Abstract The Singapore Court of Appeal’s decision in Prabagaran a/l Srivijayan v. Public Prosecutor represents a substantial development in Singapore’s law on the doctrine of severability in constitutional review. An examination of Prabagaran reveals rich theoretical underpinnings relating to the nature of legislative intent. The case rightly locates the crux of the severability inquiry in secondary legislative intention: in other words, the legislature’s intention, at the time a statute was enacted, as to what should happen in the event that part of the statute is later held to be unconstitutional. This approach is preferable to the approach of asking whether excision of unconstitutional parts of the legislation would leave behind something that is ‘substantially a different law’, an approach that can lead to the judicial frustration of legislative policy. The search for secondary legislative intent is not just a matter of speculation; Prabagaran demonstrates how it may be inferred from evidence such as the structure of legislation, legislative history, and speeches in Parliament. In addition, Prabagaran highlights the importance of applicants’ identifying precisely the object of a constitutional challenge and the exact reliefs sought. 1. INTRODUCTION Article 4 of Singapore’s Constitution1 provides: This Constitution is the supreme law of the Republic of Singapore and any law enacted by the Legislature2 after the commencement of this Constitution which is inconsistent with this Constitution shall, to the extent of the inconsistency, be void. The Constitution does not make clear what the words ‘to the extent of the inconsistency’ mean. A 2012 case provided the gloss that Article 4 provides for the unconstitutional portion of the law to be severed while retaining the remaining part of the law in the statute books.3 but this does not say much: it does not explain how such severance is to be effected or how to tell whether severance is possible at all. For example, it might be the case that effective severance would require the exercise of power which the courts do not possess. Alternatively, severance might simply not be possible because what is left behind is too incoherent to count as a law. Or it might be that, because severance is possible, a finding of unconstitutionality might not help a person who seeks to have only part of a law struck down for unconstitutionality with the motive of escaping the other, constitutionally valid, part of the law. It was not until a few years later that the law on severance was substantially developed following a series of constitutional challenges against legislation relating to the sentencing of drug traffickers. The most recent of these is Prabagaran a/l Srivijayan v. Public Prosecutor (‘Prabagaran’).4 This note aims to expose and explore the rich theoretical foundations underlying the decision in Prabagaran, with a view to commenting on how litigants and courts ought to approach constitutional challenges to parts of legislation. Our hope is that Prabagaran will serve as a fertile source of theoretical ideas, particularly relating to the idea of legislative intention, as well as an illustration of how those ideas may be put into action both in Singapore and in other jurisdictions. 2. THE LEGISLATION ON SENTENCING DRUG TRAFFICKERS Prior to 2012, under the Misuse of Drugs Act5 (the ‘Act’), a person who trafficked in or imported or exported drugs6 faced the mandatory death penalty under section 33 read with the Second Schedule of the Act. Section 33 provides: the Second Schedule shall have effect…with respect to the way in which offences under this Act are punishable on conviction. The Second Schedule, in turn, contains a table listing various offences and the ranges of punishments therefor—an extract will illustrate this: Extract from the Second Schedule to the Act Section General nature of offence Punishment 5 (4) Unauthorised traffic in controlled drug containing such quantity of diamorphine being— (a) not less than 10 grammes and not more than 15 grammes Maximum 30 years or imprisonment for life and 15 strokes7 Minimum 20 years and 15 strokes (b) more than 15 grammes Death8 Section General nature of offence Punishment 5 (4) Unauthorised traffic in controlled drug containing such quantity of diamorphine being— (a) not less than 10 grammes and not more than 15 grammes Maximum 30 years or imprisonment for life and 15 strokes7 Minimum 20 years and 15 strokes (b) more than 15 grammes Death8 View Large Section General nature of offence Punishment 5 (4) Unauthorised traffic in controlled drug containing such quantity of diamorphine being— (a) not less than 10 grammes and not more than 15 grammes Maximum 30 years or imprisonment for life and 15 strokes7 Minimum 20 years and 15 strokes (b) more than 15 grammes Death8 Section General nature of offence Punishment 5 (4) Unauthorised traffic in controlled drug containing such quantity of diamorphine being— (a) not less than 10 grammes and not more than 15 grammes Maximum 30 years or imprisonment for life and 15 strokes7 Minimum 20 years and 15 strokes (b) more than 15 grammes Death8 View Large It is not surprising that several offenders who had been sentenced to death sought to challenge the constitutionality of this sentencing regime. However, these challenges failed.9 In 2012, section 33B was added to the Act, providing courts with the discretion to sentence such a person to life imprisonment and caning instead of death in certain circumstances. For present purposes, we are concerned with the following set of circumstances10: Extracts from section 33B of the Act (1) Where a person commits or attempts to commit an offence [of trafficking or importing or exporting drugs], being an offence punishable with death under […] of the Second Schedule, and he is convicted thereof, the court— (a) may, if the person satisfies the requirements of subsection (2), instead of imposing the death penalty, sentence the person to imprisonment for life and, if the person is sentenced to life imprisonment, he shall also be sentenced to caning of not less than 15 strokes; […] (2) The requirements referred to in subsection (1)(a) are as follows: (a) the person convicted proves, on a balance of probabilities, that his involvement in the offence […] was restricted— (i) to transporting, sending or delivering a controlled drug; (ii) to offering to transport, send or deliver a controlled drug; (iii) to doing or offering to do any act preparatory to or for the purpose of his transporting, sending or delivering a controlled drug; or (iv) to any combination of activities in sub-paragraphs (i), (ii) and (iii); and (b) the Public Prosecutor certifies to any court that, in his determination, the person has substantively assisted the Central Narcotics Bureau in disrupting drug trafficking activities within or outside Singapore. […] (4) The determination of whether or not any person has substantively assisted the Central Narcotics Bureau in disrupting drug trafficking activities shall be at the sole discretion of the Public Prosecutor and no action or proceeding shall lie against the Public Prosecutor in relation to any such determination unless it is proved to the court that the determination was done in bad faith or with malice. 