters or five regional councils. This latter approach, of an initiative that can repeal but not initiate legislation, is generally considered to have merit.

Switzerland allows for a constitutional initiative when 50,000 citizens propose a referendum. As with other constitutional changes, to become effective such proposals must be accepted by a majority of those voting as well as by a majority of cantons.

The referendum provisions in the proposed Canadian Constitution of 1980 did not provide for popular initiatives, but the matter seems to have been given some thought by the Trudeau government:

The Canadian public should have the right to demand constitutional change in a referendum, Prime Minister Trudeau said [at the September 1980 federal-provincial conference]…

He told the provincial premiers that an amending formula should include "some recourse to the people" through a referendum to enable voters to override provincial or federal opposition to constitutional change.

Although Trudeau didn’t elaborate, his proposal is consistent with a strategy outlined in a Cabinet document leaked to provincial delegations earlier this week [the "Kirby memo"].

The document said Trudeau should float the idea that citizens could "initiate referenda" in the event of negative action or lack of action by Parliament or legislatures on an amendment proposal.

"This would support the view that sovereignty ultimately resides in the people," the Cabinet strategy paper says.

The leaked document suggests that three per cent of the national or provincial electorate could initiate a referendum to make constitutional change. If referendums were carried in enough provinces, the amendment would be adopted.(26)

The Quebec government in 1977 declared itself favourably disposed to the initiative in principle, but felt that its introduction to Canada would need much more study:

The government is in favour of gradually extending the procedures of consultation to allow for a referendum originating with the people. It is obvious, however, that such procedures could not be immediately introduced without pushing aside a number of important factors.(27)

   C. Wording the Question

The Newfoundland Confederation referendums of 1948 are an example of how difficult it can be to find the correct question. Canadians argued that the wording of the question put commission government at an advantage. Anti-Confederationists argued that it was unconstitutional to even include the possibility of union with Canada on the referendum. Others felt that if union with Canada was presented as a possibility, union with the United States should also be added. The Canadian High Commissioner wrote to British authorities arguing that the wording of the question had been decided at the insistence of the Governor of Newfoundland: "I can only conclude that [British officials] have been prepared to buttress his position even at the cost of placing Confederation with Canada at a serious disadvantage."(28)

Denmark offers two interesting examples of how the wording issue can be addressed. In the EEC referendum of 1972, the electorate was asked to vote simply JA (yes) or NEJ (no); no question was put on the ballot, which consisted only of the words "ja" and "nej." A form attached to the ballot gave directions:

Voting concerns the bill relating to Denmark’s accession to the European Communities adopted by the Folketing: He who votes FOR the bill puts an X at YES, he who votes AGAINST the bill puts an X at NO.(29)

Denmark was, however, already used to an imaginative approach towards referendum questions. The constitution prior to 1953 required the approval of 45% of the electorate (rather than of actual voters) for constitutional amendments. This made constitutional change rather difficult given that no more than 50% of the electorate showed up for the first several referendums.

For example, a set of amendments was strongly recommended in May 1939 by prominent politicians representing the major parties, which had collectively received more than 1.2 million votes at the election one month earlier. Only 966,000 voters turned out to support the amendments, however, and a 91.9% vote of yes proved insufficient since it constituted only 44.9% of the Danish electorate.(30)

Obviously, something had to be done. The government proposed a new set of constitution amendments, which relaxed the requirements for future constitutional amendments. Since it seemed that the trick was less gaining approval than having a sufficiently large turnout, the government resorted to the "princess gambit":

When comprehensive constitutional changes were again submitted to the Danish electorate in 1953, politicians took care to include a measure believed to have great popular attraction — a change in dynastic rules making it possible for the young princess Margaret to succeed to the throne. This seemed to do the trick. A vote of 78.4% yes on a 58.3% turnout provided just enough votes in favour of the new constitution, which relaxed the conditions for amendment although it introduced new referendum provisions.(31)

Linking two separate issues is not uncommon in a referendum, but it can also backfire. The French referendums of 1961 (Algerian self-determination) and 1962 (independence) both asked essentially two questions: "Do you agree with the policy proposed, and do you want General de Gaulle to implement it?" According to another commentator,

…the worst example of the dishonest and obscure question was that put in the April 1969 referendum. It was deceptively simple: "Do you approve of the bill dealing with the creation of regions and the reform of the Senate?(32)

The bill in question was 8,000 words long, covered 14 tightly printed pages, comprised 69 articles, and involved the modification and replacement of 19 articles of the constitution. Revisions in the interim presidency provision were not even referred to in the question. Even government supporters told the opinion polls that they wanted to give separate answers to the two questions. In April 1969, 53% of the voters rejected the proposals, and de Gaulle, having made it an issue of confidence, immediately resigned.

It is difficult to find the balance between a question that is too simple, and one that is incomprehensible. Commentators have suggested that the high percentage of referendum proposals rejected by the Australian electors may result from simple frustration with the wording of the question:

The very form of the referendum ballot paper may put off some voters. The question asks: "Do you approve the proposed law for the alteration of the Constitution entitled…" followed by the short title of the bill. In 1974, for example, the first question related to "A Bill for an Act to facilitate alterations to the Constitution and to allow electors in the Territories, as well as the States, to vote at referendums on proposed Laws to alter the Constitution." Not only are the words unfamiliar to most electors, but the wording obscures the fact that the principal aim of this bill was to reduce from four to three the number of states registering approving majorities.(33)

The British EEC referendum offers an example of both how delicate the issue of wording can be, and of how little difference it can make in the long run.

