THE SUPREME COURT
Appeal Nos. 432/13 & 342/2014Denham C.J.
IN THE MATTER OF THE REFERENDUM ON THE PROPOSAL FOR THE AMENDMENT OF THE CONSTITUTION CONTAINED IN THE THIRTY-FIRST AMENDMENT OF THE CONSTITUTION (CHILDREN) BILL HELD ON THE 10TH NOVEMBER 2012
THE MINISTER FOR CHILDREN AND YOUTH AFFAIRS, GOVERNMENT OF IRELAND, IRELAND AND THE ATTORNEY GENERAL
THE MINISTER FOR CHILDREN AND YOUTH AFFAIRS, GOVERNMENT OF IRELAND, IRELAND AND THE ATTORNEY GENERAL
Defendants /RespondentsJudgment of O’Donnell J. delivered on the 24th of April 2015
1 This case requires the Court to address a delicate and important intersection in the separation of powers under the Irish Constitution. The Court must consider the exercise of the executive power of government and the orders which a court can, and should, make when it is has been demonstrated that those powers have been exceeded, and the test the Court should apply. Another issue which arises is the true interpretation of legislation passed in the exercise of the legislative power, and the manner in which the Court should approach it, when it represents a judgment by the legislature on the manner in which alleged wrongs in relation to the exercise of important constitutional rights may be addressed in litigation. It follows from the foregoing that the case also necessarily and unavoidably involves consideration of the function, and limits, of the judicial power under the Constitution.
2 It is an important starting point therefore, to observe, in this case in particular, that the Constitution, by Article 6, states that these fundamental powers of government, executive, legislative and judicial, derive under God “from the people”. That Article continues, in terms particularly relevant to this case, that it is the people’s right to “designate the rulers of the State” and “in final appeal, to decide all questions of national policy, according to the requirements of the common good”. As has been often observed, Article 6 describes government as consisting of three branches but does not itself establish or delineate a separation of powers. The subsequent provisions of the Constitution define and delimit the executive, legislative and judicial powers, and establish a system which looks, and for the most part functions, as a western liberal democracy with obvious similarities to the Westminster model. However the power of the citizen in that model to initiate change and control executive, legislative or judicial power, is extremely limited. But the functional similarity of the system established by the Constitution to that of other liberal democracies makes it more important that we recognise the distinctive emphatic assertion in Article 6, not just of a separation of powers but of popular sovereignty. Set in its historical context, Article 6 is a powerful republican statement that the powers to be exercised in the new polity cannot be traced to any royal concession or to a grant by a legislature considered omnipotent. The sole temporal source of power acknowledged by the Constitution is the people themselves. In that sense the separation of powers in the Irish Constitution while tripartite, has a fourth important, indeed overarching, component.
3 It is entirely consistent therefore with the Constitution’s assertion of popular sovereignty that it should delimit with some care the circumstances in which the people designate the rulers of the State by the exercise of their votes at election time (Article 16) and decide questions of national policy in particular in referenda when amendments are proposed to the Constitution (Articles 46 and 47). The importance of this case flows from the fact that it requires an analysis of both the legislative and constitutional norms governing the conduct of such referenda and the subsequent scrutiny of any result. That in turn necessitates, at least indirectly, a consideration of the law relating to such elections. Each case must be understood against the background of the vital role the people play in the constitutional structure.
4 The fact that it is the people who decide questions of national policy and do so in referenda, is emphasised by a consideration of the terms of Articles 46 and 47 of the Constitution. Those provisions regulate proposals to amend the Constitution. Under Article 46 such proposals are submitted by referendum “to the decision of the people in accordance with the law for the time being in force relating to the Referendum”. Article 47 for its part refers to “Every proposal…submitted by Referendum to the decision of the people”. In a sense, the phrase in Article 46 just quoted encapsulates what is at issue in this case. The decision of the people is the ultimate source of power in this democracy and the Constitution makes it clear that a decision of the people on a matter consigned to them by the Constitution trumps any governmental decision, legislative enactment or judicial determination. But what is to happen if that decision has been achieved after a referendum campaign which has not been conducted in accordance with the law relating to the referendum including, in this case, the law to be found in the Constitution itself as interpreted by the judicial arm of government in performance of its own constitutional function? That question raises fundamental issues about the structure and functioning of a State established with an eloquent assertion of popular sovereignty even if it is a piquant irony that these questions should arise in the context of a referendum in which less than 35% of the people registered to vote, actually did so. For the first 60 years of its existence the issues latent in this case remained almost theoretical, being discussed if at all, in passing, in cases dealing with attempts, uniformly unsuccessful, to restrain the holding of referenda. However, this is the second occasion in 20 years in which the issue has arisen in a most concrete way.
5 While the particular facts of this case relate to the conduct of the referendum campaign in respect of the proposed amendment of the Constitution contained in the Thirty-first Amendment of the Constitution (Children) Bill which was held on the 10th of November 2012 (and which I will refer to hereafter as “the Children Referendum” since that was the manner in which the referendum was described by the Referendum Commission established under the Referendum Act 1998 (“the 1998 Act”)), it is necessary to sketch first a little of the legal and historical background, before turning to the details of this case. To an unusual degree, the arguments in this case can only be understood against the backdrop of what occurred in the context of the contentious campaign in respect of the Fifteenth Amendment of the Constitution Act 1995 providing for the dissolution of marriage in certain specified circumstances, adopted on the 17th of June 1996 and known as the “Divorce Referendum”, and the legal challenges that arose both during and after the campaign.
6 Prior to the Divorce Referendum there had of course been a series of applications for injunctions in respect of different referenda on different grounds but there had been no challenge to a referendum result. One argument which was being increasingly ventilated was that the expenditure by the government of public finances in support of a proposal to amend the Constitution was itself unconstitutional. That claim was initially rejected, first in the context of an interlocutory application in McKenna v. An Taoiseach (No.1)  2 I.R. 1, and subsequently after a full hearing in the High Court. But in McKenna v. An Taoiseach (No.2)  2 I.R. 10 (“McKenna”), the Supreme Court, by a majority of four to one, upheld the contention and found that the public funding by the government of a campaign in support of a Yes vote in the divorce referendum campaign was unconstitutional. This decision was delivered just one week before the referendum was held. Understandably the decision attracted considerable publicity. In the event 818,842 citizens voted in favour of the proposal and 809,728 against (a margin of less than 0.56%). The provisional referendum certificate was immediately challenged on the basis of the unconstitutionality identified in the McKenna judgment. Those proceedings were Hanafin v. The Minister for Environment & Ors and the judgments are reported at  2 I.R. 321 (“Hanafin”).
7 The Hanafin proceedings were commenced under the then recently enacted Referendum Act of 1994 (“the 1994 Act”). That was a logical and probably prudent step, since s.42 of that Act provided that the validity of a provisional referendum certificate may only be questioned by petition brought in accordance with that Act. However, there were some difficulties in using the 1994 Act to address the type of claim which arose in the Hanafin case, and which recurred here. The 1994 Act was focused on essentially mechanical matters capable of affecting individual votes in particular constituencies. It referred to electoral offences, mistakes and irregularities “in the conduct of the referendum”. These matters were likely to be individual and local. Consistent with that approach, the Act contemplated that in the event that such matters were established, the Court could order the recounting of all or some of the votes in “a constituency” or the retaking of the referendum in “a constituency”. The 1994 Act was itself an updating of the provisions contained in the first Referendum Act of 1942 (“the 1942 Act”) and which had been amended once in 1992 by the provisions of s.68 of the Electoral Act 1992 (“the 1992 Act”). Indeed the Referendum Act of 1994 can be seen as a consequential measure bringing the law relating to referenda in to line with the electoral acts which had been most recently and comprehensively revised in the 1992 Act, rather than as a separate provision addressed to the particular issues arising in a referendum. In particular, the 1994 Act does not address the type of systemic nationwide problems which can arise in referenda, and in particular those which did arise in the Hanafin case and again here.
8 In Hanafin a Divisional Court of the High Court (Murphy, Lynch and Barr JJ.) heard considerable evidence on behalf of the petitioner but dismissed the case at the close of the petitioner’s case. The Court divided somewhat as to its reasoning. The majority (Murphy and Lynch JJ.) held that the words “conduct of the referendum” (s.43 of the 1994 Act) were only capable of applying to those aspects of the organisation and taking of the poll and other matters which were entrusted to the returning officer whose duty under the statute it was to “conduct” the referendum. The majority considered that the government’s action in funding a publicity campaign was not itself capable of coming within the meaning of the term “obstruction of or interference with or other hindrance to the conduct of the referendum” (s.43 of the 1994 Act), on an ordinary reading of the text and that no broader interpretation was necessitated since the wrongdoing identified in McKenna was capable of being addressed and remedied, without setting aside the result of a referendum, most obviously by a declaration of unconstitutionality such as that granted in McKenna, which had been immediately obeyed. Barr J. took a different view on this point holding that a broader interpretation had to be given to the words so that the government support of an advertising campaign on behalf of one side of a referendum must be held to be capable of being an interference, obstruction, hindrance or irregularity in the conduct of the referendum. In any event, Murphy and Lynch JJ. considered that even if the governmental conduct could be an interference with the conduct of the referendum within the meaning of the Act, the evidence was not such as to establish that the advertising campaign had “materially affected the result of the referendum”. Barr J. for his part was prepared to hold that the evidence did not establish on a prima facie basis that the will of the people had not been properly ascertained and freely expressed in accordance with the law. Barr J. also expressed the view (p. 375) that a referendum result could only be set aside if it was established beyond reasonable doubt that the result could not be seen as reflecting the will of the people and that the onus on a petitioner was very heavy. Murphy and Lynch JJ. did not consider it necessary to determine what standard of proof applied, but held that the petition had failed to satisfy even the lower of the two standards.
9 The petitioner appealed to the Supreme Court and succeeded on nearly all issues save the final and determinative one. The Court, having unanimously rejected the contention that an appeal did not lie, also held unanimously, that a broad interpretation should be applied to the Act and in particular the concept of “irregularity in the conduct of the referendum” (s.43) so that the government’s actions in supporting a Yes vote by the expenditure of public funds was capable of being such an irregularity or interference with the conduct of the referendum. The Court also rejected the contention that the petitioner was required to establish a case beyond a reasonable doubt. Furthermore, some members of the Court expressed disagreement with the approach which the Divisional Court had taken in disposing of the case at the close of the petitioner’s case. Nevertheless all members of the Court rejected the appeal on the grounds that that High Court was entitled to come to the conclusion which it had upon the evidence, that it had not been established that the result of the referendum had been materially affected by the wrongdoing. That this was, for some members of the Court at least, a close call, is illustrated by the judgment of O’Flaherty J. who concluded that the verdict of the people enjoyed a presumption of constitutionality which had not been displaced, a point he considered “as narrow as the verdict in the disputed referendum” (p. 438).
10 Subsequent to the decision in McKenna, the Oireachtas enacted the Referendum Act 1998 which permitted the establishment of a Referendum Commission charged with providing neutral information to the electorate on the content of any proposed amendment to the Constitution, a function successive Commissions have discharged during subsequent referenda. In 2012 however, in the context of the Fiscal Compact Referendum, it was apparently decided to take a different approach. The decision in McKenna had made it clear that the prohibition on governmental action related to the expenditure of public funds and resources in support of one side of the referendum. The government however was entitled to campaign, and indeed to provide information, although this was said in a context where there was no independent body discharging that function. The good sense of the establishment of an independent commission is obvious on a moment’s reflection. It is much easier to identify a category of neutral “information “ in theory, than it is to draw that line in practice, particularly when the person providing the information is the person proposing the amendment, and has concluded that it is necessary. Relying on the dictum that the provision of information by the government was not itself a breach of McKenna, and, it appears, the fact that a number of surveys seemed to show a desire on the part of the electorate for further information, it was, apparently, decided to mount a governmental information campaign in parallel to that of the Referendum Commission. Although that decision was criticised at the time, it was not the subject of any legal challenge during that campaign.
11 When in due course the government advanced a proposal to amend the Constitution, and in particular Articles 41 and 42, in the Children Referendum, it appears it was decided to conduct a further information campaign. This campaign was challenged by Mark McCrystal in proceedings commenced during the course of the campaign. Although the challenge was rejected in the High Court, this Court upheld the challenge and issued a short ruling explaining its decision on the 14th of November 2012, two days before the proposed referendum. The text of the ruling is set out at paragraph 25 of the High Court Judgment on the petition in this case and the full Supreme Court Judgments were delivered on the 11th of December 2012: Mark McCrystal v. The Minister for Children and Youth Affairs & Ors  IESC 53 (“McCrystal”)).