3. EARLY FALSE STARTS: THE IMPORTANCE OF IDENTIFYING THE OBJECT OF CHALLENGE The new legislation spawned further litigation. Some of it consisted of challenges to the Public Prosecutor’s exercise of his discretion so as not to certify under section 33B(2)(b) that an offender had substantively assisted the Central Narcotics Bureau (hereafter, to issue a ‘Certificate’); these challenges failed.11 Other challenges were brought against the legislation itself rather than against administrative action taken pursuant to the legislation. These merit closer study because, rather than being challenges to one rule, they were challenges to a complex statutory scheme. The latest of these challenges (resulting in the most comprehensive judicial treatment of the topic of severance) is Prabagaran, which was decided by Singapore’s Court of Appeal. Before studying that case, however, it is worth briefly outlining the two cases leading up to it (which were decided by the same court). (A) Quek Hock Lye v. Public Prosecutor The first such case was Quek Hock Lye v. Public Prosecutor (‘Quek’).12 The applicant had been sentenced to the then-mandatory death penalty (and had his appeal against conviction dismissed) prior to the enactment of section 33B. After the enactment of section 33B, he had sought to have his sentence re-considered pursuant to section 27(1) of the Misuse of Drugs (Amendment) Act 201213 (the ‘Transitional Provision’): 27 .(1) Where, on the appointed day, the Court of Appeal has dismissed an appeal brought by a person for a relevant offence, the following provisions shall apply: (a) the person may apply to the High Court to be re-sentenced in accordance with section 33B of the principal Act […] However, the Public Prosecutor had refused to issue a Certificate to him, rendering him ineligible to be sentenced under section 33B to imprisonment and caning instead of death. He sought to argue that the Transitional Provision violated the constitutional guarantee of equality before the law because, of all those who had previously been convicted of capital offences, some (that is, those who met the criteria in section 33B(2)) could now be spared the death penalty, whereas others (like him) could not.14 The court dismissed this argument. It added that, if it were to succeed, the effect would be that the Transitional Provision would be struck down, leaving the applicant with no way to have his sentence of death changed.15 In other words, the application, ‘even if successful, could not benefit the Applicant’.16 Notably, the applicant did not claim that, for instance, part of the Transitional Provision was unconstitutional or that its effects when read in conjunction with section 33B were unconstitutional—perhaps such arguments could have addressed the court’s concerns. Nor does he appear to have made submissions on what the effects of a finding of unconstitutionality would be. His only retort was that, should the application succeed, ‘Parliament [would have] a chance to reconsider the law’17—it is not clear what this means or how it would have helped him. (B) Yong Vui Kong v. Public Prosecutor This weakness in the way in which the applicant in Quek put his case echoes that in another case, Yong Vui Kong v. Public Prosecutor (‘Yong’).18 The applicant, who had been sentenced under section 33B(1)(a) to life imprisonment and caning, argued, inter alia, that his caning sentence was in violation of the constitutional right to equality in that, according to section 325(1) of the Criminal Procedure Code (‘CPC’)19 (the ‘Caning Restriction Provision’), women and men aged above 50 could not be caned: 325. (1) The following persons shall not be punished with caning: (a) women; (b) men who are more than 50 years of age at the time of infliction of the caning […] The trouble was that the applicant had framed his challenge as a challenge to the Caning Restriction Provision only. The court pointed out that, even if this challenge were to succeed, the result would not be that the ‘caning regime as a whole [w]ould be struck down’. Again, because ‘the violation of [the constitutional guarantee of equality], if any, stems from the enactment of [the Caning Restriction Provision]’, a successful constitutional challenge would merely lead to the striking down of the Caning Restriction Provision; the applicant’s caning sentence would remain.20 (C) Analysis In Quek and Yong, valuable opportunities to develop the law had been lost. Up to this point, it had always been assumed that Article 4 of the Constitution contemplates a constitutional challenge to one provision of a statute: no more, no less. This assumption could have been challenged: it could, for example, have been argued that the object of challenge could be a group of statutory provisions that form part of the same statutory scheme, or part of a statutory provision. To elaborate: a. In Quek, it appears that the applicant took issue with the Transitional Provision ‘read with s 33B of the [Act]’,21 yet sought to challenge the constitutionality of the Transitional Provision only. The applicant did not seek to mount a broader challenge the entire sentencing regime as a whole, for example by seeking to argue that the rest of the sentencing regime was inseverable from the allegedly unconstitutional part, or by seeking to argue that the Transitional Provision and/or its effects were too narrow in scope. b. In Yong, the applicant did not seek to mount a narrower challenge by claiming that, though the law on caning was not unconstitutional in toto, it may be severed according to the fact patterns to which it would apply and held unconstitutional as applied to some of those cases. For example, the applicant might have argued that, without prejudice to the existence of the Caning Restriction Provision (which exempts women and older men from being caned), statutes providing for caning of offenders would be unconstitutional to the extent that they apply to younger men.22 In short: the applicants sought to challenge parts of statutory schemes, but had challenged the wrong parts by framing their challenges too broadly or narrowly. This would most likely not have made a difference to the ultimate outcome since the courts ultimately dismissed the constitutional challenges. Nonetheless, what these cases make clear is that it is important to identify correctly the statute, or part thereof, which is under challenge. This is particularly so when the ground of challenge is inequality23 because the right to equal treatment says nothing about what that treatment is—for instance, men and women would be treated equally, not only if neither were liable to caning, but also if both were. In such a case, the applicant cannot simply state that the statute is void because it leads to inequality; the applicant must state exactly what effect of the statute gives rise to the inequality, and hence what the remedy ought to be. The key development is that the courts implicitly agreed that it would in principle be possible to strike down the Transitional Provision (in Quek) and the Caning Restriction Provision (in Yong). These were provisions dictating how other statutory provisions were to be applied; in other words, each case represented a challenge to a provision that was but one part of a statutory regime. While the courts questioned the practical logic of the applicants’ seeking to sever these statutory regimes into parts and challenge some of those parts, the courts did not deny that such severance was in principle possible. 