Though it had been suggested that more than one question should be posed on the ballot paper, and that these questions might include one asking if the voters preferred to leave the matter to Parliament, the government felt that the advantage lay in simplicity. There would be only one question on the ballot, and the wording proposed in the White Paper was:

The government has announced the results of the renegotiation of the United Kingdom’s terms of membership in the European Community.



Debate ensued. At the beginning of February, pollsters had tried out different formulations which had produced a great variety of results, and they finally concluded that if the issue was clear, by the end of a highly publicized campaign, the wording would not matter. Despite these findings, there were vehement arguments in Cabinet over the ballot. The Foreign Office wanted to see a long preamble to the question, while the anti-Market faction demanded that the words "Common Market" replace "European Community." The Referendum Bill finally adopted the question as worded in the White Paper, though the government gave in to the demands of the anti-Marketeers and agreed to add the words, "the Common Market" in brackets, to the question. The final ballot read:


Even this wording led to some administrative concern when it was found that the printers had set "Common Market?" separately on a new line.(34)

In Australia, fiddling with the wording is not common because of the standard form of the ballot paper, but the two conscription referendums of 1916 and 1917 were strangely worded indeed. The 1916 question ran as follows:

Are you in favour of the Commonwealth having, in this grave emergency, the same compulsory powers over citizens in regard to the requiring of their military service for the term of this war, outside the Commonwealth, as it now has in regard to military service within the Commonwealth?

This was narrowly defeated, and the Hughes government tried again at the end of 1917, with a question that was terser but no more helpful to the elector: "Are you in favour of the proposal of the Commonwealth government for reinforcing the Australian Imperial force overseas?" This was defeated by a rather larger margin.(35)

This can be compared with the Canadian effort at wording the 1942 referendum on conscription:

Are you in favour of releasing the government from any obligation arising out of any past commitments restricting the methods of raising men for military service?

Perhaps the most interesting aspect of the comparison is that the voters in both countries did seem to know what they were voting for.

The issue of wording the question is not irrelevant in post-war Canada. At an international conference on referendums, held in England in 1979, the co-editor of the definitive text on referendums made the following comment:

I am less upset than some people by the bogusness of the question that may be put in Quebec. There was a good deal of talk in Britain about the wording of the question in the 1975 referendum, and there was a poll study that asked the question hypothetically in eight different ways, with major differences in responses to the different formulations. People were upset at this. The wording of the question about entry into the Common Market or the European Community did not matter. At the end of an election campaign, where things had been discussed fully, people knew perfectly well what they were doing. They were voting yes or no. Nobody went to the ballot box and actually read the words.(36)


   D. Effect of a Referendum

      1. Referendums and Plebiscites

Although the words "referendum" and "plebiscite" are often used interchangeably, their origin is different. The term "plebiscite" originated with Roman law, and described a law enacted by vote of the plebians, or commoners, at the request of a plebian magistrate.(37) It now describes a vote by the electorate of a political jurisdiction, expressing their opinion on some specific issue or question. The word "referendum" identifies a similar method of "referring" an important legislative measure to a direct vote. The term did not appear in English until the late nineteenth century, but the Swiss cantons had decided issues ad referendum 200 years earlier.

Under this definition, plebiscites are essentially consultative and are not legally binding on a government, however great the political or moral obligation to respect the result. Referendums, in the strictest sense of the word, create law directly with no subsequent government intervention. The distinction is not always accepted, however, and many texts use the term "referendum" to describe non-binding consultative votes as well as binding referendums.(38)

For example, "The ABCs of a Referendum in Quebec" defined the purpose of a referendum as being "to consult the electorate on a given subject through polling similar to that of a general election."(39) The federal government, in its 1978 pamphlet "Canada Referendum Bill: An Explanation," suggests that "modern usage in both languages clearly favours ‘referendum’ and the word ‘plebiscite’ seems to have lost all currency."(40)

On the whole, however, lawyers tend to draw more of a distinction between referendums and plebiscites than do political scientists. Perhaps this explains why, notwithstanding the Canadian and Quebec pamphlets suggesting little difference, a 1989 article drew a clear distinction:

A referendum is a legally binding verdict of the people which must be acted upon by the government in the form of a law. Courts have held that due to the principles of parliamentary government, Parliament or provincial legislatures cannot be legally bound by such direct lawmaking by the people. Only municipal councils may truly use referendums in the sense that they may be held legally accountable for the result.

A plebiscite is little more than an expression of public opinion, it is advisory only and not binding on government, whether federal, provincial or municipal.(41)

Perhaps one of the reasons for confusion is that:

The distinction sometimes made between a referendum and a plebiscite — between a popular vote that is immediately legally normative and binding, and a popular vote that carries merely moral authority — is one that seems limited to the English-speaking, common law world.(42)

      2. Binding Referendums

Binding referendums are rare in parliamentary democracies, and are best suited to countries with a tradition of direct democracy, such as Switzerland:

We have a different view of referendums in Switzerland because we do not think that we are a parliamentary democracy. We call ourselves a direct democracy; therefore, constitutionally, it is clear that the ultimate power is with the people … The Supreme Court of Switzerland is never asked whether a popular referendum is permissible. The final power is with the people. They are higher than any branch of government. The Supreme Court would never declare any referendum unconstitutional. Let me give two examples. The female vote was not introduced in Switzerland until 1971 … Is this unconstitutional? It is not, as long as the people say it is not. It is not possible for a Supreme Court to say that the people are violating basic human rights.(43)

Similarly, a constitutional initiative launched to expel foreign workers from Switzerland was rejected by nearly 60%; however, if it had been successful, the courts could have done nothing to protect the foreign workers, even though they had legal work permits.