12 It is necessary to observe at this point that the decision in McKenna has from time to time been criticised and dissatisfaction expressed with it. Some examples are recorded in Noel Whelan “Changing The Rules of the Political Game” in O’Callaghan Barniville and Ruane (eds.) Law and Government - A Tribute To Rory Brady (Dublin; Round Hall; 2014), and Ruane, “Reflections on Procedural Rights in Constitutional Referenda” (2012) 48 Irish Jurist 1, which also sets out and discusses the legal arguments. Indeed Whelan states that the decision to conduct the public information campaign in the context of the Fiscal Compact Treaty Referendum sought to test the boundaries of the McKenna decision. However, despite these matters, it is important to note that no attempt was made in McCrystal to argue that McKenna was wrongly decided, or to invite this Court to depart from it or qualify its conclusions. I do not mention these matters to invite such a challenge or necessarily cast doubt on that decision. It is important however to record and repeat the fact that McCrystal was an application, rather than a consideration and an endorsement, of McKenna.
13 The evidence in McCrystal was revisited in this case with the benefit of further discovery, and showed, in my view, a very clear breach of the McKenna principles. Nor was this something which only emerged after the fact, and in the course of litigation. One particularly telling piece of evidence was that during the course of the campaign, RTÉ had refused to carry certain advertisements as part of the information campaign because it considered that they could be a breach of the obligation on RTÉ of fairness and balance.
14 In the event, on polling day on the 14th of November 2012 some 33.5% of the electorate voted in the referendum and divided 615,731 to 445,863 (58%/42%) in favour of the proposal. The provisional referendum certificate was then challenged by petition initiated by Ms Joanna Jordan. She also commenced parallel plenary proceedings contending that if the Act was to be interpreted as imposing upon her the burden of demonstrating that the outcome of the referendum was altered by virtue of the unconstitutional information campaign, then she contended in the light of Hanafin, that such a burden was impossible to discharge, and that the Act if so interpreted would be unconstitutional.
15 Section 42 of the 1994 Act requires that a provisional referendum certificate may only be challenged by petition, and furthermore, that a petitioner must obtain the leave of the High Court to commence any such proceedings. Under section 42 in order to grant leave, the Court must be satisfied that there is prima facie evidence of grounds under s.43 (eg. an electoral offence, obstruction, interference, mistake etc) and furthermore that the matter is “such as to affect materially the result of the referendum as a whole” (s. 42(b)). In this case the parties agreed that the leave application should be heard at the same time as the substantive application to avoid the necessity of two lengthy hearings, and furthermore agreed that the evidence in the petition could be taken as the evidence in the constitutional challenge. This was probably a sensible course in this case since there is a high degree of overlap in the legal arguments on the questions of leave, the petition, and the constitutional arguments. An unconstitutionality had already been established in McCrystal, and Hanafin had decided that such unconstitutionality was capable of being an interference with the conduct of a referendum. However the leave stage created by the legislation is an important threshold step reflecting the fact that a challenge which seeks to reverse what is prima facie the will of the people expressed in a referendum, is a very significant step not least because the grant of leave prevents the amendment from taking effect. In most if not all challenges to the result of a referendum, it would in my view be important that a separate leave application be held, and rigorously assessed. It was however appropriate to hear the petition and plenary proceedings together since in my view the interpretation issue in particular could not properly be resolved without considering the constitutional arguments, since a major argument was that the Act should be construed in a particular way to comply with the Constitution. Hanafin itself is an illustration of this approach. It was held in that case that a broad interpretation was required to be given to the 1994 Act to accord with the Constitution to permit the issues created by the recognition in McKenna, of constitutional wrongdoing with the potential to affect the referendum campaign and process, to be addressed within the petition procedure.
16 The appropriate sequence therefore would seem to be to consider what the statute appears to mean applying normal canons of interpretation; to consider then if such an interpretation would render the Act unconstitutional; and if so, to consider if a broader or different interpretation is possible which would be consistent with the Constitution; if a constitutional interpretation is possible it must be adopted, but if no other interpretation is possible the Act must be declared unconstitutional. It is certainly easier to do all this in one reasoning process rather than to artificially separate the question of interpretation from the question of constitutional validity. Indeed, as the Court observed at the outset of this case, the question could be also approached from the other direction: what does the Constitution require (or permit) when there has been an unlawful (and indeed unconstitutional) interference with a referendum? Is the 1994 Act consistent with that conclusion? In either case does the Constitution require that the provisional referendum result be invalidated and a new poll directed?
The Evidence in the High Court
17 The High Court heard extensive evidence, which was carefully tested by searching and detailed cross-examination. I gratefully adopt the detailed description of the evidence contained in the judgment of my colleague Mr Justice MacMenamin. Among the useful points to emerge was agreement that the Children Referendum could be considered a “low salience issue” (in contrast for example to the Divorce Referendum). This is an issue which is obscure to the voters and about which they know little in advance. As a result it was argued that members of the public would be more dependent on, and likely to be influenced by, official communications, particularly it might be said, those presenting themselves as neutral and informative. The evidence as to the degree of preparation for the referendum and the expenditure involved, and the extent to which the wording of the amendment and its presentation was the subject of research and public surveys in advance, was also instructive. It is useful to remember that the McKenna constraints have been considered only to apply to the period between the passage of the Bill to amend the Constitution and the poll date, which in recent times has been quite a short period. In this case the Bill to amend the Constitution passed both Houses of the Oireachtas on the 3rd of October 2012, and on the 8th of October, polling day was set for the 10th of November. The evidence of the government’s spending on its information campaign (€1.1million) is also significant. The details are set out at paragraph 21 of the High Court Judgment, but the bulk of these funds were spent on polling, advertising and design and delivery of the booklet, all of which occurred in a short time period. This is a substantial amount of (public) money in its own right, but the potential significance of the spend in a short period of time in the context of a low salience referendum campaign is illustrated by the fact that after the monies allocated to the Referendum Commission, this was by far the largest single spend on the campaign and comfortably exceeded the combined spend of all the political parties and voluntary groups campaigning in support of the proposed amendment. Another useful point of comparison is that the two political parties comprising the government spent a total of €200,000 on their own behalf on the referendum campaign but the government itself spent €1.1 million on its information campaign. This was all the more significant in the context of the Children Referendum since the No side had virtually no resources with which to conduct any advertising or publicity campaigns. Whatever view is taken of the McKenna decision, it cannot be gainsaid that this is a very significant expenditure both in its own terms and in the context of this referendum campaign.
The High Court Judgment on the Petition
18 The judgment of the High Court ( IEHC 458) first considered the legal test to be applied. Addressing itself to Hanafin, it concluded that the onus was on the petitioner on the balance of probabilities to establish a material effect on the result of the referendum as a whole. In particular, the High Court concluded at paragraph 60 that having regard to the importance of the values identified in Hanafin:
This test it will be noted, involves the proposition that in order to succeed, the petitioner had to show that the referendum would have been defeated were it not for the unconstitutional conduct identified in McCrystal.
“…anything less than evidence which, on the balance of probability, demonstrates that the unconstitutional behaviour in this case affected the outcome of the Referendum and had such an impact on voters as to produce a majority in favour of the proposal, means that the petitioner must fail in her petition”. (emphasis added)
19 The first phase of the case involved evidence from individual voters. Four voters were called, and the evidence was that one contended that he had changed his vote from a No to a Yes having read the government booklet, and one individual who was going to vote No abstained. There was other evidence from campaigners. The trial judge’s conclusion from this evidence (paragraph 73) was that while it was useful and honest it was also speculative and subjective. He considered that it was clear that:
20 The Court also considered evidence from political scientists, which attempted to quantify the impact of the governmental campaign. Much of this evidence was given on behalf of the petitioner by Dr Michael Bruter, a reader in political science specialising in electoral psychology at the London School of Economics and Political Science. He expressed the view that in the context of a low salience referendum, the perceived nature of a source was a major determinant of how influential the information would be. Evidence was also given by Professor Paul Whiteley, Professor of Governance at the University of Essex (recounted from para. 94 of the High Court Judgment). He considered that information offered from the department of state having expertise in the area would be particularly influential in the context of a low salience campaign. He acknowledged that he could not give a precise assessment of the extent of the influence the information had on the decision of voters because “the data was simply not available”. Notwithstanding however the absence of a “detailed study capable of giving a precise estimate of the extent of the influence of the general campaign” he considered he could form a judgment of a rough kind about what the effects were. While not possible to determine the “percentage affect” of the booklet he was satisfied to conclude that the information “may very well have been quite influential”.
“…without something more tangible and objective, this evidence would be entirely insufficient to justify the granting of leave to present a petition or to establish that the overall Referendum result had been materially affected by the government information campaign”.
21 This evidence was contested on behalf of the respondents by Professor Michael Marsh, Professor of Comparative Political Behaviour at Trinity College, Dublin. Professor Marsh’s principal area of expertise is electoral behaviour including how voters make their choices in elections and referenda. A central part of his evidence, as set out at paragraph 91 of the High Court judgment was that it was:
He was emphatic that “it was not possible to come to a clear view that the unconstitutional expenditure of money by the government resulted in victory for the yes side that would not otherwise have occurred”. His views were summarised at paragraph 92:
“…not possible to assess definitively what factors determined the outcome of the children referendum or to determine whether any individual factors such as the material published by the department had an effect on the level of the ‘yes’ or ‘no’ vote.”
Indeed, even assuming a study could have been carried out at the time of the conduct of the campaign, Professor Marsh acknowledged that there would still have been problems with the approach and that “[a] causal sequence can be very difficult to establish” (para. 93 of the High Court Judgment). To an extent Professors Marsh ‘s evidence was directed towards the negative and somewhat unsatisfactory objective of showing that it was not possible to prove or disprove decisive effect.
“He was strongly of the view that without a properly designed study, implemented at the time of the conduct of the campaign itself, it was not possible to assess what factors had a particular effect on the outcome.”
22 The judge set out his conclusions on the evidence at this point in the case. He observed that Professor Whiteley had accepted that all the experts were agreed that “the supporting data was not available to enable a study to be carried out to determine exactly the extent of the influence of the government’s publication” (para. 95 of the High Court Judgment). Accordingly, the High Court Judge expressed its view at paragraph 96 that the evidence of the political scientists was “completely insufficient to establish that the votes cast in the Referendum had been materially affected by the information distributed by the government (including the booklet)”. He considered that Dr Bruter’s evidence was the subject of reasonable criticism by Professor Marsh and that both Professor Marsh and Professor Whiteley had agreed that there was “insufficient data to enable the type of analysis that might make it possible to determine what influence, if any, the information had on the course of the campaign and to what precise level”.
23 After the evidence was given, but before the case was determined, it emerged that the Referendum Commission had carried out its own post referendum survey in a sensible and praiseworthy attempt to measure the impact and effectiveness of its own information campaign and the relative impact of various methods used for gaining the attention of voters. Some of the questions asked in that survey made reference to the governmental information campaign and in particular the government booklet. The raw data from the survey was still available, and accordingly, Dr Bruter engaged in what the trial judge described, accurately in my view, as the interesting and imaginative attempt to derive from that data evidence of the impact of the governmental campaign. This led to the point where he contended that there was evidence of a statistically significant association between receipt of the government booklet (or perhaps more accurately recall of receipt of the booklet) and voting Yes in the referendum. He sought to quantify this connection at 9.3%.
24 The figure of 9.3% appears to be derived from the raw data contained in the Behaviour and Attitudes survey conducted on behalf of the Referendum Commission which recorded that of those who received the Referendum Commission Guide, 62% said they voted Yes and 38% No: 64% of those who received the government booklet voted Yes and 36% No, and 65% of those who received both guides voted Yes and 35% No. Furthermore, 64% of those who received neither guide voted Yes and 36% voted No. It is clear therefore that the survey recorded a higher percentage of Yes votes among those who had obtained the government guide (64%), than among the population generally (it will be recalled that 58% of voters supported the amendment). As a result Dr Bruter said initially “[o]verall, receiving the booklet made people over 9.3% more likely to vote yes”. Put more precisely, he agreed this meant that he considered that there was a “statistically significant relationship between saying you received the Government Booklet and saying that you voted yes in the referendum” (paras. 100 – 102 of the High Court Judgment).