4. THE PRABAGARAN CASE As we have seen from Quek and Yong, having properly identified the object of challenge, the applicant must make clear what he wants the court to do to it and must ensure that the remedy will be of practical benefit to him. In Prabagaran, the applicants’ submissions on this issue were much more sophisticated. Prabagaran involved four people who had been sentenced to death for drug trafficking and who had failed to avail themselves of section 33B of the Act. They sought to argue that, because part of the scheme was unconstitutional, they ought to be sentenced to life imprisonment and caning instead. The court’s concern was that, if they were challenging section 33B of the Act, such a challenge would not result in changes to their death sentences: if the applicants are correct in their contentions that s 33B is unconstitutional, then this court would have to disregard s 33B as if it was never enacted and each of the applicants would have to be sentenced under the Second Schedule. We could not see how this argument could assist them.24 Therefore, the applicants stressed that their position was not that all the court could do was to strike down section 33B in toto. Instead, they thus pleaded several alternative cases, evidently seeking to cover all bases25: i. Sections 33B(2)(b) and 33B(4) were unconstitutional and should be struck down, leaving the rest of section 33B intact26; ii. Sections 33B(2)(b) and 33B(4) were unconstitutional and should be struck down, together with the Second Schedule, since all of these work in ‘inextricable tandem’27; iii. The entire sentencing regime under the Act was unconstitutional and should be struck down; and iv. Section 33B should be ‘cured’ by deleting section 33B(4) and the words ‘the Public Prosecutor certifies to any court that, in his determination’ in section 33B(2)(b).28 The court held that the constitutional challenges failed. The Court also remarked, however, that that, even if they had succeeded, they would not have assisted the applicants because none of the reliefs sought would be available. The only relief available, even if the alleged unconstitutionality had been made out, would have been to strike down section 33B as a whole—no more and no less. 5. THE TEST FOR SEVERABILITY: THE SEARCH FOR LEGISLATIVE INTENTION (A) Two Types of Legislative Intention: Primary and Secondary In reaching this conclusion, the court began by identifying the test for severability. Applying the remarks of the US Supreme Court in Alaska Airlines Inc v. Brock,29 the court held that a legislative provision is: a. Severable if ‘the truncated statute, with the unconstitutional portion excised, will operate in the manner that the legislature intended’30; b. Severable if the legislature ‘would have enacted the truncated statute with only the remaining provisions’31; c. But, conversely, not severable if the truncated statute ‘cannot function without [the unconstitutional part], at least in a manner that Parliament could not have contemplated’.32 Therefore, declared the court, in the exercise of severance, legislative intent is paramount. The reason for this is clear: to allow the courts to do so in a manner that is contrary to the intent underlying the passage of the provision in question would effectively confer upon the judiciary legislative powers and violate the principle of separation of powers. […] it must be shown to be Parliament’s intention behind the enactment of an Act that is found to be partially in breach of the Constitution that it should nevertheless continue to be given effect even after the severance and invalidity of some portions.33 As we will see, this passage is crucial to understanding the court’s approach. It places the focus squarely on the legislature’s intention as to whether what remains after severance should survive as law. Let us call this intention the ‘secondary intention’ of the legislature. It is distinct from what we will call the ‘primary intention’, which is simply the intention that the statute take effect exactly as it was enacted. The problem lies in how to ascertain the secondary legislative intention. We will now explain why, difficult as this exercise is, it is the right one to undertake. We will then explore how the court in Prabagaran went about performing this task. (B) Is the Search for Secondary Intention Misplaced? (i) Criticisms of the Search for Secondary Intent The approach of searching for the legislature’s secondary intention has been criticized as being purely hypothetical. For example, the High Court of Australia remarked in an early case: ‘What a man would have done in a state of facts which never existed is a matter of mere speculation, which a man cannot certainly answer for himself, much less for another’.34 In a similar vein, the High Court of Singapore had previously remarked that a similar act of severance in the context of contract law would not be in line with the drafters’ intentions; it was said to ‘amoun[t] to a unilateral variation of the parties’ obligations imposed by the court with the benefit of hindsight’.35 Such criticisms typically favour a test for severability that purports to focus on the text of the legislation rather than the secondary intention of the legislature, presumably because the former is more certainly ascertainable. Formulations of such a test include: a. ‘whether the statute with the invalid portions omitted would be substantially a different law as to the subject matter dealt with by what remains from what it would be with the omitted portions forming part of it’36; and b. whether the court, were it to effect severance, would be ‘effecting no change in the substantial purpose and effect of the impugned provision’.37 Prabagaran has demonstrated that such criticisms are misplaced, for two reasons: they are based on a misunderstanding of the nature of secondary intention; and they can amount to judicial frustration of legislative policy motivated by the erroneous assumption that legislative intention has no role to play merely because it is unclear. (ii) Misunderstandings of the Nature of Secondary Intention First, such criticisms are based on a misunderstanding of the nature of secondary intention. As the approach taken in Prabagaran shows, the search for secondary intention is not about what the legislature would have done upon its later learning that part of the legislation was unconstitutional (let us call this the legislature’s ‘hypothetical intention’); rather, it is about what the legislature has already, at the time of enactment, intended is to be done in the event that the legislation turns out to be unconstitutional. The former is an improper imputation to the legislature of an intention that, by definition, it never had. But the latter is perfectly permissible in principle; the real problem is how to ascertain the secondary intention of the legislature. One might respond as follows. The legislature’s primary intention has been expressed in the normal way in which the legislature expresses its intention, viz. through legislation; there is no such expression of the legislature’s secondary intention, if indeed it exists; therefore, the court’s understanding of the legislature’s secondary intention is at best based on scanty evidence and at worst a complete fabrication. But