Switzerland had 297 referendums between 1848, when it became a federation, and 1978. Votes on obligatory referendums, or proposals that came from the Parliament, have usually been positive, with 100 "yes" votes to 38 "no" votes. Votes that have been demanded over parliamentary proposals under the optional referendum procedures are more often negative than positive — 51 "no" and only 34 "yes" votes. Votes arising from popular initiatives are almost always negative, with 67 "no" votes to seven "yes" votes. On the whole, therefore, the Swiss use of the referendum appears to be very functional.

One area of concern, however, is that Switzerland has a much lower turnout than that in neighbouring countries, possibly as a result of the large number of votes. Moreover, the average percentage of the population that votes has declined over the years:

1880-1913       58%

1914-1944       61%

1945-1959       54%

1960-1969       43%

1970-1978       42%.(44)

Overall, direct democracy seems to run smoothly in Switzerland, although it does not appear to be an exportable commodity:

Switzerland, then, is the home of the referendum. It is the country that has used referendums most. It has made referendums an integral part of its political life. It has shown that, at least in a small, sophisticated country, direct democracy can work with almost none of the ill consequences which have been ascribed to it in political argument elsewhere. Switzerland provided the model on which American and other evangelists for direct democracy based their cases in the 1890s and 1900s. And the Swiss can claim that the original continues to work better than its imitators.(45)

Binding referendums are not possible under the present Canadian constitution:

Under the present constitution, a referendum has no direct legislative authority since any such interpretation would trespass upon the powers and functions of the Lieutenant-Governor … This constitutional obstacle in no way diminishes the political value of a referendum since the first binding law of democratic process is that of the clearly expressed majority, and a government can always commit itself explicitly to accepting the result of a referendum.(46)

      3. Abrogative Referendums

The abrogative referendum allows the electorate to reject or repeal a measure passed by the legislature.

The constitution of Denmark provides that one-third of the members of the Folketing may call for a referendum on a bill after it has been passed. For the bill to be defeated by the referendum, a majority of votes cast must be in the negative and these must represent at least 30% of eligible voters. The abrogative provision does not apply to certain bills, most notably those dealing with fiscal issues or treaty responsibilities.

Article 75 of the Italian constitution allows legislation to be repealed by a referendum called on the demand of 500,000 voters or five regional councils. Specifically excluded are "fiscal or budget laws, amnesties or pardons, or laws authorizing the ratification of international treaties." The implementing law was not passed until 1970, and there had been nine referendums as of 1986. The first referendum, in 1974, repealed the divorce laws, but none of the later referendum proposals was successful. Four concerned anti-terrorist legislation, two were proposals to restrict access to abortions, and the others concerned state financing of political parties and automatic wage indexation.

There are some indications that referendums on constitutional amendments tend to negate bills that are particularly badly drafted. Perhaps this is simply because the voters sense that a bill which cannot be logically explained has some internal problems:

It seems to me that if a referendum is confined to the approval or rejection of a bill passed by Parliament, the bill’s supporters will be more concerned to seek defensible legislation rather than cobble together a bad bill and push it through willy-nilly. One of the reasons why the Scotland Act and the Wales Act were not more strongly supported in the referendums was that they were terrible pieces of legislation.(47)

De Gaulle went down to defeat in the 1969 referendum partially because "the constitutional experts of the Council of State described the bill as ‘one of the worst drafted ever to come before the council’."(48)

Overall, the abrogative referendum appears to be among the most successful and least disruptive types of referendums. As experience in Italy and Switzerland shows, abrogative referendums are normally unsuccessful, so they are seldom destabilizing. Their effect on the principles of representative democracy is minimal, since they only take place after the legislative process is complete. Legislators cannot take refuge in abrogative referendums merely because they do not wish to deal with a difficult issue. Nonetheless, such referendums are an ultimate check by the electorate on the legislators. The mere possibility of an abrogative referendum confers additional legitimacy on constitutional amendments, since a highly unpopular amendment would doubtless be challenged.

      4. Plebiscites or Consultative Referendums

A distinction is sometimes made between plebiscites and consultative referendums, but it is a difficult line to maintain. Arguably, a plebiscite approves or legitimizes an actual government initiative, whereas a consultative referendum is purely advisory. In fact, given the moral and political weight of a referendum of any sort, plebiscites and consultative referendums tend to merge.

Referendums held to consult the public are not legally binding, but they do tend to be politically binding. The Quebec referendum of 1980 did not technically bind the Quebec government, far less the federal government, but there was no question as to its political and moral significance.

Because consultative referendums gauge public opinion rather than legitimize a specific action, they can leave the government considerable flexibility to respond. A consultative referendum was held in Sweden in 1980 on whether the country should continue to use nuclear power, and all parties declared they would abide by the outcome. The results were indecisive but assisted in fashioning a compromise, and the various parties were left with more flexibility than if the referendum had been legally binding.

However, governments do not always continue to feel themselves bound by a past referendum. In 1955, an overwhelming majority of Swedes rejected a proposal for driving on the right-hand side of the road. Nonetheless, in 1963 the government implemented right-hand driving without holding another referendum.