25 There are, it should be said, of number of valid issues which may be taken with this exercise and the extrapolation from the survey. Since for example, 64% of respondents identified themselves as Yes voters, then either the survey was not representative of the population, or there was an element of over reporting which is not uncommon in opinion polls. Furthermore the correlation, if it existed, was perhaps at best between a recollection of receipt of the booklet and of voting Yes, rather than between the contents of the booklet and the individual vote. However 9.3% was in some ways, and to borrow a phrase, a “magic number” since it allowed the petitioner to argue that it showed, even to the most demanding standard, that the result of the referendum had indeed been altered by the government information campaign which had been found unconstitutional in McCrystal. I understood this to be because on one view of these figures, if accepted, it could be said that this difference was sufficient to carry the referendum (I am not convinced that any of the steps in this calculation are watertight even as a matter of mathematics, but for reasons which will be addressed later it is not necessary to resolve that matter here). A 9.3% difference was on any view, a figure which, if accepted, provided some concrete assessment of the impact of the information campaign and allowed the petitioner to argue that the impact was indeed decisive.
26 As the trial judge found, the arguments in this matter were detailed and technical and of rather narrow focus. There was further sophisticated evidence given as to the methodology employed, which for example discussed whether weighted data should have been used, and as to whether the results were reliable. Again, the evidence of the respective experts was subjected on both sides to well prepared and searching cross-examination. In the event, the trial judge accepted the evidence of Professor Marsh and the criticisms he advanced, and accordingly did not accept that the petitioner’s evidence had demonstrated a material effect on the outcome in the sense of demonstrating that without a government information campaign the amendment would have been rejected. Although strenuous efforts were made to revive these controversies in this Court, this was a conclusion of fact, given after considering expert evidence advanced in the High Court, and was a conclusion to which in my view, the trial judge was amply entitled to come. It must accordingly be taken as forming the factual substratum upon which the legal issues in this appeal are to be debated. This conclusion was however fatal to the petition, since it meant that on the test which the Court set itself, it was satisfied that the acknowledged unconstitutionality established in McCrystal had not materially affected the outcome of the referendum as a whole.
The Constitutional Argument on the Interpretation Issue
27 The next step was to consider if the test applied was consistent with the Constitution. In this respect the trial judge again accepted the respondents’ arguments, which placed some reliance on an unreported judgment of the Supreme Court (Henchy J: Budd and Griffin JJ. concurring) which had rejected an appeal pursuant to the Electoral Act 1923 (“the 1923 Act”) in respect of a challenge to the election result in East Mayo in the 1973 election: Dillon-Leetch v. Calleary & Ors (unreported; Supreme Court; 31st July 1974) (“Dillon-Leetch”). In that case there had been an allegation that the secrecy of the ballot had been breached – at least in theory – because of a failure to maintain supervision of the ballot boxes at break times. Henchy J. cited Woodward v. Sarsons and Sadler (1874-75) L.R. 10 C.P. 733 (“Woodward v. Sarsons”) at p. 750,
Henchy J. continued, “this interpretation does not appear to have been dissented from in any subsequent case. The most recent approval of it is to be found in Morgan & Ors v. Simpson & Anor  1 All E.R. 241”. Henchy J. considered therefore that what he described as an “electorally ineffective” breach of a principle (albeit in that case the constitutional principle of secrecy) should not, without more, lead to the invalidation of the result. This he considered was a conclusion based on the provisions of s.39 of the 1923 Act which, in terms essentially identical to section 13 of the Ballot Act 1872 (“the 1872 Act”), provided a saver for elections in which, while there may have been a breach of a statutory provision, were conducted in accordance with the principles of the Act, and where such non-compliance did not affect the result of the elections.
“It is said that section 13 [of the Ballot act 1872] though it is in a negative form, assumes as an affirmative proposition that a non-compliance with the rules, or any mistake in the use of the form would render an election invalid, unless it appeared that the election was conducted in accordance with the principles laid down in the body of the act and that such non-compliance or mistake did not affect the result of the election. If this proposition is closely examined, it will be found to be equivalent to this, that the nonobservance of the rules or form must be so great as to amount to a conducting of the election a manner contrary to the principle of an election by ballot, and must be so great as to satisfy the tribunal that it did affect or might have affected the majority of the voters, or, in other words, the result of the election. It therefore is, as has been said, an enactment ex abundante cautela declaring that to be the law applicable to elections under the Ballot Act which would have been the law to be applied if this section had not existed.”
28 Importantly for present purposes, Henchy J. also considered that such an outcome was consistent with and required by the Constitution itself:
29 Although the dispute between the political scientists must be taken to have been resolved by the determination of the trial judge, the general evidence was relevant in a number of different ways to the legal arguments of both sides. Thus the petitioner argued first, as already discussed, that the evidence passed the test set by Hanafin however high and demanding that might be. Second, she argued that if the evidence did not meet the standards set by the High Court Judge that that was a misinterpretation of Hanafin so as to create a virtual impossibility of proof for a petitioner. Third, she argued that if indeed the test applied by the High Court followed correctly from Hanafin, then Hanafin was wrong in setting such a test which required proof to a point which it was argued, was acknowledged to be virtually or actually impossible. Finally, if the test applied followed from the 1994 Act, then that Act was unconstitutional in requiring an impossible standard. The respondents for their part sought to escape the pincer movement implicit in the petitioner’s arguments by arguing at one and the same time that the evidence so extensively debated before the High Court did not meet the required standard, but nevertheless the fact that evidence of that nature could be adduced showed that it was in principle possible to provide evidence of material effect on the result of a referendum, and accordingly, there was no true impossibility of proof. The failure of proof, it was argued, was specific to this case, but not inevitable. The petitioner’s arguments, it was argued, confused difficulty with impossibility. There was nothing wrong with a high and demanding standard being set before the vote of the people should be set aside. Indeed, that was consistent with the decision in Hanafin and moreover required by the Constitution.
“If an election is upheld because although a breach of secrecy has or may have occurred it could not have affected the result the court will be overlooking a breach of the constitutionally recognised principle of the secrecy of the ballot in elections for Dáil Eireann: see Article 16 s.1 ss.4 of the Constitution. This, however, is because the courts will not allow an electorally ineffective breach of that principle to be used to set aside the correctly exercised constitutionally right of the rest of the citizens forming the electorate in the constituency to elect their representatives in Dáil Eireann: see Article 16, s.1 ss.2 of the Constitution. To hold otherwise would be as much an inversion of constitutional priorities as to hold that an election in a particular constituency should be set aside merely because a voter or particular voters voted more than once (in breach of the prohibition on Article 16 s.1 ss.4 against so doing) but not so as to have effected the result of the election. In such cases, where the court is confronted with the inescapable necessity of giving effect to one or other of two constitutional principles or rights, it should opt for the one that is more essential to the smooth and harmonious operation of the Constitution in the light of its basic assumptions and primary aims:”
30 The High Court, in its judgment on the petition delivered on the 18th of October 2013 ( IEHC 458), held that there was indeed a prima facie case pursuant to s.42 of the 1994 Act and accordingly granted leave to the plaintiff to present the petition, but dismissed the petition on the merits. On the 20th of June 2014, having heard further argument on the constitutional issue, the trial judge delivered a further comprehensive judgment dismissing that claim ( IEHC 327). It is no disrespect to the meticulous judgments delivered by the trial judge to observe that in some respects the legal arguments were narrowed and simplified in that case by the fact that the High Court was bound by the decision in Hanafin. In this Court however, the appellant, as she was entitled to do, has argued that Hanafin was wrongly decided and should not be followed, and the arguments have ranged further.
31 On this appeal the petitioner/appellant makes a number of interrelated arguments that have in turn raised additional issues of some complexity. It is argued that on a true construction of the 1994 Act, the petitioner was not required to demonstrate material effect on the result of the referendum as a whole. As this argument was developed it was acknowledged that it had not been made, at least in the same terms, in the High Court. Accordingly, consideration of this argument also involves consideration of the circumstances in which the Court can entertain an argument not made in the High Court. The petitioner also argues that either the test of material effect on the result of the outcome should not be interpreted to mean a different result would have ensued, or that on proof of an unconstitutional interference with a referendum the onus should shift to the State respondent (either as a matter of law or because it was essentially a matter within the power of the State to prove) to demonstrate that there had been no material effect on the outcome. It is also argued that the trial should have been conducted as an inquiry with no onus of proof on the petitioner.
32 All of these arguments are based to some extent on the proposition that in the light of the secrecy of the ballot, it is effectively impossible, as a matter of reality, to prove one way or another whether there had been an impact on votes to the degree where it could be demonstrated that the votes in the referendum would have been different or conversely, that in the absence of the government’s unconstitutionality the result would have been the same. Thus, whoever bore the burden of proof of material effect or its absence, would inevitably lose. These arguments have echoes of arguments made in Hanafin and the decision in that case. The petitioner accordingly also argues that the trial judge misunderstood or misapplied Hanafin, or that if he had correctly applied the principles in Hanafin, then the decision was wrong and should not be followed. To that extent the argument also therefore requires the Court to address the circumstances in which it could depart from an earlier decision of the Supreme Court in relation to constitutional issues.
The Interpretation Issue
33 Counsel for the petitioner put forward an intricate argument as to the true interpretation of the 1994 Act in the light of the common law of elections and the statutory developments since the famous Ballot Act of 1872 which introduced the secret ballot. Counsel frankly admitted that the argument had not been put in this way in the High Court, but I am satisfied that it is proper for this Court to address it. It is closely related to issues of interpretation which were argued in the High Court. It is a legal argument as to the interpretation of the Act and is not dependent on evidence, and has no implications for the manner in which the High Court hearing proceeded. Further, the constitutional challenge requires that the Act be properly interpreted before consideration be given to the question of whether or not the Act as so interpreted is consistent with the Constitution. It would be unsatisfactory, to say the least, if the Act was found to be unconstitutional on the basis of an erroneous interpretation, and equally undesirable if the petitioner’s claim on a matter of great importance not only to her, but to the State generally, were to be held to fail and a potentially unconstitutional amendment permitted to stand in the basic law of the State because a legal argument of some force, was not permitted to be made. This accordingly, is one of those occasions for the application of the observation attributed to Judge Learned Hand that inspiration is not so common place a virtue that it is to be rejected merely because it comes late. In accordance with the approach set out in Lough Swilly Shellfish Growers Cooperative Society Ltd v. Bradley & Ivers  IESC 16, I would be prepared to permit this argument to be made and considered.
34 It is necessary to state in some detail the petitioner’s argument in this regard, not only to understand and address it, but also because an understanding of the electoral law, its history and application is useful in resolving the larger constitutional questions in this case since, the right to vote in parliamentary elections enjoys the same constitutional status as the right to vote in a referendum. In the language of Article 6, it is the choice by the people of the rulers of the State under Article 6 just as voting in a referendum can be seen as the people deciding in ultimate appeal all issues of national policy. Furthermore, as already observed, the provisions of the Referendum Act of 1994 are largely patterned on the Electoral Acts which themselves must be understood against the background of the pre-existing electoral law. There is thus a close parallel between both the statutory provisions involved and the constitutional values engaged in each case.
35 The starting point for counsel’s argument was an attempt to undermine the authority of the decision of the Supreme Court in Dillon-Leetch. As already observed the judgment in that case cited the dictum from Woodward v. Sarsons and further observed that the dictum did not appear to have been dissented from and had itself been recently approved in the decision of the High Court in Morgan & Ors v. Simpson & Anor  1 All E.R. 241 (“Morgan v. Simpson”). Dillon-Leetch addressed the provisions of s.9 of the Electoral Act 1923 which was in the following terms:
As Henchy J. observed, this saving provision can be traced to s.13 of the Ballot Act 1872. It is also worthwhile at this point to observe that that section is in essentially similar terms to the provisions contained in s.48(2) of the Referendum Act 1994 (itself drawn, it appears, from s.9 of the Electoral Act 1992). The formulation of this saving clause has caused difficulties of interpretation since it is in negative terms (“no election shall be declared invalid” in the 1923 Act (s. 39) and “the Court shall not order a referendum to be taken again” in the 1994 Act (s. 48(2)) and does not, except by implication, contain any positive statement as to the circumstances in which election should be declared invalid. The fact that the provision also contains two conditions, conduct in accordance with the principles of the Act and no affect on the result, makes the interpretation more difficult, particularly when it is sought to transpose the negative statement into a positive.
“No election shall be declared invalid by reason of a non compliance with the rules … if it appears to the tribunal having cognisance of the question that the election was conducted in accordance with the principles laid down in the body of its act, and that such non compliance … did not affect the result of the election.”
36 As it happens, this difficulty of interpretation manifested itself in the Court of Appeal decision in Morgan v. Simpson ( 3 W.L.R. 517). The Court of Appeal of England and Wales emphatically overturned the decision of the High Court which had been referred to, and to an extent relied upon, in Dillon-Leetch. In doing so the Court went further, expressly disapproved of the dictum of Lord Coleridge C.J. in Woodward v. Sarsons also referred to in Dillon-Leetch, and set out at paragraph 27 above, Lord Justice Lawton referred to the dictum as not making sense (page 171) and Lord Denning M.R. agreed (page 163).