conversely, as Prabagaran illustrates, it is unobjectionable, and indeed quite proper, to refer to the legislature’s secondary intention if it has expressly made this intention clear, or if such secondary intention can be inferred from evidence. This is particularly so in Singapore, where it has been accepted that the legislature may communicate its intention not only through legislation but also through speeches made in Parliament.38 (iii) Judicial Frustration of Legislative Policy Second, criticisms of a search for secondary intention are essentially centred on the problem of ascertaining such intention, which is a question of evidence; but they go on to advocate alternative approaches which abandon the search for secondary intention altogether. This may lead the court to a conclusion that totally frustrates legislative policy, overstepping the judicial boundaries by showing fidelity to neither secondary nor primary legislative intention. To see why this is so, we will now examine the argument that the question of severability can be answered by reference only to the text of legislation and not to legislative intent itself. The following metaphor from Pitt v. Holt (albeit in a different context) provides useful language with which to describe this sort of argument: there is a difference between a rule of a ‘monolithic character’ and a ‘bundle of benefits of different characters’; ‘it is obviously easier to sever part of a bundle than part of a monolith’.39 In Owners of the SS Kalibia v. Wilson (‘Kalibia’), the High Court of Australia said that if Parliament had enacted that certain specified things, say A, B, and so on down to Z, might lawfully be done, the first half-dozen being within its legislative power and the remainder outside it. There the bad can be separated from the good and excised, and if there be left a law not substantially or radically different, dealing effectively with so much of the subject matter as is within the legislative power, the Act will be good, minus the invalid provisions eliminated; but if the ‘specified things’ were described by a broad phrase such as ‘coasting trade’, then that phrase will be held to be ‘indivisible’ and no such excision will be possible.40 The reason for this was said to be that the latter would show legislative ‘intention to put [A, B… Z] on the same footing’.41 In other words, to use the metaphor from Pitt v. Holt,42 the court held that ‘A, B…Z’ was a severable bundle, but ‘coasting trade’ was an inseverable monolith. Severing a bundle was, in the court’s view, merely an act of ‘constru[ing] the Statute so as to render it constitutional’, whereas purporting to sever a monolith would be ‘re-writ[ing]’ the statutory text.43 But this distinction is somewhat misleading, for the following reason: whether the law is a bundle or a monolith, severance will ultimately result in a change to the law, viz. the deletion of legislative content, which cannot be described as mere construction of a statute. The real question is how acceptable a change it is. Moreover, the mere fact that the wording of the legislation suggests that the law is a bundle is not conclusive of the extent to which severance will change the character of the law. To see why this is so, consider the court’s assertion that the use of a collective noun such as ‘coasting trade’ would indicate an ‘intention to put [A, B…Z] on the same footing’.44 This is correct. But does it follow that the words ‘A, B…Z’ would not show such an intention? In truth, the words ‘A, B…Z’ could evince one of at least two possible legislative intentions: [LI1] ‘that the rule apply to A, and to B… and to Z’45; or [LI2] ‘that, in the operation of the rule, A, B…Z be treated equally/together/for the same reasons’. Now suppose the law is held to be unconstitutional as applied to (say) Z. It is more justifiable if the legislature’s intention is [LI1], and less so if it is [LI2], to sever the part of the law relating to Z and hold the law valid insofar as it applies to A to Y. Consider the more concrete hypothetical example provided in Jumbunna Coal Mine v. Victorian Coal Miners’ Association (‘Jumbunna’), in which a legislature only has powers to ‘make laws for the government of the tropical part of South Australia’. How would the following two types of law stack up against this provision? [L1] A law ‘providing for the government of all tropical Australia as one whole’; [L2] A law ‘providing for the government of the tropical part of South Australia, and also…that the same provisions should apply to the tropical part of Queensland, and to the tropical part of Western Australia [etc.]’46 [L1] would evince legislative intention [LI2]; it would not be possible to sever the parts relating to South Australia from other parts. A possible justification of this is that the nature of task of governance changes radically and fundamentally depending on whether it is an entire country or only a part thereof which is being governed. But which legislative intention does [L2] reflect: is it [LI1] or [LI2]? The answer depends on the legislature’s intent with regard to the words ‘that the same provisions should apply’: a. On the one hand, it could simply be drafting shorthand, as though the legislature was simply ‘copying’ certain provisions from under the heading ‘South Australia’ and ‘pasting’ them, mutatis mutandis, under the headings ‘Queensland’ and ‘Western Australia’. b. On the other hand, it could evince legislative intent that it be significant that the laws of those three states be harmonized. In this latter case, deleting [L2] altogether might well show more fidelity to legislative intent than deleting only the words ‘and also…that the same provisions should apply to the tropical part of Queensland, and to the tropical part of Western Australia [etc.]’. The point is that the courts cannot abandon the search for secondary legislative intention and merely seek to uphold whatever legislation remains to the extent that it would be substantially the same law—which is the crux of the bundle-monolith distinction (i.e. that severing part of a legislative bundle does not render it substantially a different bundle). Rather, the question is whether what remains after severance would have been enacted as law at all. This is so for the following reason. The court’s ultimate duty is one of fidelity to the Constitution. The courts must be concerned with upholding constitutionally compliant legislation as much as they are with striking down unconstitutional legislation. The former, in turn, must be upheld on the basis that legislation is an instance of the legislature’s constitutional power to express its intention into law. Consequently, whether or not what is left after severance would be ‘substantially a different law’47 with a different ‘substantial purpose and effect’48 is inseparable from the question of what secondary legislative intent is, because whether the law is totally different—in other words, the salience of the differences between the law as enacted and the law as it is after excision—is itself a question of legislative policy for the legislature to decide. The content of what would be left behind following severance therefore cannot be the crux of the court’s inquiry; it is at most evidence—and not necessarily conclusive evidence—of secondary legislative intent. (iv) The Importance