Arguably, the plebiscitary referendum is a means to legitimate a government initiative or decision, while the consultative referendum discovers what type of future government action the public would prefer. However, the difference between the two can be illusory:

The consultative referendum is a slippery device. It may remain genuinely advisory if rarely used. If invoked often or in relation to issues which polarize opinion (for example, adhesion to the EEC in the case of Norway in 1972), its consultative status becomes illusory. To defy its verdict would be to flout the most important basis of legitimacy in contemporary democratic states, that of popular consent.(49)

   E. Subject of Referendum

      1. Constitutional Amendment

The constitutions of a number of countries make provisions for referendums to be held when constitutional amendments are proposed.

  • In Australia, constitutional amendments must be approved by referendum, receiving a majority of all votes cast as well as a majority of the votes in a majority of the states (or four out of six states).
  • In Austria, significant constitutional amendments must be approved by referendum if one-third of the National Council or the Federal Council (the Upper House which represents the "Lander" or provinces) request such a vote.
  • In Denmark, constitutional amendments must be approved by a simple majority of those voting in a referendum, representing at least 40% of eligible voters.
  • In France, constitutional amendments must be confirmed by either a three-fifths majority of both Houses sitting together or by a referendum.
  • In Ireland, a bill amending the constitution must be submitted to a referendum after passing both Houses of Parliament.
  • In Italy, a referendum on a constitutional amendment is required only if requested by one-fifth of the members of either chamber, or by 500,000 voters or five regional councils, within three months of the publication of the amendment.
  • In Spain, a constitutional amendment can be submitted to a referendum for ratification if one-tenth of the members of either chamber so request within 15 days of the amendment passage.
  • In Switzerland, all alterations of the constitution must be affirmed by a majority of citizens. In addition, any amendment proposed by 50,000 citizens becomes the subject of a referendum (unless Parliament responds proposing its own amendment, and the sponsors withdraw their initiative). Constitutional changes must be approved by a majority of those voting, and by a majority of cantons.

Gregory Mahler, discussing the referendum procedures in the constitutional proposals of the federal government in 1980, suggests that the three most instructive comparisons for Canada are Australia, Switzerland and West Germany:

We can see, then, that the proposed Canadian amending formula would have been quite a bit different from these other procedures. The German case does not require a referendum or subsequent approval by the states, but the states are directly represented in the Bundesrat; the approval of their representatives cannot be avoided. However, an amendment could be passed over the objections of one or two or three of the German States. The Australian Constitution does not give the states an opportunity to approve or disapprove of amendments, but significant popular majorities in a referendum are required for their passage. The Swiss procedure represents a midway point: popular approval for amendments is required via a referendum and the states must approve any referendum put forth.(50)

Mahler goes on to note that the Trudeau referendum proposal for constitutional amendments would have significantly reduced the existing powers of the provinces. Specifically, the federal government could have had an amendment approved by referendum, even if it was opposed by all of the provinces; however, the reverse was not true.

      2. Territorial or Sovereignty

Where there is a conflict over the allegiance of a territory, or over the proper boundaries of a state, a referendum of plebiscite … is frequently seen as the best means of resolving the dispute. Indeed, it may seem to be the only method through which the irreconcilable creed of nationalism can be rendered compatible with popular governments.(51)

The first formal plebiscites affecting sovereignty were held in the late eighteenth century to determine if the papal territories of Avignon wished to be annexed by revolutionary France. In the nineteenth century, referendums on self-determination played a major role in the unification of Italy:

It is fair to say that the modern state of Italy was built by a series of referendums in which overwhelming majorities turned out to vote for the unification of their country.(52)

Although none of the original six members of the Common Market held referendums on its formation, the potential "second round members" did: Ireland, the United Kingdom, Denmark and, with negative results, Norway. One of the lessons of the EEC referendums in the 1970s is that sovereignty referendums can be dangerously unpredictable:

The Norwegian Social Democratic Party concurred in the holding of a referendum in the belief that the Norwegian people would vote yes; instead, they voted no. The anti-Europeans in the British Labour Party made exactly the same mistake; the 1975 referendum tied Britain even more securely to Europe. Finally, Pompidou held a referendum [to approve the admission of Britain, Norway, Denmark and Ireland] that he need not have held in order to increase his prestige; instead, the outcome had the effect of diminishing it.(53)

The United Kingdom "devolution referendum" of 1979 asked the citizens of Scotland and Wales if they wanted the establishment of their own assemblies, as provided for in the Scotland Act and Wales Act of 1978. This is discussed in more detail below under the heading "Counting the Vote."

One of the deep-seated objections to referendums is that the question put forward is often far from the real issue. In the case of devolution, the issue was whether there should be a breakup of the British State.

Professor Steiner says it is not like that in Switzerland; the devolution referendum is different from anything in the Swiss experience, because no one in Switzerland had put forward a proposition that a canton should separate from the Swiss federation. There are referendums about all sorts of other things but never on the breakup of the Swiss state. There are two entirely different kinds of issues: constitutional issues, delicate and emotional, and general issues. No one ever thought that there should be a referendum on whether the state of Oklahoma should secede from the federal union.(54)

Whatever difficulties a referendum may cause for questions of territoriality and sovereignty, it may be preferable to the alternatives. The traditional political process, based on adversarial parties, can be even more divisive than a referendum. Moreover, of all issues, those affecting sovereignty would seem to require the highest possible degree of legitimization.