37 At first sight this development (of which the Supreme Court was unaware in Dillon-Leetch), might appear to significantly undermine the authority of that judgment, but on closer analysis it is apparent that this is not so. The error at first instance in Morgan v. Simpson was to read the dictum of Lord Coleridge C.J. in Woodward v. Sarsons as meaning that an election could only be invalidated where both conditions had been satisfied, i.e. there was a material effect on the outcome and that the election had not been conducted in accordance with the principles set out in the Act. The result of that interpretation was that the court of first instance was prepared to uphold an election where irregularities complained of had affected the outcome in the sense that when corrected, a different candidate would have won. The Court nevertheless refused to set aside the election because it had not also been shown that the election was not conducted in accordance with the general principles of the Act. That issue did not arise in Dillon-Leetch which clearly came within the terms of s.39: the election had been conducted in accordance with the principles of the legislation and the irregularity had not had a material effect on the outcome of that. Accordingly, Henchy J. did not rely on the extract from Woodward v. Sarsons in the sense in which it was applied at first instance in Morgan v. Simpson and which the Court of Appeal found to be erroneous. More importantly the significance of Dillon-Leetch for this case is related to the constitutional rather than the statutory argument. Thus, whether the interpretation of s.39 was correct or not as a matter of statutory interpretation was beside the point. The burden of the dictum in Dillon-Leetch was that a requirement of proof of material effect on the outcome before an election could be declared void, was consistent with, and arguably required by, the Constitution. That conclusion was not, and could not have been affected by the dictum of the Court of Appeal of England and Wales in Morgan v. Simpson.
38 However, the decision of the Court of Appeal of England and Wales in Morgan v. Simpson is also important for a more general argument which counsel sought to advance. The judgment of Lord Denning M.R. contains a very useful and lucid account of the history of the Ballot Act 1872 and in particular s.13. He explains that prior to 1868 any election disputes were determined by parliament which had the power to determine the result, and if necessary to direct a fresh election. The approach taken by parliament was discretionary, and less formal than that which might be taken by a court. Indeed, at that time, election at common law was open and not by secret ballot, and in the words of Lord Denning M.R:
In 1868 election disputes were for the first time to be determined by the courts under the Parliamentary Elections Act of that year. Then in 1872 the Ballot Act was enacted which revolutionised the system of voting at elections and provided for voting by secret ballot and prescribed rules and set out forms of ballot papers. The degree to which elections were now prescribed by detailed statutory rules meant that there was a risk that elections could be too easily set aside for breach of those rules. It was therefore thought necessary to have a saver in the form of s.13 which has been reproduced in subsequent legislation ever since. To this extent it appears the statutory law followed the practice in parliament prior to 1868 in that parliament exercised a degree of discretion on a petition. To that extent it was perhaps correct to say that the section was enacted ex abundante cautela. However the true principle was set out by Lord Denning M.R. at page 164:
“…was disgraced by abuses of every kind, especially at parliamentary elections. Bribery, corruption, treating, personation, were rampant. These were not investigated by the courts of law. They were the subject of petition to parliament itself. Often members were unseated and elections declared invalid. If you should wish to know what happened, you will find it in Power, Rodwell & Dew Reports of Controverted Elections (1848-1853) and in Charles Dickens’ account of the election at Eatanswill [in the Pickwick Papers, Chapter 13]”. (p. 162)
39 Counsel for the appellant here pointed out that there were many statements to the same effect in earlier decisions. Perhaps one of the most notable for our purposes is the decision in In Re Pembroke Election Petition  2 I.R. 433. There the law was stated with admirable clarity by Palles C.B.:
“Collating all these cases together, I suggest that the law can be stated in these propositions:
1. If the election was conducted so badly that it was not substantially in accordance with the law as to elections, the election is vitiated, irrespective of whether the result was affected or not. … …
2. If the election was so conducted that it was substantially in accordance with the law as to elections, it is not vitiated by a breach of the rules or a mistake at the polls – provided that it did not affect the result of the election. …
3. But, even though the election is conducted substantially in accordance with the law as to elections, nevertheless if there was a breach of the rules or a mistake at the polls – and it did affect the result – then the election is vitiated. …”
40 It was argued that on this law, and indeed under the provisions of the Referendum Act 1942 had they been applicable, the irregularity here would fall within the first category, that is to say that the referendum was carried out if not contrary to the principles laid down in the Ballot Act (and now the Referendum Act) then certainly contrary to fundamental constitutional principles. Thus it could be said that it was “not really conducted under the subsisting election laws” and should be avoided whether or not a material effect on the outcome could be shown. It was not therefore an irregularity which was capable of benefiting from the saver first set out in s.13 of the Ballot Act and contained in 40(2) of the Referendum Act 1942, and substantially re-enacted in s.48(2) of the 1994 Act, and the question of effect on the outcome simply did not arise.
“Of the many classes of irregularity in the count of an election two are material in consideration of the question here:-
First – Those cases in which, by reason of the irregularity, the conduct of the election is contrary to the principles laid down in the body of the Ballot Act. In such a case, to use the words of Lord Coleridge, in Woodward v. Sarsons (1), the election is ‘not really conducted under the subsisting election laws,’ of which the Ballot Act is one of the principal and most important.
Consequently, according to the common law affecting elections, as laid down in that case, as well as by what I regard as the clear construction of the Ballot Act, such an election ought to be declared void.
The second class of irregularity is where, though the conduct of the election has been in accordance with the principles laid down in the body of the Ballot Act, and of the other existing election statutes, yet there has been a breach of the rules or forms of the schedule to the Ballot Act, or there has been a breach of some of the subsidiary provisions of those statutes; but where there has not been a corrupt or illegal practice by the candidate.
Speaking generally, an irregularity of this second class will not, per se, avoid the election. Avoidance in such a case depends not upon the statute, but upon the common law affecting elections; and avoidance or non-avoidance depends upon whether such irregularity had, or might have, affected the result. This, too, is the effect of the decision in Woodward v. Sarsons (1).” (p.448)
41 Of course the law has not remained in the state it was in 1908 or 1942 and the petitioner’s argument had to address the fact that the present law in the shape of the Referendum Act 1994 appears to impose an obligation on a petitioner to demonstrate in every case a material effect on the result of the referendum as a whole. Section 43 provides that a petition may question a provisional referendum certificate “on the grounds that the result of the referendum as a whole was affected materially by” the specified matters. The argument was made that the Act must only impose such a requirement, or should be interpreted as to only impose such a requirement, in respect of the leave stage where under s.42(3) of the 1994 Act the High Court was required to be satisfied on the ex parte application of the petitioner that there was prima facie evidence of the matter referred to in s.43 by subparagraph 42(3)(b) that the said matter was “such as to affect materially the result of the referendum as a whole”. Counsel pointed out that the first amendment to the Referendum Act 1942 was the insertion of the leave stage which was introduced by s.168 of the Electoral Act 1992 which introduced a new s.34A into the Referendum Act 1942, and which is in substantially the same terms as s.42(3) of the 1994 Act. On this argument, whilst s.43 does state that a referendum petition may question a referendum certificate on the grounds that the result of the referendum as a whole was affected materially by any of the matters set out in s.43 it does not explicitly impose a requirement that a material effect be actually demonstrated. While on the face of s.43 this might be a rather strained argument, it gains considerable support from the fact that as counsel showed, and the respondents were forced to accept, there is a tension, indeed inconsistency, between s.43(1) and s.48(2).
42 Section 48(2) operates as a saver to preclude the Court from ordering a repeat of a referendum where the referendum has been conducted in accordance with the principles contained in the Act, and when there is no material effect on the outcome. But if s.43 requires a petitioner to positively demonstrate material effect in order to be in a position to succeed in a petition, at which point s.48(2) potentially comes into play, there could be no circumstance in which the saver could ever apply in fact, since it requires the demonstration that the act or error did not affect the result of the referendum as a whole, as the opposite has necessarily been proven. Put in reverse, if indeed it could be demonstrated that there was no effect on the outcome of the referendum as a whole, as required to satisfy s.48, then the petition should fail at the first hurdle under s.43(1), and there would again be no circumstance in which s.48(2) could ever take effect. To interpret s.43(1) as requiring a demonstration of material effect to succeed in a referendum petition would therefore render s.48(2) effectively redundant. Since the law leans against redundancy, it was argued, the only logical interpretation which could reconcile s.48(2) with s.42 and s.43, was an interpretation which limited the onus on the petitioner to demonstrate material effect at the leave stage where undoubtedly it was unambiguously required by s.42(3)(b). That had however been satisfied in this case, and accordingly on this interpretation, the petitioner was entitled to succeed since the respondent could not on this analysis benefit from s.48(2). That saver could not apply to the error or wrongdoing in this case since it could not be demonstrated that the referendum had been conducted in accordance with the principles of the Act if those principles were to be understood as including the constitutional principles outlined in McKenna. In any event, the onus under s.48(2) was on the respondent to show no material effect.
43 Furthermore, and for good measure, counsel pointed out that if s.43(1) was to be read as imposing a general requirement of a demonstration of material effect, and was to have the consequential effect of rendering s.48(2) redundant, it was, to say the least, curious that in 1998, when it came to enact the Referendum Act creating the Referendum Commission, the Oireachtas, by s.12 of the Act, inserted an identical saver to that contained in s.48(2) and this time, by an amendment to s.43 itself. Section 12 of the 1998 Act reads as follows:
The objective of the new subsection seems to be to extend the same protection to actions of the newly created Referendum Commission as already exists for other officials such as returning officers. If however, s.43(1) is to be understood as imposing a requirement of material effect, then it appears to follow that the Oireachtas, in 1998, went to the trouble of inserting in s.43 itself a new s.43(3) which was redundant, or at least ineffective.
“Section 43 of the Act of 1994 is hereby amended by the insertion of the following subsection after subsection (2):
“(3) A provisional referendum certificate shall not be questioned by reason of a non-compliance by the Referendum Commission with any provision contained in the Referendum Act, 1998, or mistake made by the Referendum Commission if it appears to the High Court that the Referendum Commission complied with the principles laid down in that Act and that such non-compliance or mistake did not materially affect the result of the referendum.”
44 I quite see the logic of this argument, even if it requires a degree of historical excavation to appreciate its full force. On the other hand, if the argument is accepted, then it seems to run counter to both the words and logic of ss.42 and 43. In particular, the apparent plain words of s.43(1) would in effect be removed, and a diametrically contrary meaning assigned to s.43(1) to that which the words suggest. Either interpretation therefore involves considerable surgery in relation to the Act. There is no doubt that the argument advanced by the petitioner demonstrates that the drafting of the 1994 and 1998 Acts has resulted in a clumsy, confusing, incoherent and self-contradictory text which shows insufficient awareness of the operation of the Act and the background history of electoral law, but the Act is the law, and must be interpreted as best we can. As a matter of statutory interpretation alone, I have reached the conclusion that I am in no doubt that the 1994 Act should properly be interpreted as requiring that in order to succeed, a petitioner must establish not just an irregularity or defect or other matter set out in s.43, but also that the referendum as a whole was affected materially thereby. This must be established on a prima facie basis to obtain leave and on the balance of probabilities at the full hearing of the petition.
45 My reasons for coming to this conclusion are that the words of s.43(1) are clear and taken on their own, unambiguous. Furthermore, a requirement of material effect is consistent with the structure of the Act and the proceedings envisaged by it. It is entirely logical that if the question of material effect is to be addressed at the leave stage, it must also be a requirement at the substantive hearings. There would be little or no sense in requiring an applicant to establish on a prima facie basis that the irregularity was such as to have material effect, but not require material effect in fact to be proven at the substantive hearing. Indeed, there would be manifest objections in logic and fair procedures if that was the course required by the Act. Moreover, it seems clear that every development in the law relating to elections and referenda has been to tighten the procedure, and raise the threshold for proceedings seeking to set aside the result of an election or referendum. Section 13 of the Ballot Act is itself an early example. The introduction of the leave requirement itself in 1992 is another. One further instance is the difference between s.40(2) of the 1942 Act and s.48(2) of the 1994 Act. Under the 1942 Act the saver only operated if it could be demonstrate that the irregularity or other matter complained of did not have an effect on the outcome of the referendum in the individual constituency, whereas by 1994 that had been altered to a requirement that the poll would not be retaken in any constituency if it was shown that the irregularity did not affect the outcome of the referendum as a whole and presumably therefore, even if the irregularity affected the outcome in the particular constituency. It was entirely consistent with this development of the law to understand that the additional words introduced into s.43 that the result of a referendum “as a whole was affected materially by” were deliberately included in 1994 to impose a requirement that the petitioner demonstrate such an effect.
