On September 16, 2003, the official opposition introduced a motion “to reaffirm that marriage is and should remain the union of one man and one woman to the exclusion of all others, and that Parliament take all necessary steps within the jurisdiction of the Parliament of Canada to preserve this definition of marriage in Canada.” The motion presented members of the governing Liberal party with a dilemma: most of them had supported an almost identical motion in 1999, but the government’s new policy was that the definition of marriage should be changed to include same-sex unions. The prime minister suggested that those members could vote differently in 2003 in good conscience, because a vote for the motion would be a vote against the Charter of Rights and Freedoms. Why? Because “all necessary steps” might include invoking the notwithstanding clause, and to invoke the notwithstanding clause is to undermine the Charter.

The prime minister’s gambit worked: by the narrowest of margins (the Speaker casting the tie-breaking vote against it) the House of Commons rejected an amendment to remove the reference to “all necessary steps,” leading to the rejection of the main motion by a vote of 137-132. The successful transformation of a motion about the definition of marriage into a de facto referendum on the notwithstanding clause is indicative of a growing convention that it should never be invoked by any legislative body. The rapid decline in the clause’s perceived legitimacy as an instrument of government is one of the key developments in the post-Charter era of rights talk.

The notwithstanding clause in section 33 of the Charter provides that both Parliament and the provincial legislatures may expressly declare that legislation shall operate “notwithstanding” the Charter’s constitutional protection of fundamental freedoms (section 2), legal rights (sections 7 to 14) and equality rights (section 15). Although legislative declarations to this effect automatically expire after five years, they may be renewed indefinitely.

Section 33 was the product of hard political bargaining and compromise. When the First Ministers met on November 2, 1981, for a final round of constitutional negotiations, eight provinces still opposed the federal government’s patriation plan. During the course of those negotiations, Saskatchewan Premier Allan Blakeney argued forcefully for a legislative override provision that would apply to everything in the Charter except language rights, democratic rights and fundamental freedoms. This proposal attracted the attention of other dissenting provinces, and they also pushed for the extension of the override provision to include fundamental freedoms. Sensing the opportunity for agreement, Prime Minister Trudeau indicated his willingness to accept this proposal, subject to the premiers’ agreeing to a five-year time limit on any specific override clause. In what Roy Romanow and two other participants would describe as a “classic example of raw bargaining,” the federal government and nine provincial governments agreed to this provision without which the negotiations might have failed.

The circumstances that produced section 33 inhibited the public development of a coherent theoretical justification for the legislative override. The most extensive public discussion of this provision occurred on November 20, 1981, when then Justice Minister Jean Chrétien introduced the constitutional resolution containing the Charter into the House of Commons. Even then, Chrétien’s remarks on section 33 covered only eleven paragraphs and were aimed primarily at assuring the House that it did not “emasculate” the Charter. The only theoretical point that Chrétien stressed in these remarks was that section 33 would be an infrequently used “safety valve” that would ensure “that legislatures rather than judges would have the final say on important matters of public policy.” Section 33, Chrétien argued, would allow legislatures “to correct absurd situations without going through the difficulty of obtaining constitutional amendments.”

Despite Chrétien’s explanation of the circumstances that might lead to the use of section 33, the first government to invoke the notwithstanding clause did so with quite different purposes in mind. On June 23, 1982, the Quebec National Assembly passed legislation (Bill 62) amending all existing Québec statutes to include a notwithstanding clause. The Quebec government thus used section 33 to make a pre-emptive strike against an agreement to which it had refused to give its assent.

Despite this unexpected use of section 33, most observers still considered it a viable part of the constitution. Nowhere is this more evident than in the Supreme Court’s January 1988 abortion decision. The political context of the decision meant that there was at least the possibility that the Conservative government of the day could find public support to override a judicial declaration of a constitutional right to abortion. This possibility presented the Court with a strategic dilemma. On the one hand, maintaining its Charter-based institutional authority to participate in controversial policy debates meant that the Court could not simply avoid the abortion issue, as it had in 1975. On the other hand, faced with uncertainty about whether judicial nullification of the federal abortion policy would trigger a legislative override, the justices confronted the possibility that the Court might “lose” its first direct confrontation with Parliament over a highly visible policy issue. In the long-term, this outcome could have seriously undermined any future claims the Court might make to constitutional supremacy.

Chief Justice Dickson’s solution to the dilemma was to nullify the existing law while maximizing the set of alternatives to legislative override. He did this by discovering administrative flaws in the operation of abortion law, while making it quite clear that it was “neither necessary nor wise” to “explore the broadest implications” of liberty in analyzing the abortion provisions. One plausible explanation for this cautious approach was the viability of section 33. That viability suffered a significant blow less than one year after the abortion decision.

After the Supreme Court struck down the commercial signs provisions of Quebec’s Bill 101 in December 1988, Robert Bourassa announced his intention to enact new language legislation (Bill 178) that would be insulated from judicial review by the notwithstanding clause. The decision had important consequences. It cost Bourassa three members of his Cabinet; it undermined political support for the Meech Lake Accord outside Quebec, dealing a fatal blow to the chances for its ratification; and it led Prime Minister Brian Mulroney to attack the notwithstanding clause’s legitimacy. Speaking before the House of Commons, the prime minister called section 33 “that major fatal flaw of 1981, which reduces your individual rights and mine.” Section 33, Mulroney continued, “holds rights hostage” and renders the entire constitution suspect. Any constitution, he concluded, “that does not protect the inalienable and imprescriptible individual rights of individual Canadians is not worth the paper it is written on.”