of Secondary Legislative Intent All this may seem rather artificial. It is not necessarily the case that the thought ever crossed the legislature’s mind that any legislation it makes could possibly be unconstitutional. The truth is that secondary legislative intent may often be, to an extent, fictional. But to the extent that it is a fiction, we have shown that it is an important one. It serves to focus our intention squarely on what the legislature intended, which it is the duty of the courts to put into action. In other words, even if we ask whether what would remain after severance would be ‘substantially a different law’, what we ought to mean by this is whether it would in the legislature’s view be ‘substantially a different law’. But this, in turn, is simply a proxy for secondary legislative intent, as we have defined it. Therefore, any remarks by the legislature as to what it considered to be essential features of the law are simply evidence from which the legislature’s secondary legislative intent may be inferred. Prabagaran shows us how, and on what basis, this process of inference may be performed. 6. HOW SECONDARY LEGISLATIVE INTENT MAY BE INFERRED: PRABAGARAN AS A CASE STUDY In Prabagaran, there was no direct evidence that the legislature had had a secondary intention in the following form: ‘If sections 33B(2)(b) and/or 33B(4) or parts thereof are held unconstitutional, then we wish as follows: …’ Yet the court inferred secondary legislative intention, both as to: a. the severability of section 33B as a whole; and b. the severability of parts of section 33B from other parts of section 33B. Looking at the text of the legislation in vacuo merely for the purpose of wondering whether the law would have been rendered ‘substantially a different law’ by severance would, as we have seen, be purely speculative. But Prabagaran shows us that scrutinizing the text and structure of the legislation can still be a useful task if it is done with a view to ascertaining secondary legislative intention. This is especially if this task is performed in the light of legislative history, which is another source of evidence of secondary legislative intention. (A) Evidence That Section 33B Was Severable from the Rest of the Act In Prabagaran, the court held that section 33B aimed to create an exception to the policy laying down the mandatory death penalty; and that the exception was severable from the rule. In other words: in enacting s 33B Parliament never intended a “major sea-change” to the mandatory death penalty regime. The change intended was a narrow and specific one. If, for whatever reason, that change is unconstitutional, then that change will not be effective and nothing in the existing law will be affected.49 This argument was supported by the structure of the legislation. It did not have the following structure: [S1] ‘Any person who trafficks in drugs in quantities above x, and is not a mere transporter of the drugs who has substantively assisted the Central Narcotics Bureau, shall be sentenced to death’. Rather, the Act had the following structure: [S2] ‘(i) Any person who trafficks in drugs in quantities above x shall be sentenced to death; (ii) but rule (i) shall not apply to a mere transporter of drugs who has substantively assisted the Central Narcotics Bureau.’ It appears that in the court’s view, the difference was that [S2(ii)] was a ‘carve-out’ from the general rule in [S2(i)], which provided an ‘additional sentencing option’ to that in [S2(i)].50 In other words: the applicants have each committed an offence which attracted the death penalty and that punishment would have to be imposed unless they could show that they satisfy the requirements provided in s 33B.51 From the ‘carve-out’ structure, the court concluded that the legislature positively intended the exception to stand or fall independently of the general rule. This is a lesson in legislative drafting: the difference between something being an exception to a rule, and something not even being covered by that rule in the first place is significant not only to issues such as the burden of proof52 but also to issues of severability in the event of partial unconstitutionality. Might it be said that this was not a foregone conclusion, i.e. that there is ambiguity similar to that between [LI1] and [LI2] in our hypothetical examples above? Probably not. The reason for the confusion between [LI1] and [LI2] is that [L1] and [L2] had the same legal effect, viz. to provide for the governance of a number of parts of Australia; it was the underlying legislative policy behind the two that differed. By contrast, in Prabagaran, [S1] and [S2] would arguably not have had the same legal effect. [S2] would place the burden of proof of the elements in limb (ii) on the accused, while [S1] would place the burden on the state to disprove those elements. In any event, even if the structure of the legislation was not conclusive, the court was fortified in its conclusion that the legislature intended section 33B to be severable for two other reasons: a. First, the Second Schedule to the Act (which prescribes the mandatory death penalty) existed first, and ‘[had been] in operation for an appreciable period of time’; whereas section 33B (creating exceptions to the mandatory death penalty) was added by way of a later addition to the Act.53 b. Second, the Minister moving the Bill that later became the Amendment Act said in the course of debates in Parliament: …We are maintaining the mandatory death penalty for the drug offences where it currently applies, but are making measured and carefully defined exceptions to allow for the courts to impose life imprisonment instead for couriers in cases of abnormality of mind or where substantive cooperation has been provided.54 While these reasons are persuasive, neither can, in and of itself, have been conclusive: a. The first reason would not have distinguished between legislative intent to develop the law on sentencing incrementally by adding an exception to a rule, and legislative intent to effect an overhaul of the entire sentencing regime by modifying the rule and its policy basis. b. The second reason was a statement of the Amendment Act’s intended practical effect, but does not explain how it would necessarily translate into the conclusion of law that the rule and the exceptions were therefore severable from one another. This is an interesting case study on how the legislature may be held to have communicated its secondary intention as regards the severability of legislation. More generally, the relative importance of the three types of evidence will depend not only on the particular legislation in question but also on the tools available within the legal system by which legislative intention is communicated generally. (B) Evidence that Sections 33B(2)(b) and 33B(4) Were Not Severable from the Rest of Section 33B An illustration of this is the fact that the legislative text did not provide any clue as to whether parts of section 33B were severable from the rest of section 33B. The court did not, however, therefore jump to the conclusion that section 33B must therefore stand or fall as a whole merely because, for example, it was intended to operate as a whole. Instead, the court appears to have proceeded on the basis that, in principle, parts of a legislative provision could be