46 By contrast the interpretation contended for by the petitioner emerges from the Act only after considerable close analysis and against a historical background of quite complex and recondite election law. There is no doubt that once a provision such as s.48(2) is related to s.13 of the Ballot Act 1872 and indeed the pre-existing law, it becomes clear that such a provision is only a saver, and furthermore, that it has no practical effect if s.43(1) creates a substantive requirement of material effect whatever the irregularity or breach of which complaint is made. But the very process of understanding the point gives some insight as to how such an incongruous result could arise and does not support the conclusion that the Act must be read to elide any requirement of showing material effect at the substantive hearing. It seems likely that the background to s.48(2) was not sufficiently understood during the drafting process of the 1994 Act. Indeed in fairness to the drafter it should be said that the point now made emerged only after the provisions of the 1994 Act relating to referenda petitions had been exhaustively scrutinised by both the High Court and the Supreme Court in Hanafin, and by the High Court again in the extensive hearings in this case. If s.48(2) had been understood with the clarity which has now emerged after a painstaking analysis, it would of course have been more elegant to remove it since it appears to be completely superseded by s.43(1). However inelegance in drafting is a much lesser sin than inconsistency, contradiction and a failure to achieve the intention of the Oireachtas which would follow from the petitioner’s argument.
47 Furthermore, there is in my view a significant difference between interpreting the Act to require material effect and thus rendering s.48(2) ineffective and interpreting the Act as containing no requirement that the petitioner show material effect at the hearing of the petition, and leaving that matter to the limited saver under s.48(2). It is apparent that on a superficial reading, s.48(2) was designed to achieve essentially the same object as the other changes to the codes made since 1942, that is, to impose a more demanding standard for the invalidation of referendum results, and avoid invalidation for technical or electorally ineffective irregularities. The same objective underlies the introduction into s.43 of a requirement of proof of material effect on the outcome of a referendum. The provisions are therefore at least superficially consistent and it is perhaps understandable therefore that s.48(2) might be left in place. Indeed even if the potential impact of s.43(1) was adverted to during the drafting process it would not be surprising if s.48(2) was left in place on an assumption that it did no harm to, and did not appear to contradict or thwart, the intention of the Oireachtas. Section 43(1) on this view does not contradict s.48(2) or alter its meaning. If the saver in s.48(2) came to be applied or interpreted it would be given the same interpretation as its statutory predecessors. If s.43(1) imposes a requirement of a demonstration of material effect in every case, then it supersedes s.48(2) and may mean that it is impossible to conceive of circumstances in which it would take effect as a matter of fact. On the other hand, the interpretation advanced by the petitioner would not merely render the words of s.43(1) ineffective: it would instead require that they be removed, and not merely as surplusage, but so that the section would have a meaning exactly opposite to what it appeared to say. If the petitioner is correct, then a provisional certificate could be questioned, and set aside, on the grounds set out in s.43(1) even if there was no material effect on the result of the referendum as a whole. This is a much more dramatic interference with the text of the Act than the probable redundancy of s.48(2) and moreover would be plainly contrary to the intent of the Act, as I understand it. I do not consider that it is necessary to rely on s.5 of the Interpretation Act 2005 in this regard but if it is necessary to consider that section, it would plainly support the conclusion to which I have come.
48 The unhappy drafting of the Act does however, have the consequence that it cannot properly be said that the interpretation to which I would come as a matter of statutory interpretation, is the only possible interpretation. Accordingly, it is necessary to address the effect of the double construction rule and consider if the Constitution requires that the interpretation advanced by the petitioner while less likely, should nevertheless be accepted because the more likely interpretation of the words to require a showing of material effect, would be unconstitutional. This involves considering the question set out at the outset of this judgment. Why should any showing of material effect be necessary once it is shown that the referendum has not been conducted in accordance with the rules – and here rules which were derived from and required by the Constitution?
49 I do not think it adds much to the analysis of this question whether or not the interference with the referendum here can be characterised as a breach of the Constitution as interpreted in McKenna. That is because any interference with the conduct of a referendum is a serious interference with the constitutional right, the right of each citizen to vote, and even more importantly, with a solemn constitutional process. Of course if the breach was intentional and advertent, it is possible, indeed likely, that different considerations would apply. It might then be relevant that the conduct involved more than one breach of the Constitution. That however, does not arise here, and the question must be addressed at the level of principle, as to why once an interference with a vital and fundament constitutional process can be shown, and which can be demonstrated to be neither technical nor de minimis, the result should not be set aside and referendum re-run without having to go further into the question as to whether that interference had any decisive impact on the outcome of the referendum, an inquiry which on any view is time consuming, difficult and contestable, and indeed on the petitioner’s case, virtually if not entirely, impossible. After all, the Palles C.B. judgment in the In Re Pembroke Election Petition case, and the judgment of Lord Denning M.R. in Morgan v. Simpson shows that as a matter of history, once a serious breach of electoral law is shown, an election result is invalidated without requiring any showing that it had an impact on the outcome of the election. There are many examples in administrative law where once a breach of the procedures or law governing a decision making process is shown, the decision is set aside because it has not been reached in accordance with the law, and even if it might be said with considerable force that the same substantive decision would be reached by an untainted procedure. Why then, when there is a serious non-technical interference with perhaps the most important constitutional process, should it be necessary to go further and show some material effect on the result? This is a central question in this case not least because if it were established that a requirement of material effect was inconsistent with the Constitution, then it would follow that the less plausible, but possible, interpretation of the Act advanced by the petitioner, would have to be adopted.
50 For these purposes it is, I think, preferable to address first the general question of whether it is permissible, once a serious breach has been shown, to also require that a challenger show that the breach was, in the helpful phrase of Henchy J., electorally effective. I will leave until later the question of the precise content of any such test, if permissible. While the two questions are interrelated, it facilitates analysis to deal with the more abstract question first.
51 If the only question for the Court was whether the proposed amendment had been submitted to the people “in accordance with the law” for the time being in force as required by Article 46(2) then the answer would be clear. That indeed was the question properly posed in McCrystal, and addressed in that case. Only one constitutional value is engaged in such a case. However as Henchy J. pointed out in Dillon-Leetch, admittedly obiter but in a passage which is in my view as perceptive as it is persuasive, that is not the issue, or at least the sole issue, here. When the question is the setting aside of a vote, whether at an election or referendum, then the issue involves a balance between two different constitutional values. A referendum did take place and the people participated. They voted for or against, or did not participate, but in each case, their decision was the exercise of their sovereign constitutional entitlement. In many cases the individuals may have been completely unaware of or disregarded the government information campaign. If conscious of it at all, they may have acted in defiance of it or because of annoyance with it, but in each case a decision was made which it is for a citizen of Ireland alone to make. That has a very significant constitutional weight which must be put in the balance. A decision of the people, in the terms of Articles 46 and 47 was given. To set aside a referendum result simply because of a breach of the law, but particularly the regulations relating to provision of information, would be to demean the significance of the individual voter and the value of their individual vote, whether that vote was exercised or not.
52 The people of Ireland who participate in an election campaign or a referendum, whether they vote or not, are not mere ciphers whose free will in relation to a vote can be lightly disregarded. Their participation, or lack of it, in a referendum campaign is not to be treated as no more than leaves to be blown one way or another by the breeze generated by the skill of the pollster or advertiser or the designer of the information booklet who for example in this case decided, no doubt after much thought, to describe the referendum as “the Children’s Referendum” rather than as the Referendum Commission did, “the Children Referendum”. While that was one component of what was found sufficient to constitute a breach of the McKenna principles, the Constitution, in my view, permits, and indeed requires, that something more be shown before the outcome of the referendum is set aside on that basis. Even if one were to assume for a moment that some voters were decisively influenced to vote for the referendum by the contents of the government’s information campaign, and that this was not counterbalanced by those who decided to vote against or abstain because of the finding of unconstitutionality, it still follows that every other voter was not so influenced, and their votes have a very significant constitutional value. Those votes constitute the decision of the people. As Henchy J. put it:
53 This approach is, I consider, consistent with and required by the modern mass participation democracy established by the Constitution. While I recognise the benefits of recourse to the authorities, and have found them useful here, there is a limit to what can be learnt from the older cases. It is only necessary to consider the difference in the electorate in the Pembroke Urban District Council election which was the subject of the petition considered by the Courts of King’s Bench in 1908, and the electorate which was entitled to vote in the Children Referendum. In 1908 universal suffrage had not been achieved. Until 1898 only freemen and the propertied classes could vote in local elections and even after the significant expansion of the franchise in 1898, the electorate was still very limited. Furthermore, women were not permitted to vote until 1918 and then only those over 30 years of age. In many ways the elections prior to the introduction of the secret ballot, and which were the subject of much of the early and fascinating electoral law, were more akin to votes in private clubs, among a very limited franchise, and which could easily be re-run. Modern elections are by contrast exercises in mass democracy, expensive and onerous to run, and carrying with them an inevitable risk of individual error but also having a significant internal self-righting mechanism achieved by the sheer numbers participating, or eligible to participate, and the collective intelligence generated thereby.
“the courts will not allow an electorally ineffective breach of … principle to be used to set aside the correctly exercised constitutional right of the rest of the citizens forming the electorate in the constituency to elect their representatives in Dáil Eireann … see Article 16.1.2 of the Constitution. Otherwise it would be as much an inversion of constitutional priorities as to hold that an election in a particular constituency should be set aside merely because a voter or particular voters voted more than once (and breached the prohibition in Article 16.1.4 against so doing) but not so as to have an effect on the result of the election”.
54 By the same token the re-running of a referendum in modern times is not something to be lightly undertaken, and not a simple, or necessarily effective, remedy. The re-running of an election or a referendum is not like a laboratory experiment where it is possible to alter a single variable (in this case the government information campaign), maintain all other values constant, and re-run the experiment without additional cost or doubt. Instead, a re-run election is normally undesirable since it gives rise to an inevitable risk of uncertainties and doubts. If the result is the same it was largely a pointless exercise. If the result is different, then some voters will inevitably have reasons to feel disenfranchised. Some of the broader issues are well discussed in an article, “Remedying Election Wrongs” (2007) 44(2) Harvard Journal on Legislation 265 by Professor Steven F. Huefner and quoted in the decision of the Supreme Court of Canada in Opitz v. Wrzesnewskyj  3 S.C.R. 76, at page 100 :
55 Some of those concerns can be illustrated in the context of the present referendum. If the referendum is to be re-run, the past cannot be erased. The actors in the referendum including the voters cannot be put back into the situation in which they were prior to the 12th of November 2012. To take but one example, the monies expended by the various campaigns and the effect of the expenditure in the 2012 campaign, is long since dissipated. A number of voluntary bodies participated significantly in this referendum, which is normally a healthy thing in a democracy, and expended significant funds which will not be available for a second referendum. If the outcome of a new referendum were to reject the proposed amendment, in a very different campaign, perhaps by a narrow margin, then such bodies might legitimately feel that they had been deprived of the benefit of their participation in the campaign and a successful outcome. They and others would have no reason to have faith in the outcome of the second referendum and the stability and certainty which an election process is designed to provide, at least as to outcome, would not exist.
“. . . a new election can never be run on a clean slate, but will always be colored by the perceived outcome of the election it superseded. New elections may also be an inconvenience for the voters, and almost certainly will mean that a different set of voters, with different information, will be deciding the election. Moreover, there can be no guarantee that the new election will itself be free from additional problems, including fraud. In the long term, rerunning elections might lead to disillusionment or apathy, even if in the short term they excite interest in the particular contest. Frequent new elections also would undercut democratic stability by calling into question the security and efficiency of the voting mechanics.”
56 Modern mass participation democracy is designed above all to achieve a result and one that can be trusted by the public whether any individual member participated or not. The sheer weight of numbers involves a significant quality of self-balancing and self-correction so that individual incidents become less important in the overall mix. Thus the vote, or abstention, of a foolish and uninformed person has the same weight as that of the Nobel prize winner. The vote of the person who has studied and who has agonised on the issues has the same value and impact as that of the person who votes casually or on a whim. The person who fully intends to vote but is delayed has the same impact on the outcome as the person who pays no attention and never intended to vote, or the person who defiantly abstains on the grounds of principled objection to the issue or a candidate. Some people may vote on a plain misunderstanding of the issue. In some cases, as indeed has occurred, people may be confused by the formulation and led to vote in a manner contrary to their intention.