As Benjamin Disraeli noticed, there is no finality in politics, and least of all over issues involving the volatile and sensitive emotions aroused by ethnic nationalism. One should not expect, therefore, to discover an instrument which can achieve finality in a democratic state. When that point is understood, then the referendum becomes, to paraphrase Winston Churchill, the worst possible instrument to resolve territorial disputes — expect for all others.(55)

      3. Moral or Political Issues

Referendums on moral or political issues tend to involve a deliberate transfer of decision-making:

Governments have been reluctant to settle issues on which they were themselves divided; they have wanted to avoid responsibility for decisions which would be unpopular with a significant section of the public. Referendums have offered a way of passing the buck. The EEC referendums in Norway and Britain, the Leopold III referendum in Belgium, the divorce referendum in Italy, and the various prohibition referendums in Scandinavia and Australasia exemplified this. So, in a different way, did the military conscription referendums (Australia, 1916 and 1917; Canada, 1941; New Zealand, 1949).(56)

Often, moral issues such as divorce, prohibition or abortion can cut across party lines and a referendum becomes a means of avoiding parliamentary disruption. Similarly, referendums can involve political problem-solving when an issue, such as EEC membership in Britain, threatens to disrupt party unity.

      4. Bill C-9, the Canada Referendum Act (1978)

Despite the traditional reluctance of Canada’s federal government to hold referendums, Bill C-9, the Canada Referendum Act, was introduced in 1978. This was soon after the Quebec legislature had passed the enabling legislation under which the 1980 provincial referendum was held, and was presumably intended to allow the federal government to hold its own referendum, on its own question, if the necessity arose.

The main provisions of the Canada Referendum Bill proposed in 1978 were:

  • referendums could only be held on constitutional questions;
  • referendums would be advisory or consultative and not have the direct force of law;
  • questions would provide a yes-no choice;
  • referendums could be held in one or more provinces;
  • referendums would be initiated by a government resolution in the House of Commons proposing a particular question. Debate would be for a maximum period of 40 hours; referendums could be held only on questions approved by the House of Commons and the Senate;
  • writs of referendum could only be issued within 45 days of Parliament’s approving the resolution to hold a referendum; the official campaign period would be 47 days;
  • all federal and provincial parties or other groups intending to spend more than $5,000 during the official campaign period would be required to register; registration could be at any time from the tabling of the resolution to the seventh day before polling; and
  • these registered parties or "registered referendum committees" could have a constituency representative with rights similar to candidates for the revision of voting lists and the scrutiny of the vote.

   F. Regulating the Referendum Process

On the whole, existing electoral law provides for the compilation of voter registers, the appointment of returning officers, the establishment and manning of polling stations, and the rules for secrecy of the ballot. While it is sometimes suggested that age restrictions should be relaxed for a referendum, this is not normally done. On the other hand, there may be a reason for giving a referendum vote to some classes of the electorate who, for reasons ranging from the need for absolute neutrality (judges) to a legal disability (prisoners), are denied the vote in a general election.

Some problems that arise with referendums are quite distinct from those in ordinary elections and largely have to do with identifying the various interest groups so that funding, media and scrutineering criteria can be fairly applied.

      1. Grouping the Interests

Generally, the most difficult issue with regulating the referendum is defining the recognized groups for purposes of applying general financial and media criteria:

The purpose of umbrella arrangements is to ensure fair play between the two sides in terms of money and of media access. If there is to be public subsidy of referendum campaigns, it is necessary to define who is entitled to receive it. If there is to be a limit on expenditure so that the richer side does not buy its way to victory, it is necessary to define who may spend money and be made accountable. If booklets presenting the yes and no cases are to be distributed by the authorities, it is necessary to define who is to prepare these statements. If the broadcast media are to offer balanced time, it is convenient to have some formal definition of who the representatives of the balanced arguments are.(57)

The British referendum on the EEC had a number of unique features, many of which related to the recognition of umbrella groups representing each point of view. A subsidy of £125,000 was made to each of the umbrella organizations, since it was evident that the pro-Europeans had greater resources than did those in opposition. Moreover, Cabinet solidarity was relaxed for the three months of the referendum; 16 members of the Cabinet campaigned for EEC membership while seven campaigned against.

The use of umbrella groups, initiated by the United Kingdom referendum of 1975, was adopted by Quebec, as were many other aspects of the British referendum. In fact, the government of Quebec was surprisingly laudatory about the British experience in a 1977 paper, "Consulting the People of Quebec," issued by the Minster of State for Electoral and Parliamentary Reform:

It will be noted that the provisions which have been retained, though owing something to many countries with strong traditions of democracy, spring above all from the experience of Great Britain in 1975, when she held a referendum on whether to keep her ties with the European Common Market. The parliamentary system, the long-standing democratic tradition, and deep-rooted sense of fair play are British concepts which have all been absorbed into Quebec’s own political tradition, and the referendum mechanisms which Great Britain built on these concepts seem to the Government of Quebec to be an invaluable guide in improving our own institutions.(58)

The Quebec Referendum Act provided that the members of the National Assembly could register with the Chief Electoral Officer the option they favoured within seven days of the adoption of the question or bill containing a question. Those in favour of each option then set up provisional committees. If, after seven days, no member of the National Assembly had registered in favour of an option, the Chief Electoral Officer could then invite between three and 20 electors to set up a provisional committee.

The Canada Referendum Bill, on the other hand, would have required all groups spending more than $5,000 in referendum expenses to register with the Chief Electoral Officer, but would not have limited the number of groups that could register.