severed notwithstanding that they had been enacted as part of a whole. This is why the court’s methodology was to determine the secondary legislative intent as regards the particular parts of section 33B. The court found the answers in speeches made in Parliament, from which it concluded that section 33B was primarily intended…to disrupt the activities of drug trafficking syndicates by providing an incentive for offenders to provide information which would enhance the capabilities of law enforcement agencies in the war against drugs (emphasis added).55 Moreover, it appears that the court accepted that Parliament enacted section 33B on the basis that ‘the Public Prosecutor is better placed [than a court] to decide’ whether such information had been provided.56 In other words, the requirements in sections 33B(2)(b) and 33B(4) that a. the offender had ‘substantively assisted the Central Narcotics Bureau in disrupting drug trafficking activities’; and b. the Public Prosecutor be the one to determine whether the offender had rendered such assistance, were considered by Parliament to be central to section 33B. Therefore, the severance of these sub-sections would result in something ‘fundamentally different from what Parliament intended’57—the legislature had evinced a ‘clear and express intention against the result of the applicants’ proposed severance’.58 This finding is of crucial theoretical importance, as it highlights two key points made above: a. The question is not whether section 33B sans those sub-sections would be a fundamentally different law from section 33B as it was enacted, but rather whether it would in the legislature’s view be a fundamentally different law.59 This is why the court asked what would remain would be ‘fundamentally different from what Parliament intended’60 (emphasis added). b. The court’s task is to seek the secondary legislative intent that the legislature had had at the time of enactment, and not the hypothetical legislative intent that the legislature would form were it to learn after enactment that the legislation turned out to be unconstitutional.61 This is why the court emphasized that Parliament had had a ‘clear and express intention against the result of the applicants’ proposed severance’62 which was inferable from what had been said in Parliament. For completeness, it is worth stressing that the court was engaging in a fact-finding exercise, involving the weighing up of various pieces of evidence which did not all point in the same direction. For example, the applicants referred to several other speeches from Members of Parliament which they claimed showed, contrary to the court’s conclusion, that the legislature intended to create a new ‘holistic discretionary death penalty framework’ rather than a ‘mere carve-out’ to the general rule that the death sentence be mandatory.63 It is not our intention to comment on the correctness of the court’s conclusion on this point. Rather, our point is simply that, because the severability inquiry involves the search for secondary legislative intention, it engages the various theoretical debates as well as legal doctrines relating to the nature of legislative intention and the means by which it may be ascertained.64 7. WHAT, IF ANYTHING, IS TO BE SEVERED? There is a good reason for the approach of inquiring into the severability of parts of statutes and not only the whole. It is that, taking the contrary approach to its logical conclusion, an entire Act—even a wide-ranging one (such as Singapore’s Penal Code, which sets out the general part of the criminal law and defines most standard criminal offences)—could be struck down merely because of one objectionable provision, no matter how minor.65 However, proceeding on a part-by-part basis carries certain risks, arising from the fact that it is not true that, the more parts of legislation that are struck down, the better for the applicant. To the contrary, it might be better for others in the applicant’s position in future if the application fails than if it succeeds in part. In the simplest type of case involving severance, there is a law in two parts, A and B. The applicant, seeking to avoid the effect of B, claims that B is unconstitutional. As an alternative, the applicant also claims that: A is to be struck down for unconstitutionality; A and B are inseparable; therefore B is to be struck down too. That case is straightforward. The more difficult case is one such as Prabagaran, in which the applicant wants to avoid only part of the effect of B—in other words, B is sub-divisible into more parts, only some of which the applicant wishes to avoid. In such a case, it cannot be assumed that the applicant would necessarily wish to strike down as much of the law as possible. To illustrate this, consider the law in question in Prabagaran as broken down into parts: [A] Anyone who trafficks drugs in quantities above x shall be sentenced to death66; [B] However, some of those traffickers referred to in [A] may be sentenced to life imprisonment and caning instead, provided that the conditions in [C], [D], and [E] are met, namely67: [C] The Public Prosecutor issued a Certificate to the effect that the trafficker substantively assisted the authorities68; [D] The trafficker’s acts are restricted to transporting [etc.] drugs69; [E] The court exercises its discretion in favour of the accused.70 Of these parts, not two, but three parts, were in contention: [A], [B], and [C]. The applicant had essentially three alternative arguments: a. [C] is unconstitutional, and [A] and [B] are inseverable from [C] (and [D] and [E] are meaningless in the absence of [B]). So all five parts should be struck down. Consequently, the sentence should be quashed. b. [C] is unconstitutional, so [C] should be struck down. Consequently, the applicants should be re-sentenced according to the principles in [A], [B], [D], and [E] only. c. [C] is unconstitutional, so [C] should be modified by removing the words ‘The Public Prosecutor issued a Certificate to the effect that’. Consequently, the applicants should be re-sentenced according to the principles in [A], [B], [C] as modified, [D], and [E]. What the applicant would not want was for [B] (and consequently [C], [D], and [E]) to be struck down, but [A] left intact. That would leave a mandatory death penalty regime with no exceptions. It would not only not help the applicant, but harm other drug traffickers who would otherwise at least stand a chance to escape the death penalty. Given this, it is worrying that the court described the applicants’ initial contention as being merely that ‘s 33B is unconstitutional’,71 which would lead to precisely the result just described. It is unclear whether this accurately reflects the applicants’ (initial) submission or whether it reflects a misunderstanding on the courts’ part. It does not matter, for our objective is not to assign blame—in any event, the applicants eventually clarified exactly what reliefs they sought—72 but to underscore the importance of an applicant making absolutely clear what he seeks to be held unconstitutional and severed, and what he wishes to be left intact. If the importance of this point was not clear from Quek and Yong, it ought to be perfectly clear now. 8. CONCLUSION While Prabagaran is undoubtedly presently Singapore’s leading case on the doctrine of severability in constitutional review, it does not purport to put forth an exhaustive account of the law on severability. Moreover, it offers but one illustration of such a multifaceted doctrine—one wonders, for instance, how the court would have reacted had the evidence of secondary legislative intention been less clear (or had the permissible sources of evidence of the same been different from what they are in Singapore). Nonetheless, we have shown that, beyond the Singapore jurisdiction, the law on severance is necessarily based on a theory of legislative intention, which in turn necessitates a theory of evidence of legislative intention. It is hoped that we have provided some food for thought on these theories by analysing a case that we hope will turn out to be fertile ground for future development. 