57 There will always be the possibility of error in the conduct of elections involving millions of individuals. Statements will be made which are exaggerated and even false. True statements may be misunderstood. Furthermore, if we think closely enough about it, the system includes the possibility for error and mistakes, at least at the margins. Whether it is the hanging chads of the Florida vote in 2000 or the uncounted or miscounted bundles in our own system, or indeed the system for the allocation of surpluses in the proportional representation system, all modern systems of election contain elements of randomness which mean that every election process falls far short of the perfect sterile laboratory standard, and any re-run is inevitably likely to be imperfect perhaps in different ways. Confidence in the system relies on relative weight of numbers if not curing, then overwhelming most imperfections and rendering them irrelevant. Quite apart from the importance in democratic theory of individual participation in elections and referenda, the system of modern mass participation democracy recognises the phenomenon of collective intelligence – the idea recognised many years ago that a collective judgement, if sufficient numbers are involved, will reduce and eventually remove the impact of individual extreme opinions. The immune system of democracy has therefore at least some resistance to individual error. A reluctance to re-run a vote is also based on the relative difficulty of mobilising the population at large for a sufficient time to engage with the issues in an election or referendum in the expectation that all the individual votes for all the myriad individual reasons will produce a collective result. This is not a process that can be reproduced easily.
58 It is also the case that while any individual election result may be decisive and even dramatic, it is in the life of the State, always to some extent, provisional. Another election is always on the way - where the result may be reversed. While this is perhaps less true for referenda, there are nevertheless a number of examples of the same proposal being put to the people sometimes with the same result (proportional representation) and sometimes with different results (divorce, the Nice and Lisbon referenda). Thus in a functioning democracy if there is sufficient dissatisfaction with the outcome of any referendum, then there remains the possibility of a further poll. Democracy works best when a result is accepted which lets people proceed with the victors enjoying their success and leaving the losers to lick their wounds and plot their revenge or indeed acquiesce in the result and move on. It is also far from ideal that the outcome of a referendum should remain in suspended animation while the parties, and the courts, investigate at length the complexities and nuances of voting patterns and issues. It is for example extremely unsatisfactory that an amendment apparently adopted by a significant majority of the people should be ineffective for a lengthy period. That is what has occurred here and there is no constitutional remedy for the people in general who have been deprived for that period of the amendment for which they voted, or those individuals who during that period would wish to have relied on the amendment, which in the light of the decision of the High Court and this Court, was adopted by the people. It would be particularly undesirable if for example the outcome of a general election were left in abeyance for even a fraction of that time. The challenge to the outcome of a referendum or election is not costless. On the contrary it imposes significant costs on society which cannot be recouped. All these considerations lead me to the same result arrived at so lucidly and concisely by Henchy J. more than 40 years ago: the Constitution not only permits but requires that before a poll, whether in an election or referendum, be set aside, that it be demonstrated not just that there has been a breach of the law, even the constitutional law, but also that the breach, to adopt Henchy J.’s words, has been electorally effective, that is that it has had an impact on the result of the poll. That much, at least, is in my view required by the value the Constitution places on the decision of the people in a referendum. But to prevent any challenge to a result, or adopt a test which renders a challenge effectively impossible would not be consistent with the obligation to uphold the law designed to permit the true ascertainment of the peoples’ decision on a matter of national policy, and the constitutional values which underlie it.
59 The central difficulty in this case is that the Court is required to identify a test which sets a balance between two values which have significant constitutional weight, and which may on occasion conflict. Very significant weight must be given to the importance of upholding the law, and in particular the constitutional law, regulating the holding of elections and referenda. Those laws exist in part to prevent a potentially beneficial collective intelligence becoming a damaging group think. They also exist to promote trust in the process and confidence in the result. On the other hand, as has just been discussed, there is a significant constitutional weight to be given to the decision of the people. The Court cannot allow one to decisively outweigh the other. While some might assert that courts should not properly engage in what can be seen as a political question, withdrawal from decsionmaking is not an option which is consistent with the balance set by the Irish Constitution, which after all, goes to some lengths to regulate by its own terms the electoral process. Even those theories of judicial review which lay emphasis on the separation of powers and the importance of recognising both the limits of the judicial power and the consignment by the Constitution of certain matters to the other branches, nevertheless stress the importance of positive enforcement of constitutional guarantees in respect of voting since that is a fundamental underpinning for the legitimacy of decisions of the representative branches. The Court must find the point of constitutional equilibrium. The democratic process should not be too readily stopped and ballot results upset by courts, but seriously flawed processes must not be beyond effective challenge by citizens. For present purposes it is enough to observe that the Constitution permits and indeed requires that more be established than a breach of the rules regulating a referendum or election and that a threshold of the electoral effectiveness must be passed before a referendum result is invalidated. It will however be necessary to address at a later stage, what precisely is required to pass that threshold. However before doing so it is necessary to address the argument advanced by the petitioner that it is effectively impossible to demonstrate material effect on the outcome of the referendum and accordingly that the test must either be adjusted or struck down as unconstitutional.
60 The petitioner claims that if the Act requires some demonstration of material effect, that she cannot or should not be required to demonstrate that the outcome of the referendum would have been different, that is, that without the government’s information campaign the result rather than Yes by 58% to 42%, would have been No. That she says is an impossible task and in so arguing she relies on a number of dicta from the decision in Hanafin and in the High Court judgment in this case . Setting such a test she argues is a form of catch 22 which forces citizens who have been the innocent victims of a constitutional wrong, to pursue an essentially futile though costly claim which cannot ever result in the successful overturning of a referendum. On this argument, it is to add insult to injury to accept the argument of the respondents (who the petitioner says it must be remembered gave rise to these proceedings by wrongly expending public funds on an unbalanced information campaign) that in order to assert a claim a petitioner must have commissioned a detailed, rigorous, comprehensive and therefore expensive public opinion survey, and moreover one conducted during the course of the referendum, and before any unconstitutionality is identified. Even then, it is argued that this case clearly demonstrates that the results of any such survey would be open to contest. The petitioner claims therefore that it should be enough to show an impact on the vote which is more than minimal. Alternatively it is argued the onus should shift to the respondent and in this case the wrongdoers, to show on the balance of probabilities that there has been no material effect on the result. It was further argued that because proof of material effect was something within the power of the State which could certainly more easily have conducted surveys either contemporaneously or in the aftermath of the referendum, the onus should shift for this reason. Finally, a different argument was that the Court should not have applied any onus of proof but should rather have conducted the petition as a form of inquiry.
61 These arguments are different ways of confronting the problem of proof in this case (and the question of what is to be proved and to what standard) which necessarily involves a consideration of the decision in Hanafin. While the petitioner understandably presents her arguments in the first place as merely distinguishing Hanafin or advancing arguments not addressed in Hanafin, counsel for the respondent has shown convincingly in my view, that the issue of difficulty of proof was central to Hanafin and that therefore, to accept the petitioner’s argument, at least in its strongest form, involves overruling that decision departing from it or qualifying it. It is thus necessary to consider the test the Court should apply when invited to depart from a prior decision of this Court involving the interpretation of the Constitution.
62 Counsel for the respondent placed reliance on the well known decision in Mogul of Ireland Ltd v. Tipperary (North Riding) County Council  I.R. 260 (“Mogul”). It stated:
This is a valuable statement that has guided this Court since it was first articulated in 1976. Furthermore, as the word “normally” indicates in the formulation of the test, this is not and could not be a bright line rule. There is an inevitable area of flexibility which cannot be defined with any greater precision. Just as judges can in good conscience disagree as to whether a previous decision was erroneous, they can disagree as to whether any error was clear or not. Nevertheless the decision is an important guide to the posture which this Court should adopt when invited to depart from a previous decision.
“A decision of the full Supreme Court (be it the pre-1961 or the post-1961 Court), given a fully-argued case and on a consideration of all the relevant materials, should not normally be overruled merely because a later Court inclines to a different conclusion. Of course, if possible, error should not be reinforced by repetition or affirmation, and the desirability of achieving certainty, stability and predictability should yield to the demands of justice. However, a balance has to be struck between rigidity and vacillation, and to achieve that balance the later Court must, at least, be clearly of opinion that the earlier decision was erroneous.
Even if the later Court is clearly of the opinion that the decision was wrong, it may decide in the interests of justice not to overrule it if it has become inveterate and if, in a widespread or fundamental way, people have acted on the basis of its correctness to such an extent that greater harm would result from overruling it than from allowing it to stand. In such cases the maxim communis error facit jus applies...”. (p. 272)
63 However the decision in Mogul occurred in the context of a question of statutory interpretation. It is obvious that the same considerations may not apply, or at least not apply with the same force, in the area of constitutional interpretation. There must be less scope for application of the principle of communis error facit jus in the area of constitutional decision-making. The risks inherent in erroneous interpretation of the Constitution are necessarily greater. Furthermore, even without error it may be necessary to depart from a previous interpretation. As Walsh J. observed in McGee v. The Attorney General & Anor  1 I.R. 284: “[i]t is but natural from time to time that prevailing ideas … may be conditioned by the passage of time; no interpretation of the Constitution is intended to be final for all time” (p.319). It is conceivable that viewed in the light of experience or in novel conditions which did not apply at the time of the original decision, it may become apparent that the decision should no longer represent the interpretation of the Constitution. Thus it is, I think, clear that the principle in Mogul must apply with some modification in the constitutional context. I agree therefore that the observations of Keane J. (with whom Denham J. agreed) in Society for the Protection of Unborn Children (Ireland) Limited v. Grogan & Ors (No.5)  4 I.R. 343:
64 However the doctrine of precedent is itself part of the administration of justice contemplated and required by the Constitution, and provides qualities of certainty and predictability which are important and indeed vital parts of that system. These are qualities of enduring importance, perhaps more so where there is a multi-tiered superior courts system. No one for example, considers that courts at a lower tier in the judicial hierarchy should be entitled to depart from decisions of the High Court or Court of Appeal or this Court, merely because it involves a constitutional matter. Furthermore, the Constitution is an exercise in delicate checks and balances, and part of that balance is historical as well as institutional. In the modern phrase, a constitution is to some extent an ongoing conversation across the generations. While Bunreacht na hÉireann is emphatically the peoples’ Constitution, it must be remembered that the youngest living person who can be said to have adopted the Constitution in 1937 will be 100 years old next year. When we speak therefore of the people adopting and giving to themselves this Constitution, that necessarily comprehends also the acceptance of the Constitution by subsequent generations, and the accretions to it by the process of constitutional interpretation.
“It is to be observed that Mogul of Ireland v. Tipperary (N.R.) C.C. was a case concerning the interpretation of a particular statutory code affecting a limited and defined section of the community (those whose property had been maliciously injured and the local authorities obliged to compensate them), a construction which had been adopted and applied for many decades and had been unanimously upheld by the former Supreme Court.
The contrast with the present case is plain. In this case, the court is concerned with the constitutional rights, not merely of every woman of child bearing age and of the embryos and foetuses which they are carrying, but of the even wider range of citizens who will be affected …
I do not think, however, that the absence of dissent in the earlier decision can be regarded as foreclosing any further judicial determination of the issue at the level of this Court, particularly when constitutional rights and their protection are in issue. So to hold would be seriously to encroach on the role of this Court as the guardian of last resort of the Constitution and might oblige it to perpetuate an erroneous interpretation of the Constitution.” (pp. 388-389)
65 The force and value of the Constitution comes in part from its age, and the respect which is paid to our predecessors who attempted to express an enduring statement of those core truths that they held to be fundamental and applicable in whatever circumstances. Of course the Constitution speaks in the present tense and would be of little value if it was not capable of addressing important contemporary issues. But as Murray J. observed in Sinnott v. The Minister for Education & Ors  2 I.R. 545:
Part of the value of the Constitution is that it is at one and the same time both old and ageless, and it gets some of its strength from the fact that it has been tried and tested through different generations in different circumstances. The process of constitutional interpretation is part of that historical continuum and an important exercise in the intergenerational conversation and solidarity. The power to declare an act unconstitutional, or to promulgate a binding interpretation of the Constitution, is one which is imposed upon the judiciary by the Constitution, and that duty must be performed without fear or favour. But it is a weighty power, and it is an appropriate recognition of the significance and reach of the power, that judges approach the task with the caution and humility of which the doctrine of precedent is one part. The very act of deciding that a previous considered decision of the Supreme Court is wrong is its own demonstration that courts can be wrong. That should give a court pause for thought, which is what respect for prior precedent does. I fully accept therefore that this Court must be free to adopt a new interpretation of the Constitution particularly in the light of experience or novel developments, and should certainly not engage in the process derided by Jonathan Swift of repetitions of the worst decisions, or of reasoning which is now clearly wrong. When in conscience this Court comes to a conclusion that a prior decision is wrong and cannot stand, it is its clear duty to say so, for the reasons outlined by Keane J, and the fact that there may often be conscientious and vigorous disagreement cannot deter it. But the fact that a previous Supreme Court, whether post or prior to 1961, or indeed 1937, has given thoughtful consideration to an issue, is a factor of substance not to be lightly disregarded, and it is necessary to carefully analyse the reasoning of a prior decision, and any supporting argument, before departing from it.