The reasons given by the Quebec government for adopting the umbrella group approach were: to give equal treatment to each option under electoral practices that limited maximum expenditures; and to avoid the artificial situation of placing political parties in charge of an issue that might not break down along party lines. The equal treatment issue involved financial aid from the state and limits on external funding as well as spending.

Notwithstanding the accepted legitimacy of umbrella groupings, Quebec’s legislation provoked some criticism because, although the intent was to prevent distortion by wealthy interests:

The consequence was to place the argument almost exclusively in the hands of the province’s two main parties. People who wanted to urge a yes or no vote for reasons not approved by their side of the campaign seemed liable to be largely silenced.

When it was tested in 1980, the Quebec law worked more smoothly and proved less draconian than many had feared. There were no serious quarrels within the umbrella organizations — though it is arguable that the Liberals used their position to consolidate their complete preeminence over the other small parties as the opposition to the Parti Quebecois. Other bodies managed to get a voice in the referendum by institutional advertising and by conferences, as well as by speeches in the federal parliament and other newsworthy activities. The press and its correspondence columns provided a forum for all sorts of views.(59)

      2. Financing and Media Access

Once the umbrella interest legislation is in place, the main issues remaining are the extent of government contributions to the groups, reporting requirements and limitations on campaign contributions, ceilings on campaign expenditures, guaranteed access to the media (primarily television), possible publication of position papers at government expense, and the length of the campaign.

A decision has to be made on whether there should be public disclosure of contributions and expenditures. Since referendums often involve a highly emotional issue, the arguments for and against disclosure differ somewhat from those in general elections. It has been suggested that the "chilling effect" of disclosure may be greater in a referendum than in an election and several countries, notably Denmark and Norway, have rejected disclosure and reporting requirements on precisely these grounds.(60) A compromise may be to set the level of contributions that must be reported at a high enough level to keep reporting and enforcement burdens to a minimum.

Another generally contentious issue is whether there should be a ceiling on expenditures. On the one hand, expenditure ceilings keep costs within manageable limits, ensure that referendums cannot be "bought" by the richest side, and presumably increase public confidence in the result. On the other hand, ceilings that are set too low can prevent a truly effective information campaign. Moreover, expenditures ceilings are more difficult to enforce in a single issue campaign, when it is difficult to draw the line between campaign advertising and "lifestyle advertising." If, for example, Canada held a referendum on whether to ban smoking entirely, would a national advertising campaign by the Cancer Society on lung cancer be included in the expenditure ceiling?

Two related issues are whether there should be restrictions on contributions from organizations with an interest in the outcome of the referendum, and whether there should be a ceiling on contributions from any one source. Finally, there is the question of a "floor" for campaign costs, or the amount that should be contributed by government to ensure that all sides are able to run an effective information campaign.

   G. Counting the Vote

The criteria for a successful result in a referendum can vary considerably:

The decision may be by simple majority of those voting — or there may be some more stringent requirement, for example, two-thirds of those voting (Gambia), 50 per cent of the qualified electorate (West Germany), 45 per cent of the qualified electorate (Denmark until 1953), 40 per cent [of the qualified electorate] (the Scottish and Welsh referendums of 1979), or 50 per cent of those voting in a majority of component states or cantons (Australia and Switzerland). In the vast majority of cases, a simple majority has provided the verdict, although as King Leopold III of Belgium found in 1950, a 57 per cent vote is not enough to give acceptance and legitimacy to a constitutional head of state.(61)

The turnout in a referendum can also be of critical significance. In the case of the Newfoundland referendum of 1948, the United Kingdom made it clear that it would regard any majority, however small, as binding. On the other hand, in 1972, President Pompidou attempted to capitalize on General de Gaulle’s creative use of referendums by calling for the approval of entry into the European Community by Great Britain, Ireland, Norway and Denmark. Although almost 70% of the voters approved, two-fifths of the electorate stayed home and did not vote at all. This was the highest abstention rate in any French national election since the war. By holding an unnecessary referendum in the hope of enhancing his prestige, President Pompidou diminished it instead.

Where the results of a referendum are binding, either in law or by political agreement, the question arises as to whether a proposal should require the consent of a certain percentage of the entire electorate as well as a majority of those actually voting. The example most cited is the Scottish devolution referendum of 1979. In 1978, the Labour government introduced bills calling for devolution of some powers to new assemblies in Scotland and Wales, if approved by the voters in referendums held in those areas but not in England itself. A rebellion by backbenchers in early 1978 forced an amendment that "if in the opinion of the secretary of state less than 40 per cent of the Scots or Welsh entitled to vote voted for the act in question, the secretary would be required to lay a repeal order before Parliament." The election was held on 1 March 1979, and turnout became critical. There was no difficulty in Wales, which rejected devolution by 80% of voters and 47% of the eligible electorate, but the Scottish result was less clear:



Per Cent of
Votes Cast

Per Cent of

Did not vote




The government tried to delay but narrowly lost a vote of confidence on 28 March 1979. The experience seems to have dampened British enthusiasm for referendums, particularly since some commentators suggest that it would now be politically impossible to hold a referendum in Britain without incorporating the 40% rule.

Denmark now requires that 40% of all eligible voters approve a constitutional amendment. Referendums on non-constitutional bills, however, work the other way around; to be defeated, a majority of the votes cast must be in the negative, and these must represent at least 30% of eligible voters.