This sequence of events severely undermined the political legitimacy of section 33. Indeed, in March 1998 the Alberta government learned a very hard lesson about the politics of section 33. On March 10, Alberta introduced a bill to compensate victims of provincial eugenic sterilization laws that were in effect from 1929 to 1972. One element of the bill was a provision to prohibit victims from suing for additional compensation, and the government proposed to shield that provision from judicial review through the notwithstanding clause. In purely legal terms there was nothing particularly unusual about this provision: provincial workers’ compensation and nofault automobile insurance regimes also prohibit individual lawsuits as a quid pro quo for a simplified system of guaranteed compensation. On an emotional level, however, wielding the notwithstanding clause against this vulnerable group smacked of meanspiritedness. As a result, one day after introducing the bill, the provincial attorney-general withdrew it under intense political pressure. Alberta Premier Ralph Klein explained the decision to withdraw the bill in the following terms: “It became abundantly clear that to individuals in this country the Charter of Rights and Freedoms is paramount and the use of any tool…to undermine [it] is something that should be used only in very, very rare circumstances.” It thus came as no surprise that the Alberta government summarily dismissed the idea of invoking the notwithstanding clause after the Supreme Court’s decision one month later that its human rights act must be read as providing protection on the basis of sexual orientation.

Is there a justification for section 33 beyond the pragmatic argument that it was necessary to get a deal in 1981? Opposition to any use of the notwithstanding clause to override unacceptable judicial interpretation and application of Charter rights is the product of an historical accident and three conceptual errors. The historical accident is that Canadians experienced a use of section 33 that they found objectionable before the Supreme Court rendered a politically unpopular Charter decision. Unfortunately, the political leaders who took advantage of Quebec’s inclusion of a notwithstanding clause in Bill 178 to condemn section 33 cannot now rely on the legislative override in circumstances where it might be beneficial.

One of the conceptual errors underlying the opposition to the legislative override involves a misunderstanding of the constitutional role of legislatures and courts in liberal constitutional theory. There is nothing in that theory that assigns the task of constitutional interpretation exclusively to courts: legislatures also have a legitimate and important role to play. The second conceptual error stems from a basic misunderstanding of the legislative process as being characterized by the haphazard adoption of measures motivated by majority tyranny. To be sure, legislatures can act both irrationally and arbitrarily; and judicial review provides an important check on these pathologies of legislative behaviour.

Nevertheless, judicial supremacy may be a cure worse than the disease, since courts suffer from their own institutional pathologies when it comes to evaluating complex policy choices. Finally, the opposition to section 33 is fuelled by a basic misunderstanding of the nature of Charter adjudication. Although Charter cases raise fundamental questions about rights or moral principles, the outcome in these cases almost never hinges on the resolution of those questions. In most cases, the dispute is reduced to one about whether the legislature has chosen the least restrictive means for achieving an important policy objective. Yet, even if Charter cases did involve serious disputes about fundamental moral principles on a regular basis, there would be no reason to leave the resolution of these disputes in the exclusive hands of Supreme Court justices. Elevation to a nation’s highest court does not transform any individual into a moral philosopher. Indeed, there is nothing in legal training or in the practice of law that imparts superior judgment in such matters.

As the constitutional scholars Peter Hogg and Kent Roach have separately argued, the notwithstanding clause is a crucial element of the Charter that distinguishes rights-based judicial review in Canada from its American counterpart. Section 33, they argue, facilitates dialogue between courts and legislatures rather than judicial finality or supremacy. However, its effectiveness is only as great as the willingness of legislatures to use it. A constitutional convention against its use would reduce the possibility of interinstitutional dialogue.

The same-sex marriage issue has brought the notwithstanding clause to the top of the political agenda in a significant way. In fact, in March 2000 a private member’s bill— the Marriage Amendment Act— passed in Alberta that defined marriage exclusively as an opposite-sex union and contained a notwithstanding clause to protect that definition fromCharter review. Although probably unconstitutional on federalism grounds, the bill indicates the potential level of legislative resistance to changes in the definition of marriage in at least some quarters.

The prime minister was surely correct in September, therefore, when he concluded that “all necessary means” might include use of the notwithstanding clause. But he was not correct that this would represent an attack against the Charter. The notwithstanding clause is part of the Charter, the Charter would not exist without it, and the Supreme Court has on several occasions recognized the legitimacy of its use.

Whether it should be invoked in the same-sex marriage debate is, of course, a matter of political judgment tempered by a good faith consideration of constitutional obligation and sound public policy. However, neither the original understanding of the Charter, nor its actual text, anticipates blind legislative deference to judicial interpretation. If courts declare that the definition of marriage should change, but Parliament determines that a plausible alternative interpretation of the Charter leads to a different conclusion, then the notwithstanding clause is the constitutionally authorized mechanism for asserting that alternative interpretation.