“Agreeing as I do with the view that the Constitution is a living document which falls to be interpreted in accordance with contemporary circumstances including prevailing ideas and mores, this does not mean, and I do not think it has ever been so suggested, that it can be divorced from its historical context. Indeed, by definition that which is contemporary is determined by reference to its historical context.” (p. 680)
66 In considering prior precedent, the fact that individual judgments were delivered in a case such as Hanafin can lead to misunderstanding. It is not a question of picking dicta from one or more judgments and presenting those as a binding authority. The weight of a case as a precedent will depend on those areas, necessarily narrower, in which it can be said that the Court or at least a majority of it, was agreed on an issue which was central to the decision in the case. It is also important that any decision must itself be read against the background of the facts of the case. One of the most significant features in Hanafin was that the evidence for the petitioner in the High Court signally failed to advance the petitioner’s case. Furthermore, the petitioner was forced to call one witness to prove certain opinion poll results, who would otherwise have been a primary witness for the respondent. His evidence was inevitably exploited by the respondent on cross-examination and it is plain from the judgments in the Divisional Court that the evidence of that witness, Mr Jones of MRBI, was found to be very persuasive and damaging to the petitioner’s case. Thus, by the close of the petitioner’s case, the Court had heard a significant aspect of the respondent’s case. The result was that the petitioner’s evidence at the close of his case could properly be characterised by counsel for the respondent in the Supreme Court as being in a “state of disarray”. This is important background against which both the legal arguments and some of the comments in the judgment should be understood.
67 I have no doubt that counsel for the respondent is correct in his analysis that it was argued in the Supreme Court in Hanafin that the petitioner should not bear an onus of proof; that it was enough to demonstrate that there was an impact on the referendum which was not trivial; that any impact on the referendum should not have to be decisive but should merely be significant in the conduct of the referendum; and that once interference with the conduct of the referendum was shown, the onus should shift to the respondent to disprove or to demonstrate that there had been no material effect on the outcome of the referendum. It is also clear that the Supreme Court collectively and unanimously held that the petitioner should bear an onus of proof and, notwithstanding some confusion of language, that the standard of proof should be the balance of probabilities. The Court also held that the onus moreover was on the petitioner to show impact on the outcome and result of the referendum and not merely an impact more generally. Quite apart from the detailed analysis of the report of counsel’s argument and the judgments in the Supreme Court, these arguments all necessarily follow from the posture of the case in the Supreme Court, and the state of the evidence, and in any event the limited capacity of this Court to review evidential findings. Furthermore, I am satisfied that these conclusions are correct, and for reasons which have already been addressed, consistent not merely with the Act, but also with the requirements of the Constitution.
68 While it is true that some of the judges, most notably O’Flaherty J., expressed a preference for hearing proceedings as an inquiry, that was by no means a finding of the Court, or even the individual judge, because the Court unanimously dismissed the appeal notwithstanding the fact that the case had not proceeded as an inquiry but had been disposed of in a fashion in which the rules of adversarial procedure had been rigorously applied. It is, I think, notable that the Referendum Acts contemplate some capacity in the Court to secure evidence even if not produced by the parties themselves but it does not appear to me that much in practical terms is achieved by describing the process as inquiry rather than proceedings. An onus must be discharged by the proposition that the referendum was materially affected by the interference with its conduct and inevitably, parties to the petition will adopt rival positions which they will advance and contest. That indeed is how most proceedings in these courts are dealt with and it is part of the administration of justice contemplated by the Constitution. Notwithstanding its costs and limitations, the adversarial process is an effective and trusted method of finding facts and testing and scrutinising propositions of law. There is no reason to think that the Constitution prohibits the utilisation of such a process in the case of a referendum petition, subject to the qualification already mentioned.
69 Nor do I think there can be any substance in the petitioner’s contention that the date of the referendum ought to have been postponed. It is accepted that that would have required the passing of special, and emergency legislation by the Oireachtas. That is a matter which the Constitution consigns to the legislative branch, which in our system means that the executive branch is also involved. The Constitution does not allow the courts to compel the passage of legislation which is a matter within the discretion of the legislature. Though the mechanism of the Referendum Act, as interpreted, provides a remedy in the case of wrongdoing which has material effect on the result of the referendum since it permits the referendum to be re-run. If such were to occur, and were to be based on a finding that was in some sense time sensitive, so that the perceived impact on the outcome of the referendum might have lessened if there had been a postponement, then of course in hindsight in such a case it might have been wiser for the legislative and executive branches to pass legislation to defer the holding of the referendum. But the fact that a referendum was held on a particular day and not postponed cannot affect the decision whether on that occasion the result was materially affected by the alleged wrongdoing. It either was, or was not. Furthermore, since the passage of legislation proposing an amendment to the Constitution, the fixing of a referendum date and a referendum campaign are all part of an important constitutional process, any interference with a date set by the legislative branch could itself be open to challenge. Accordingly, I would not depart from the decision in Hanafin that there is an onus on a petitioner to show a material effect on the outcome of the referendum as a whole.
70 I have however more difficulty with the petitioner’s contention that the test in Hanafin of showing a material effect on the result of the referendum in the sense of demonstrating that it must have been different, was applied in the High Court in this case to set for the petitioner an impossible standard. Indeed that term was used in Hanafin where a number of judges referred to the fact that having regard to the constitutional obligation of the secrecy of the ballot, it was impossible to say that the outcome must have been different if the government breach had not occurred. The petitioner also points, with some merit, to the facts of the present case as illustrating the point that a citizen who wishes to challenge the outcome of a referendum where there has been proven to be an interference with the conduct of the referendum, should not be required to prove or seek to prove the impact of such interference only by the production of sophisticated and elaborate surveys of voting intention which will in any event become the subject of academic contest about methodology and the conclusions to be drawn therefrom. Such a process will generate abstruse academic debate leading almost inevitably to a conclusion not that the outcome was or was not been affected, but rather that it was impossible to say. The petitioner argues with some force that this would be to create a constitutional catch 22 standard and cannot be required by the Constitution, particularly in cases where the starting point of this case, and indeed Hanafin, was that an interference with the constitutional process of the referendum had been proven.
71 It is true that some of the judges in Hanafin used the language of impossibility and also referred to the difficulty posed by the secrecy of the ballot. Thus, in the High Court Barr J. said at page 378: “…in my view it is impossible to divine with any reasonable degree of accuracy what effect the advertising itself might have had on influencing some voters to vote ‘Yes’ rather than ‘No’.” He went on to say that ”In my view it is impossible to asses with any confidence the conflicting effects of such factors on the ultimate decisions made by the voters as to whether they would support or reject the proposed amendment of the Constitution”.
72 I think counsel for the respondent is correct to say that this is properly understood as referring to the evidence in the case itself rather than a general state of impossibility. However in the Supreme Court Hamilton C.J. said:
This is an accurate statement of the limits of direct evidence in this or any other case, but does not go so far as to suggest that it is impossible by other evidence to ascertain on the balance of probabilities the influence of certain factors on the referendum. Denham J. (as she then was) made no reference to this issue. Blayney J. said at page 444:
“Because of the secrecy of the ballot, it is not possible to ascertain, by direct evidence, the factors which influence the people in casting their votes, what their motives were in casting their votes and the reasons why they cast their votes in a particular way”. (p. 430)
This statement refers to difficulty and not impossibility, and in any event possibly goes too far in suggesting that the secrecy of the ballot prevented any evidence being obtained as to how people voted or as to why they voted in the way they did. The secrecy of the ballot does not preclude evidence being volunteered by individual voters as to how and why they voted, and opinion polls of the voting intentions and exit polls as to voting reasons would be admissible, as in my view would expert evidence as to the conduct of a campaign.
“…to prove that the result of the referendum as a whole has been materially affected becomes extremely difficult because, as was submitted by the Attorney General, the secrecy of the vote prevents any evidence being obtained as to how people voted or as to why they voted in the way they did.”
73 The judgments of O’Flaherty and Barrington JJ. go perhaps the furthest in this regard. At page 457 Barrington J. said:
“It seems to me to be impossible to assess on any scientific basis what effect the Supreme Court ruling, and the reaction of the Government and the various parties to it had upon public opinion.”
This is clearly a reference to the facts of the particular case. But at page 458 he said:
This is a rather puzzling concluding observation, since no other judge considered that the issue was one of the admissibility of evidence (or indeed its acceptability). But I do not think it can or should be read as asserting that it is impossible in all cases to prove material effect on the outcome of the referendum. Indeed, at page 457 in the introduction of this section of his judgment, Barrington J. observed that the question of the result of the election being materially affected by the unconstitutional conduct of the government was essentially a matter for the Divisional Court which heard the evidence of the petitioner and the witnesses, and the Supreme Court could not interfere with that assessment. That appears to have been the basis upon which he rested his judgment.
“We know how they voted. We don’t know why they voted the way they did. We are not competent to interrogate them as to do so would amount to breaking the secrecy of the ballot. Primary evidence being thus excluded it seems to me that we cannot accept secondary evidence from experts who seek to second guess why the people voted in the way they did.”
74 Finally, the petitioner laid stress on two passages in the judgment of O’Flaherty J. At page 437 he said:
At page 436:
“Can it be said the ‘Yes’ side were not influenced to the extent of that small percentage at least by the Government’s advertising campaign? Of course, that cannot be said: it is something incapable of proof.”
Again, while this is framed by reference to the evidence in the particular case, I accept it is capable of suggesting a more general proposition.
“…how is the verdict of the people to be assessed in the circumstances that there was no way to prove by the evidence adduced, or to rebut by any evidence that was held in reserve, that the advertising campaign did or did not materially affect the result of the referendum?”
75 When the judgments of the Supreme Court are read together, I think however that they must be understood in the context of the particular evidence in that case. I do not think that the judgments of the Supreme Court can fairly be read for the self-contradictory proposition that while the Act is required to be read broadly to ensure that a referendum can be retaken in the event of a constitutional wrong affecting the referendum as a whole, and in particular a wrong which was capable of affecting materially the outcome of the referendum, such an order could never be made because such material effect is impossible to prove. Nevertheless the language used in some of the judgments does give rise to problems of interpretation.
76 This difficulty of interpretation is compounded by an uncertainty in the decision in Hanafin as to what precisely, a petitioner is required to prove. There is no doubt that the petitioner must establish a material effect on the outcome of the referendum and not just an impact on it: while not stated explicitly, that necessarily follows from the rejection of the argument made at page 402 that “affect” should mean “substantially bear upon the result” rather than “altering” it. But the judgments of the Supreme Court in Hanafin do not expressly address the question of what must be proved to establish material effect and in particular whether a petitioner must prove as a matter of probability that the result must have been different without the wrongful interference, or merely that there is a real, or reasonable, likelihood that the outcome might have been different or a real risk to that was so.
77 Counsel for the respondent first argued boldly that a petitioner was required to show that the outcome of the referendum must have been different. Subsequently he refined that submission by suggesting that it should be read against the standard of proof and thus meant that a petitioner must show on the balance of probabilities that the outcome must have been different. This he said amounted to showing that the outcome would probably have been different, and was he argued, not substantially different to any test of likelihood or risk of a different result. I do not think that this is so. In my view, while all this is a matter of degree and nuance and the application of different language may lead to the same result in concrete circumstances, nevertheless there is a difference between proving (on the balance of probabilities) that a result must or would have been different, and proving (again on the balance of probabilities) that there is a real likelihood or real risk that it is so. There is a difference between what must be proved and the standard by which it is proved.
78 I do not think that the judgments in Hanafin can be understood as addressing this issue and resolving it definitively. It is of some significance that Denham J. in her judgment referred to the test then applicable in extradition cases in Finucane v. McMahon  1 I.R. 165 at p. 226 where it had been held that courts should not extradite a person if there is “a real danger” that a citizen’s constitutional rights would be breached. It is plain from the reference to this test in Hanafin, and the fact that there is no dissent or debate upon the test in other judgments, that the specific issue identified here was not debated or discussed in any detail in the argument or judgments in that case, still less decided. This is not surprising because it is clear that all the judges considered that the evidence of the petitioner failed to reach any threshold and did not reach the stage where it could be said to raise a real or substantial likelihood that the outcome of the referendum might have been different. Accordingly, this is a matter which in my view, the Court must address from first principles, albeit with the guidance available to be gleaned from the decided cases.