Quebec considered this issue when drafting its referendum bill, but decided that "since referendums are used for consulting the public, it would be pointless to include in the law special provisions requiring a certain majority vote or rate of participation."(62)

Another mechanism for dealing with the problem of low turnouts is to provide for compulsory voting. Voting has been compulsory in Australia since 1924. Although turnouts dramatically improved, the effect on the outcome of the subsequent referendums is unclear.

Only half the enrolled electorate voted in the referendums of 1906 and 1911, although six in ten voted in 1910, and more than seven in ten in 1913 and 1919. It is hard to believe that the increase to 91 per cent which occurred in 1926 had no effect at all on the outcome of the vote, for the most plausible a priori reasoning suggests that the dragooning of the uninterested and ignorant to the polls would have increased the number of those who like to vote no. There have been more successes in referendums since 1924, however, and there has continued to be a large number of near misses. No finding can be returned.(63)

Italy has semi-compulsory voting and a great many other countries deal with the problem by having referendums coincide with national elections, when political interest is highest. This also holds down the costs of the referendum, and causes the least additional interference with the business of government. However, political passions tend to run higher at election time, and the referendum can become confused with other issues.

The 1950 Belgium referendum (referred to above at fn. 61) was a consultative plebiscite to determine whether or not Leopold III should be restored to the throne. The problem was not one of a total majority; it was that Leopold III had a majority in the Flemish half of Belgium but only a minority in the French half. In such circumstances, the referendum results did not provide the answer, and Leopold III decided to abdicate.

A similar situation could have arisen in the Quebec referendum of 1980 if the results had been more evenly divided. Before the referendum, René Lévesque expressed his hope that the referendum proposal would not be narrowly defeated: "…any figure around the 45-50 per cent mark obviously signifies that the majority of francophones voted yes… It would mean that a francophone majority had clearly expressed a ‘yes’ but that they had been blocked by a majority." Even with a 60-40% rejection of the proposal, he did not feel the situation was entirely clear: "To this day, no one knows if our two-fifths vote for the ‘oui’ represented 49 or 51 per cent of French Quebec."(64)

Referendum results can differ dramatically according to geographical location and social background, even in comparatively cohesive societies. In the tightly fought Newfoundland referendum, which approved Confederation by a slim majority of 52 to 48%, geographic patterns were obvious:

The Responsible Government vote was concentrated mainly in the Avalon Peninsula; outside the Peninsula, every district but one voted for union. The division seems to have been partly on denominational lines, for the population of the Avalon Peninsula was strongly Roman Catholic, and the Roman Catholic Archdiocese had taken a recognizably anti-Confederation position.(65)

The above situations highlight the difficult question of how votes should be counted in a referendum. It has been suggested that "it was absolutely critical for the Belgium referendum that the results were available at a local level as well as for Belgium as a whole, because they demonstrated that one-half wanted Leopold III and the other half did not."(66) The issue, however, is not a simple one.

Britain, for example, has always counted and recorded votes in larger units than most other democratic countries, but in the 1975 EEC referendum the practice was carried even further so that no member of Parliament would be embarrassed by an evident discrepancy between his own views and those of his voters. Originally, the intent was to have all votes counted in London and announce only the national vote, largely because of concerns that a different result in Scotland, Wales or Northern Ireland could produce serious political difficulties. This idea was eventually abandoned for both administrative and political reasons, and votes were recorded by county rather than constituency.

The Quebec government noted the problem when preparing its own referendum legislation:

In an election, the votes are first counted in each polling station, and then pooled to give a total for the electoral district. This system could obviously be retained for referendums. It should be noted, however, that there is a risk that members may find themselves opposed to their constituents’ views. This problem could be avoided if the example of Great Britain were followed, and the results from each polling station combined on a regional level, which would take in more than one district, or even on a Quebec-wide level. The government is prepared to accept any one of these three systems.(67)

Sometimes the problem with a geographic or social division is not the way the votes are counted but a minority’s refusal to participate. For example, the "border poll" of 1973 asked the people of Northern Ireland whether or not they wanted to remain part of the United Kingdom; 98% of those who voted said they did. The results of the referendum drew mixed reviews, however, since it was obvious that many Catholics had refused to participate.

(1)  For a list of national referendums, see "Appendix A Nationwide Referendums," in David Butler and Austin Ranney, eds., Referendums: A Comparative Study of Practice and Theory, American Enterprise Institute for Public Policy Research, Washington, 1978, p. 227-237.

(2)  The referendum procedures for a number of countries are described in Habib Massoud’s "A Survey of the Referendum Processes in Various Countries," Library of Parliament, 1986; therefore, only the relevant provisions are described in the body of this paper.

(3)  A list of referendums and plebiscites held in Canada can be found in J. Patrick Boyer, Lawmaking by the People: Referendums and Plebiscites in Canada, Butterworths, Toronto, 1982, p. 39-41.

(4)  Thomas Foley, in "Reflections on Referendums," The Referendum Device, A Conference sponsored by the American Institute for Public Policy Research and the Hansard Society for Parliamentary Governments, Washington, 1981, Austin Ranney, ed., 1981, p. 10.

(5)  David Butler, "The World Experience," The Referendum Device (1981), p. 75.

(6)  Ronald Butt, in "Reflections on Referendums" (1981), p. 12.

(7)  Robin Day, in "Reflections on Referendums" (1981), p. 9.

(8)  Thomas Foley, "in "Reflections on Referendums" (1981), p. 10-11.