9. APPENDIX: THE MANNER IN WHICH SEVERANCE IS EFFECTED Because of its conclusions on severability and the outcome of the constitutional challenges, the Singapore courts have never had to confront the issue of how severance is to be effected. Nonetheless, the court in Prabagaran made some potentially concerning remarks, on which we will now briefly comment. (A) Textual and Substantial Severance One might argue that implicit in the courts’ remarks (and, perhaps, counsel’s pleaded cases) in Quek and Yong is an assumption as to how the striking down of unconstitutional legislation is to take place: namely, by deleting words. The courts did not explore the question of whether striking down can take place by other means, such as by modifying or even adding words. Such an assumption, had it been made, would be erroneous. Nothing in the words ‘to the extent of the inconsistency’ in Article 4 of the Constitution requires that the ‘extent’ be measured according to the individuation of provisions, clauses, or phrases in the statute. It is possible, for example, that a single statutory provision is unconstitutional to the extent that it covers certain types of case. In other words, to use the language of the majority of the House of Lords in Director of Public Prosecutions v. Hutchinson (‘Hutchinson’), though the provision might not be ‘textually severable’, it might be ‘substantially severable’.73 In such a case, according to English case law, the court’s powers are not limited to ‘running a blue pencil through the offending part’ of the statute.74 The logic is that, otherwise, the court’s powers would be constrained by the rules of English grammar and the division of the legislation into sentences and sections, which would be ‘purely artificial’.75 Rather, the court can ‘modify the text in order to achieve severance’.76 As the court in Hutchinson impliedly emphasized, this is not the same as modifying the law.77 Rather, it is an act of modifying the text in order to keep the law ‘unchanged in its legislative purpose, operation, and effect’,78 but changed only in scope.79 For example, if (hypothetically) the Caning Restriction Provision (with which Yong was concerned)80 were found to be unconstitutionally discriminatory on grounds of gender, such discrimination could be remedied by adding the words ‘who are more than 50 years of age at the time of infliction of the caning’ after ‘women’. (B) Did Prabagaran Reject the Possibility of Substantial Severance? It is unclear what the status of substantial severance is after Prabagaran. However, the court appears to have made certain remarks to the effect that substantial severance would not be possible in Singapore. To see why, it is necessary to examine the applicants’ pleaded case in more detail. Of the various alternative forms of relief they sought,81 one called for an ‘amend[ment]’ of section 33B, ‘such that it is the court, and not the [Public Prosecutor], that determines if an offender has substantively assisted the Central Narcotics Bureau in disrupting drug trafficking activities’.82 This attempt at ‘amend[ment]’ was premised on Article 162 of the Constitution, which provides that certain laws can be construed…with such modifications, adaptions, qualifications and exceptions as may be necessary to bring them into conformity with this Constitution. The applicants further submitted that Article 162 was a ‘guide’ to the operation of Article 4, and guided the way to the use of the ‘well-known techniques of severance, reading in, reading down and striking out’ which were part of a process known as ‘remedial interpretation’.83 As authority for the availability of such techniques of ‘remedial interpretation’, the applicants cited a Hong Kong case, HKSAR v. Lam Kwong Wai (‘Lam’).84 In response, the court made the following remarks on the courts’ powers under Article 4: a. The court remarked that Lam was inapplicable, as Hong Kong’s Basic Law had no equivalent of Singapore’s Article 4 and Article 162. The court said that ‘there is no need for the implication of remedial powers’ into Article 162 or Article 4—this suggests that Article 4 and Article 162 themselves, on their own terms, exhaustively set out the courts’ powers when faced with unconstitutional legislation. b. The court also remarked that ‘modificatio[n], adaptatio[n], qualificatio[n] and exceptio[n]’ mentioned in Article 162 are only exercisable under Article 162 and not Article 4: the purport of Art 162 is not really that different from Art 4 in respect of the laws to which they apply, but only in respect of the manner in which they apply. Article 162 allows the court to construe all laws in conformity with the Constitution while Art 4 provides the power to void such laws…85 c. The court held that Article 162 of the Constitution does not apply to section 33B as it applies only to laws which existed or ‘had already been enacted but not yet brought into force at the commencement of the Constitution’.86 In other words, if at all the applicants were to succeed, it would be pursuant to Article 4 and not Article 162. d. Finally, the court described the power available under Article 4 as a ‘power to void’ unconstitutional laws,87 subject to the ‘doctrine of severability’.88 The effect of these remarks might be that the act of ‘void[ing]’ unconstitutional laws and the ‘doctrine of severability’ do not include the acts of ‘modificatio[n], adaptatio[n], qualificatio[n] and exceptio[n]’ or ‘reading in [and] reading down’ legislation. This would suggest that the only power under Article 4 consists of the deletion of words from a statute. It is unclear whether the court intended to express this conclusion, which would be questionable. As we have seen, the court accepted that Article 4 of the Constitution allows for the ‘doctrine of severability’.89 We have also seen that the doctrine of severability must include not only textual severability but also substantial severability—which does allow for the modification of the words of a statute. There are limits to what severance can do (for example the court cannot end up playing the role of legislator), but these limits do not foreclose the notion of substantial severability in principle. It is hoped that the Singapore courts will have occasion to clarify this point in future. 1 Constitution of the Republic of Singapore (1999 Rev Ed). 2 The Legislature of Singapore consists of a unicameral Westminster-style Parliament and the President; as Art. 58(1) of the Constitution provides, ‘the power of the Legislature to make laws [is] exercised by Bills passed by Parliament and assented to by the President’. In this paper, ‘legislature’ and ‘Parliament’ will be used interchangeably. 