79 In my view, the Constitution does not require that a material effect on the result of a referendum, or election, be interpreted as meaning must necessarily have altered the result, even as a matter of probability.
80 As a matter of logic, if it is sufficient to sustain a result in a referendum or election that a court is not satisfied, on the balance of probabilities, that the wrongdoing did not materially affect the result, then it should follow that if a court were satisfied that the wrongdoing did materially affect the result, on the balance of probabilities, so as to show that the result would have been different, the court should not order that the election be re-run, but should substitute the correct result. In some election disputes, this is what occurs. The election court will declare the correct result which is binding and effective. It is more logical therefore that an election is only to be rerun when the outcome is in doubt.
81 It seems that the development of the electoral challenges in the 20th century has meant that such challenges tended to be limited to what might be described as mechanical issues, such as the counting of disputed votes, the resolution of issues as to an individual’s entitlement to vote, and questions such as whether a particular marking shows a preference for a candidate, or should be treated as a spoilt vote. When these disputes are resolved, the impact on the result can be determined not merely as a matter of probability but as a matter of certainty. These examples may have led to a belief by analogy that a mathematical standard also had to be established even in non-mechanical cases, which are however not susceptible to such precise or certain resolution. One example is the case of corruption where it is not clear how many people were affected, how many of them voted and indeed how they voted. Another simpler example raised in argument is also instructive. Amongst the issues which might properly be described as mechanical are those cases where for example, a box of ballot papers is lost and where it may be known how many votes were cast in that box, but it will not be known in whose favour such votes were cast. Counsel referred to the test applied in Canada and referred to in Opitz v. Wrzesnewskyj described as the so called “magic number” test. If the margin between the sides on the votes counted exceeds the number of votes in the lost box, then the election will be upheld notwithstanding the irregularity. That is because it is certain that the disputed votes could not have altered the outcome. But if the margin is less than the number of the votes lost, then the election must be re-run. I imagine some similar test would be applied in this jurisdiction and that the underlying reasoning is immediately and intuitively acceptable. But it must be recognised that the test implied by such an approach is not one of certainty, or even probable impact on the result of an election or referendum, but is rather that if the votes in the box had been available, the result might have been different. It is a test of possibility and in some cases it might be quite an implausible possibility at that. Counsel properly drew our attention to subsequent academic debate in Canada as to whether it was necessary to apply the magic number test in its full rigor and whether it was not possible to refine it somewhat by reference to voting history, opinion poll results and outcome in other constituencies. This too makes sense, but it does not alter the fundamental nature of the test. An election is upheld when it can be shown as a matter of certainty that it was not affected by the irregularity, but must be re-run, however analysed, when there is a reasonable possibility that the outcome could have been different.
82 This conclusion is also supported by a number of dicta in cases where courts have addressed disputed elections and where the problem cannot be necessarily reduced to a counting of votes after resolving disputes as to the validity of any particular vote. It is particularly noteworthy in this context, in my view, that Henchy J. in the course of his judgment in Dillon-Leetch adopted with apparent approval, the test in Gill v. Reed (1874) 31 L.T.R. 69, 85 that “the informality is to such a nature as to be fairly calculated in a reasonable mind to produce a substantial effect on the election”. The extract from Woodward v. Sarsons quoted above refers to a breach of the rules which “did affect or might have affected the majority of the voters” (p. 751). When stating the test in his own words Henchy J. said:
83 Similarly in the In Re Pembroke Election Petition case Palles C.B. referred to those cases where an irregularity did not invalidate an election per se. At common law he said it was necessary to show an irregularity but also whether such irregularity “had, or might have, affected the result” (p. 449). On two other occasions in the same judgment, at pages 449 and 450, Palles C.B. repeated the test in precisely those terms, i.e. that it was necessary to show that the irregularity might have affected the result. Similarly in Levers v. Morris  3 All ER 1300, which was approved by Denning M.R. in Morgan v Simpson, Waller J. at page 229 said “[i]f this were a case where numbers of voters had been prevented from voting or where there was a substantial irregularity which might have affected the result then it would have been necessary to order a new election” (emphasis added). In the Canadian case of Opitz v Wrzesnewskyj a majority of the Court had to consider the question of when it could be said that an irregularity affected the result of an election. The majority considered that under Canadian law, an irregularity affects the result not because it alters the outcome but rather because it has an impact on anyone entitled to vote. Once that was established however, then the Court considered a court may annul the election. In exercising that discretion, the court must be satisfied that, because of the rejection of certain votes, the winner is in doubt, and it would be unreasonable for the court not to annul the election. For the purposes of this application, the “magic number” test will be used to make that determination. Finally, it is noteworthy that the European Commission for Democracy Through Law (Venice Commission) Code of Good Practice on Referendums (CDL-AD (2007)008) which was referred to in the judgments in McCrystal was also referred to in this case. One of the principles at paragraph 3.3(e) provides that there must be an effective appeal body which “must have authority to annul the referendum where irregularities may have affected the outcome” (emphasis added). None of these statements are necessarily advertent and directed to the precise issue which arises in this case, but they are instructive in that they suggest a consensus, not perhaps overly analysed or debated, but significant nonetheless, that where the outcome of an election is in real doubt by reason of some irregularity of interference with the process, it must be re-run.
“the election should not be avoided unless the court hearing the petition is satisfied the irregularity amounted to a breach of one of the principles laid down in the body of the Electoral Act 1923 and that the result of the election “has been , or could have been affected”.
84 This conclusion in my view follows as a matter of not just textual analysis and precedent, but also as a matter of constitutional principle. Modern liberal democracy involves the ascertainment of the will of the people which it is accepted will be determined by the decision of the majority of them. But democracy depends as much if not more upon the consent of the minority and their acceptance of the result. Part of the consent of the minority is based on acceptance of, and trust in, the process by which the result has been arrived at. Where an irregularity has occurred which is an interference in the conduct of an election or referendum, then there will come a point at which it can be said that a reasonable person will no longer have the requisite confidence in the outcome to allow the result to be accepted as the basis upon which society should collectively proceed. In such circumstances it will be necessary to re-run the election or referendum notwithstanding the difficulties and imperfections of that course.
85 Accordingly, I would hold that “material effect on the outcome of a referendum” involves establishing that it is reasonably possible that the irregularity or interference identified affected the result. Because of the inherent flexibility of this test, it may be useful to add that the object of this test is to identify the point at which it can be said that a reasonable person would be in doubt about, and no longer trust, the provisional outcome of the election or referendum.
86 Furthermore, I do not accept that the only manner in which material effect on the result of a referendum is by the type of expert evidence canvassed in this case, and which was addressed in such detail and such length during the course of the High Court hearing. That sets a standard which is at once too costly, elevated, demanding, complex, and contestable, and which takes far too much time, knowledge and expertise to address, to be compatible with the requirements of a review of a constitutional process such as an election or referendum. The search for the evidence which would prove effect on the result as a matter of near certainty was lengthy, costly, and ultimately frustrating in this case. Here much of the thrust of the evidence was destructive rather than constructive, addressed not towards showing that the outcome of the referendum had not been affected, but rather that it was not possible to say that it had.
87 Democracy produces its results, notwithstanding occasional technical imperfections in the process, by the sheer weight of numbers which produces a result in gross, and which normally settles the matter. Any process for review should be capable of producing a speedy, clear cut outcome. In this respect it is worth applying some simple rules of common sense. First, by the time a court has to consider the question of material effect there will have been a finding of an irregularity or interference in the conduct of the referendum. Second, it is worth recalling that the reason why such irregularities and interferences are prohibited is precisely because they have the capacity to interfere with, and distort, the outcome of an election and the public acceptance of it. In the particular context of this case the rationale of the decision in McKenna was not that the expenditure of public funds was wrong per se, but that it was wrong because it had the capacity to distort, alter and interfere with the outcome of a referendum. When such matters are established, a petitioner is a long way towards establishing his or her case particularly when addressed in the context of the test set out at paragraph 85 above. Third, evidence of opinion polls, the trends in voting, and the opinion of expert political scientists and commentators on both the progress of a campaign and the impact of the irregularity and any court determination, are all relevant considerations to be taken into account. Fourth, however, it should not be lost sight of that a referendum, like any election, is about the votes of individual citizens. While such individuals cannot be compelled to disclose their vote or the reasons that led them to do so, there is no prohibition on their giving such evidence voluntarily. Where it is alleged that significant numbers of citizens have been duped or misled in to voting in a way other than would have represented their views because of misleading statements advertising or some other interference, then it might naturally be expected when the misleading nature of such statements is revealed, that some significant number of those people would be willing to give evidence of the fact that they feel that their vote was improperly influenced and that without that influence they would have voted otherwise. Finally, in this regard, it is important to have particular regard to the margin. Any court asked to invalidate the outcome of a referendum on the grounds that it might reasonably have been affected by an irregularity or illegality, should have at the forefront of its mind the numbers taking part in the vote and the margin involved. Finally it is also important to have regard to the nature of the interference with the conduct of the referendum.
88 Applying this approach here, there is no doubt that the issuance of information booklets and other messages which had been found in a number of respects to favour the Yes side is a serious breach of the government’s duties as outlined in the McKenna case. But it does not necessarily follow that it is a serious interference in the conduct of the referendum. The majority of the statements contained in the information booklet are statements to which no objection could be made. Those that were found to lack balance, were condemned on that ground and not necessarily because they were likely to be influential still less decisive in terms of the casting of any votes. While acknowledging that the impact of partial statements, presented as neutral information, may be significant, it still seems clear that the government’s information campaign was primarily information and likely to have less immediate impact than the advertising campaign at issue in Hanafin which was designed to obtain Yes votes, and even in this regard, making every allowance for the justified scepticism with which the Court approached the evidence of the impact of advertising on voting. On the other hand, the margin here was on any view very substantial indeed. It is necessary for the petitioner to demonstrate that there was a reasonable possibility that there were either 85,000 people who voted Yes who would have voted No, or alternatively 170,000 people who did not vote but would have voted No, or some combination of the two. In my view, it is clear from the decision of the High Court that the petitioner’s evidence fell decisively short of this standard. It is in my view particularly telling that less than a handful of individuals came forward to say, with varying degrees of plausibility, that their actions at the referendum, whether voting or abstaining from voting, were affected by the contents of the government’s information campaign.
89 I acknowledge that the trial judge applying some of the dicta in Hanafin at times relied on the evidence as to the lack of data, and also that it was necessary to prove that the constitutional interference produced a majority in favour of the proposal. However it is clear that he considered the evidence fell well short of this standard and was wholly insufficient to establish the petitioner’s case on any test, and for the reasons set out above I consider this conclusion was entirely correct. Furthermore the test set out above of the view a reasonable person would take is an objective test which can and should be applied by this Court. I would accordingly dismiss the appeal on both the petition and the plenary proceedings. The result is that the provisional outcome of the referendum becomes final and the amendment voted on by the people on the 12th of November 2012 becomes part of the Constitution.
90 As a matter of principle, it is entirely unsatisfactory that for more than two years the people have not had the benefit of a constitutional change they, or a majority of those who voted, agreed to. It is also unsatisfactory that those conscientiously opposed to the amendment may have considered that the result was flawed, but beyond effective scrutiny. I hope that it will be possible in the light of this judgment to limit the number of challenges that might be taken, and if initiated, granted leave, so that only matters with a real likelihood of resulting in a setting aside of a referendum result are the subject of court proceedings. It should also be possible to simplify the issues raised in challenges to referendum outcomes so that they are easier to resolve speedily, one way or another and if there is real reason to lose trust in the outcome of the poll to have that recognised and remedied. But it must also be recognised that when an important issue arises such as that raised in this case, that complex issues will be involved, and some time will be required to hear and determine any case and any appeals. It is all the more important therefore that those with the power to do so should strive to ensure that future referenda should be resolved in the ballot box and should not give rise to disputes and differences that lead to court proceedings. The responsibility for the unacceptable interference with the will of the people which the process has occasioned in this case can be traced to a decision to spend €1.1 million of the people’s money in a very short and sensitive period. That money, if spent lawfully, would merely duplicate the efforts of the Referendum Commission established to provide neutral and fair information to the people in the context of a proposed constitutional amendment. Insomuch as any information provided by the government is different from that approved and disseminated by the Commission, there would be an inevitable danger of partiality, particularly where it was being provided by the Department of State sponsoring the proposed amendment and convinced of its merits, and when a clear distinction was not maintained, or perhaps in the circumstances was not capable of being maintained, between the importance of taking part in the referendum and the perceived importance or desirability of the proposal being put.