(9)  David Butler, "United Kingdom," in Referendums (1978), p. 218.

(10) "The miscalculations behind the defeats in France in 1969 and Norway in 1972 ended the rule of the men who proposed the referendums. In Denmark, in 1969, the proposal that was defeated was later carried when put forward more tactfully in a referendum in 1971. The Swedish defeat in 1955 was met by the government acting unilaterally in defiance of the referendum decision" (Butler, "The World Experience" (1981), p. 74 (fn. 2).

(11) Butler and Ranney, in Referendums (1978), p. 8.

(12) Anthony King, in "Reflections on Referendums" (1981), p. 14.

(13) Globe and Mail (Toronto), 7 September 1983, p. 11.

(14) Ibid., 26 June 1986, p. A14.

(15) Day, in "Reflections on Referendums" (1981), p. 9.

(16) Boyer (1982), p. 60.

(17) Philip Goodhand, in "Reflections on Referendums" (1981), p. 2.

(18) Vincent Wright, "France," in Referendums (1978), p. 150.

(19) Canadian Unity Information Office, "Understanding Referenda: Six Histories," Ottawa, 1978, p. 25.

(20) Ibid., p. 31.

(21) Sten Sparre Nilson, "Scandinavia," in Referendums (1978), p. 186.

(22) Michael Steed, in "Reflections on Referendums" (1981), p. 8.

(23) Ontario, Quebec, two Atlantic provinces having at least 50% of Atlantic population, and two of the Western provinces having at least 50% of the Western population.

(24) René Lévesque, My Quebec, Methuca Publications, Agincourt, 1979, p. 101 and 109.

(25) Wright (1978), p. 153-4.

(26) Don Butler, "Trudeau Urges Referendum Right in Constitution," Ottawa Citizen, 12 September 1980.

(27) Quebec, Consulting the People of Quebec, 1977, p. 10.

(28) Understanding Referenda (1978), p. 10.

(29) Ibid., p. 29.

(30) Nilson (1978), p. 171-2.

(31) Ibid., p. 172.

(32) Wright (1978), p. 155.

(33) Don Aitkin, "Australia," Referendums (1978), p. 131.

(34) Understanding Referenda (1978), p. 35.

(35) Don Aitkin (1978), p. 131.

(36) David Butler, "Discussion," Referendum Device (1981), p. 102.

(37) Patrick Boyer (1982), p. 12; Butler and Ranney (1978), p. 4.

(38) There is also disagreement as to whether more than one referendum should be termed a "referenda" or "referendums." Butler and Ranney (1978) consulted the editor of the Oxford English Dictionary, who answered that referenda meant a plurality of things or issues to be referred rather than a plurality of votes: "By preferring Referendums as your title, you have the angels of Rome and of the O.E.D. on your side" (p. 4-5, fn. 2).

(39) Chief Electoral Officer of Quebec, Quebec, 1980, p. 6.

(40) Minister of State for Federal-Provincial Relations, "Canada Referendum Bill: An Explanation," 1978, p. 5.

(41) Matthew Garfield, "Referendums and Plebiscites in Canada: Letting the People Choose," 14 Hearsay 29-31 (1989), p. 29.

(42) Edward McWhinney, Constitution-Making, University of Toronto Press, Toronto, 1981, p. 39.

(43) Jurg Steiner, in "Reflections on Referendums" (1981), p. 6.

(44) Jean-François Aubert, "Switzerland," Referendums (1978), p. 45.

(45) Ibid., p. 66.

(46) Consulting the People of Quebec (1977), p. 9. The problems with holding a binding referendum under the present constitution are explored more fully in Jacques Rousseau, "Notes on Changes to the Constitution and Legislation That Would Need to be Enacted to Allow a Referendum," Library of Parliament (1986).

(47) Vernon Bogdenor, in "Reflections on Referendums" (1981), p. 6.

(48) Wright (1978), p. 156.

(49) Nevil Johnson, "Types of Referendums," The Referendum Device (1981), p. 21.

(50) Gregory Mahler, New Dimensions of Canadian Federalism: Canada in a Comparative Perspective, Associated University Presses, London and Toronto, 1987, p. 73.

(51) Vernon Bogdanor, "Referendums and Separatism II," The Referendum Device (1981), p. 143.

(52) Philip Goodhart, "Referendums and Separatism I," The Referendum Device (1981), p. 139.

(53) Anthony King, "Referendums and the European Community," The Referendum Device (1981), p. 118-9.

(54) Tom Dalyell, in "Reflections on Referendums" (1981), p. 17.

(55) Bogdanor (1981), p. 158.

(56) Butler and Ranney (1978), p. 18.

(57) The Referendum Device (1981), Appendix, p. 182.

(58) Quebec, Consulting the People of Quebec (1977), p. 7.

(59) The Referendum Device (1981), Appendix, p. 183.

(60) Austin Ranney, "Regulating the Referendum," The Referendum Device (1981), p. 92.

(61) Butler, "The World Experience" (1981), p. 75-76.

(62) Consulting the People of Quebec (1977), p. 9.

(63) Aitkin (1978), p. 131-2.

(64) My Quebec (1979); Memoirs, McClelland and Stewart, Toronto, 1986, p. 309.

(65) Understanding Referenda (1978), p. 12.

(66) Butler, "Discussion," The Referendum Device (1981), p. 82.

(67) Consulting the People of Quebec (1977), p. 21.