3 Tan Eng Hong v. Attorney-General  SGCA 45,  4 SLR 476 . 4 Prabagaran a/l Srivijayan v. Public Prosecutor  SGCA 67,  1 SLR 173. 5 Misuse of Drugs Act (Cap 185, 2008 Rev Ed). 6 Sections 5(1) and 7 of the Act (n 5). The difference between the two offences is not material for present purposes. 7 That is, strokes of the cane. 8 This refers to a mandatory penalty rather than a maximum penalty: Nguyen Tuong Van v. Public Prosecutor  SGCA 47,  1 SLR(R) 103 . 9 Ong Ah Chuan v. Public Prosecutor [1979–1980] SLR(R) 710,  1 AC 648 (PC); Nguyen (n 8); Yong Vui Kong v. Public Prosecutor  SGCA 20,  3 SLR 489. A number of applications for judicial review of related decisions also failed: Thiruselvam s/o Nagaratnam v. Public Prosecutor  SGCA 13,  1 SLR(R) 362 and Ramalingam Ravinthran v. Attorney-General  SGCA 2,  2 SLR 49 (decisions to prosecute); and Yong Vui Kong v. Attorney-General  SGCA 9,  2 SLR 1189 (decision by the President not to grant clemency to an offender). 10 The parts of this provision not reproduced here pertain to another discretionary sentencing regime for offenders who were ‘suffering from such abnormality of mind…as substantially impaired [their] mental responsibility’ for their offences. 11 Muhammad Ridzuan bin Mohd Ali v. Attorney-General  SGCA 53,  5 SLR 1222. 12 Quek Hock Lye v. Public Prosecutor  SGCA 7,  2 SLR 563. 13 Act 30 of 2012. 14 Quek (n 12) . 15 Quek (n 12) . 16 Quek (n 12) . 17 Quek (n 12) . 18 Yong Vui Kong v. Public Prosecutor  SGCA 11,  2 SLR 1129. 19 Cap 68, 2012 Rev Ed. 20 Yong (n 18) . 21 Quek (n 12) . 22 There is authority supporting the view that the court may make a declaration to this effect: Dunkley v. Evans  1 WLR 1522 (DC) 1525G. 23 Art 12 of the Constitution. 24 Prabagaran (n 4) . 25 Prabagaran (n 4) . 26 Prabagaran (n 4) . 27 Prabagaran (n 4) . 28 Prabagaran (n 4) . 29 480 US 678 (1987). 30 Prabagaran (n 4) . 31 Prabagaran (n 4) . 32 Prabagaran (n 4) . 33 Prabagaran (n 4) –. 34 R v. Commonwealth Court of Conciliation and Arbitration, ex parte Whybrow & Co (1910) 11 CLR 1 (HCA) 26–27; see also 35. 35 Lek Gwee Noi v. Humming Flowers & Gifts Pte Ltd  SGHC 64,  3 SLR 27 . 36 Whybrow (n 34) 27. 37 Director of Public Prosecutions v. Hutchinson  2 AC 783 (UKHL) 811G. While Hutchinson concerned the judicial review of subsidiary legislation on the grounds that it was allegedly ultra vires the primary legislation authorizing its creation, there is no reason why the reasoning within it should not apply, mutatis mutandis, to the judicial review of primary legislation on grounds of unconstitutionality (save, of course, for the fact that no such judicial review of primary legislation is possible in English law). 38 Reference to such speeches for the purpose of interpreting legislation, in the sense of ascertaining the ‘meaning of the provision’, is explicitly authorized by section 9A of the Interpretation Act (Cap 1, 2002 Rev Ed). Arguably, the courts have gone even further and referred to such speeches even when performing tasks other than interpretation stricto sensu: see e.g. Yuen Wai Loon v. Public Prosecutor  SGHC 160,  4 SLR(R) 176 and Tan Seet Eng v. Attorney-General  SGCA 59,  1 SLR 779, in which legislative debates were referred to not to ascertain the meaning of words, but rather to ascertain the scope of their applicability. 39 Pitt v. Holt  UKSC 26,  2 AC 108 . 40 Owners of the SS Kalibia v. Wilson (1910) 11 CLR 689 (HCA) 701–702. 41 Kalibia (n 40) 699. See also Pidoto v. Victoria (1943) 68 CLR 87 (HCA) 108: ‘it would be necessary to consider whether such reading down would alter the policy or operation of the statute with respect to the cases which, after the reading down, would still remain within its terms’. 42 n 39 above. 43 Kalibia (n 40) 701–702. 44 Kalibia (n 40) 699. 45 This is especially strongly so when the intention is that the rule apply to as many of A, B…Z as possible: consider Jersey Fishermen’s Association Ltd v. States of Guernsey  UKPC 30,  1 LRC 198, –, where a law applying to the waters within 12 miles of Guernsey was, following excision, held valid to the extent that it applied to the waters within 3 miles of Guernsey; the law was not struck down in toto because the legislative intent was to ‘restrict entry and fishing in the Bailiwick’s territorial waters’ generally. 46 Jumbunna Coal Mine v. Victorian Coal Miners’ Association (1908) 6 CLR 309 (HCA) 316. 47 Whybrow (n 34) 27. 48 Hutchinson (n 37) 811G. 49 Prabagaran (n 4) . 50 Prabagaran (n 4) . 51 Prabagaran (n 4) . 52 See generally G Williams ‘Offences and Defences’  LS 2, 233–9. 53 Prabagaran (n 4) , . 54 Prabagaran (n 4) . 55 Prabagaran (n 4) . 56 Prabagaran (n 4) . 57 Prabagaran (n 4) . 58 Prabagaran (n 4) . 59 See section 5(B)(iv) above. 60 Prabagaran (n 4) . 61 See section 5(B)(ii) above. 62 Prabagaran (n 4) . 63 Prabagaran (n 4) . 64 For example, in Singapore, the tension is between the search for the intention of Parliament and not just individual MPs on the one hand (see B J Ong ‘Developments in the Law on Constitutional and Statutory Interpretation’ Singapore Law Watch Commentary (Issue 1/September 2013) 5; and the difficulties inherent in ‘comparing one Parliamentary statement with another’, especially when the statements are not ‘clear and unequivocal’ and not ‘directed to the very point in question in the litigation’: Attorney-General v. Ting Choon Meng  SGCA 6,  1 SLR 373  (see also –). In the particular context of considering Parliamentary speeches, the Singapore courts must both ‘determine whether Parliamentary debates are capable of giving assistance such that they should be “considered”; and if so,  determine what weight should be placed on them’: Tan Cheng Bock v. Attorney-General  SGCA 50  (see also –). Note also the English common-law tradition that focuses on speeches by Ministers (or promoters of Bills) in particular: Pepper (Inspector of Taxes) v. Hart  AC 593 (UKHL) 634E (Lord Browne-Wilkinson); for criticism of this provision, see Aileen Kavanagh, ‘Pepper v. Hart and Matters of Constitutional Principle’ (2005) 121 LQR 98, especially 104ff. 65 I am grateful to Mr Lim Sing Yong for this point. 66 Second Schedule to the Act (n 5). 67 Section 33B(1)(a) of the Act (n 5). 68 Section 33B(2)(b) of the Act (n 5). 69 Section 33B(2)(a) of the Act (n 5). 70 Section 33B(1)(a) of the Act (n 5). 71 Prabagaran (n 4) . 72 See the text accompanying n 25 above. 73 Hutchinson (n 37) 804–805. 74 This phrase, from Attwood v. Lamont  3 KB 571 (EWCA) 578, is what is meant in Hutchinson (n 37) 804D by the ‘blue pencil’ test. 75 Commonwealth v. Hitchings (1855) 5 Gray 482 (Supreme Judicial Court of Massachusetts) 486, referred to in Jumbunna (n 43) 318. 76 Hutchinson (n 37) 811G. 77 Hutchinson (n 37) 804C. 78 Hutchinson (n 37) 804G. 79 Hutchinson (n 37) 811F. 80 See section 3(B) above. 81 See text accompanying n 25 above. 82 Prabagaran (n 4) . 83 Prabagaran (n 4) , citing HKSAR v. Lam Kwong Wai (2006) 9 HKCFAR 574 –. 84 Lam (n 84). 85 Prabagaran (n 4) . 86 Prabagaran (n 4) . 87 Prabagaran (n 4) . 88 Prabagaran (n 4) . 89 Prabagaran (n 4) . © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: email@example.com.
Statute Law Review – Oxford University Press
Published: Jan 24, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera