HIGH COURT OF AUSTRALIA KIEFEL CJ, BELL, GAGELER, KEANE, NETTLE, GORDON AND EDELMAN JJ IL APPELLANT AND THE QUEEN RESPONDENT IL v The Queen [2017] HCA 27 9 August 2017 S270/2016 ORDER 1. Appeal allowed. 2. Set aside the orders of the Court of Criminal Appeal of the Supreme Court of New South Wales made on 8 April 2016 and in their place order that the appeal to that Court be dismissed. On appeal from the Supreme Court of New South Wales Representation B J Rigg SC with R C Pontello for the appellant (instructed by Benjamin & Leonardo Criminal Defence Lawyers) S C Dowling SC with H R Roberts for the respondent (instructed by Solicitor for Public Prosecutions (NSW)) Notice: This copy of the Court's Reasons for Judgment is subject to formal revision prior to publication in the Commonwealth Law Reports. CATCHWORDS IL v The Queen Criminal law – Murder and manslaughter – Where appellant and deceased engaged in joint criminal enterprise – Where act causing death committed in course of joint criminal enterprise – Where Crown could not exclude possibility that deceased had committed act causing death – Where appellant charged with murder or manslaughter of deceased – Whether s 18(1) of Crimes Act 1900 (NSW) encompasses self-killing. Criminal law – Joint criminal enterprise liability – Whether acts or liability for actus reus of crimes committed in course of joint criminal enterprise attributed to co-participant – Whether act of deceased causing death attributable to appellant. Words and phrases – "attribution of acts", "complicity", "constructive murder", "derivative liability", "felo de se", "felony murder", "joint criminal enterprise liability", "primary liability", "rules of attribution", "self-murder", "suicide". Crimes Act 1900 (NSW), s 18. KIEFEL CJ, KEANE AND EDELMAN JJ. The legal issue on this appeal The background, facts, and legislative provisions are set out in the judgment of Bell and Nettle JJ. We agree that the appeal should be allowed on the first ground of appeal. We also agree with the orders that their Honours propose. However, we reach the conclusion that the appeal should be allowed on the first ground for different reasons. The offences of murder and manslaughter in s 18 of the Crimes Act 1900 (NSW) require that one person kill another person. Section 18 is not engaged if a person kills himself or herself intentionally. Nor is it engaged if the person kills himself or herself in the course of committing a crime punishable by imprisonment for life or for 25 years or by an unlawful and dangerous act. This conclusion is sufficient to allow the appeal on the first ground. It is unnecessary to consider the second ground of appeal or the notice of contention, which concerned whether the killing was "malicious" within s 18(2)(a). Since we conclude that murder in s 18 does not apply to circumstances involving self-killing, it is not strictly necessary for us to consider the operation of the rules of attribution when co-offenders act in concert. It suffices to observe that we agree with the assumption upon which this case was conducted by the parties, namely, that when two or more persons act in concert to effect a common criminal purpose, it is the acts of each person to effect their common purpose which are attributed to the others. The decision of this Court in Osland v The Queen establishes that it is the acts which are attributed in this scenario, it is not the liability. Nor is it the actus reus of a notional offence. The way this issue arises As Bell and Nettle JJ explain, the Crown's case was that the appellant and the deceased were participants in a joint criminal enterprise to manufacture a large commercial quantity of a prohibited drug, methylamphetamine. A ring burner attached to a gas cylinder was lit in a small and inadequately ventilated room during the commission of that crime, causing a fire which led to the death of the deceased. The Crown relied upon the category of murder in s 18(1)(a) of the Crimes Act, commonly referred to as "felony murder" or "constructive murder", in support of its case that the appellant murdered the deceased. That was the focus of submissions on this appeal. Alternatively, the Crown alleged that the appellant was guilty of manslaughter of the deceased by an unlawful and dangerous act. Section 18 is set out later in these reasons but, in summary, the elements of that section relevant to this appeal concerning the proof of murder are the following: (1) an "act of the accused ... causing the death charged"; (2) which "was done ... during ... the commission, by the accused, or some accomplice ... of a crime"; and (3) the crime is one which is punishable by imprisonment for life or for 25 years. The Crown's case in respect of those three requirements was that: (1) the act of the accused was the lighting of the ring burner which caused the death of the appellant's co-participant; (2) that act was done during the commission by the appellant, or the co-participant, of the crime of manufacture or production of a large commercial quantity of a prohibited drug contrary to s 24 of the Drug Misuse and Trafficking Act 1985 (NSW); and (3) that crime is punishable by imprisonment for life (s 33(3)(a)). There was no dispute that requirements (2) and (3) were satisfied. However, the Crown could not prove whether it was the appellant or the deceased who lit the ring burner. So the Crown submitted at trial that even if it were the deceased who lit the ring burner, his act could be attributed to the appellant with the result that his act could be the "act of the accused" for the purposes of murder or manslaughter under s 18 of the Crimes Act. The Crown relied upon rules of attribution commonly known as "joint enterprise liability". The most elementary difficulty with the Crown case is the assumption upon which it was based, that s 18 applied in a case of self-killing. Properly construed, s 18 is not engaged in a circumstance in which a deceased accomplice killed himself or herself. It was, therefore, not engaged in this case. Questions of attribution need not arise. The origins of s 18 of the Crimes Act Section 18(1) of the Crimes Act originated in s 9 of the Criminal Law Amendment Act 1883 (NSW). In Ryan v The Queen, Windeyer J quoted with approval the comment from two Draftsmen of that Act, Sir Alfred Stephen and Alexander Oliver, that apart from banishing the expression "malice aforethought", the New South Wales definition of murder did not otherwise alter the common law. Section 9 of the 1883 Act (later s 18(1) of the Crimes Act) was therefore intended to be a restatement of the common law relating to murder and manslaughter "but shorn of some of the extravagances of malice aforethought and constructive malice". In order to understand the operation of s 18, it is necessary, therefore, to appreciate what was meant by murder and manslaughter at common law. The common law position prior to 1883 was that a homicide which was neither justifiable nor excusable was a felony. For centuries, the common law had divided the felony of homicide into three categories. As Blackstone explained in 1769, the first category was felo de se. This was a "peculiar species of felony, a felony committed on oneself". Blackstone continued: "The other species of criminal homicide is that of killing another man. But in this there are also degrees of guilt, which divide the offence into manslaughter, and murder." (emphasis in original) In the 1800 edition of his manuscript, Hale drew the same distinction. He wrote of the basic division between a felony "which concerns the loss of life happening to a man's self" and a felony concerned with the loss of life "happening to another". Hale explained that the first of these, involving a voluntary act, was felo de se (ie suicide). In the 1817 edition of Sir Edward Coke's Institutes, the distinction between self-killing and the killing of another was reiterated. Although using the looser language of "murder of a man's self" interchangeably with felo de se, it was plain that Sir Edward Coke treated self-killing separately from murder. However, in A History of the Criminal Law of England, Sir James Fitzjames Stephen sought to assimilate suicide and murder despite recognising the distinction in Sir Edward Coke's writing between the two offences. This was apparently for the rhetorical purpose of agitating for the exclusion of felo de se from homicides altogether. Nevertheless, even Stephen was forced to recognise differences between felo de se and murder. For instance, he contrasted accessories to suicide and accessories to murder when he said that the "abetment of suicide may, under circumstances, be as great a moral offence as the abetment of murder" (emphasis added) but that the abetment of suicide involves much less public danger than the abetment of murder. In 1824, in the eighth edition of Hawkins' A Treatise of the Pleas of the Crown, felonious homicide was again divided into the same three categories, separating self-killing and murder: "[(1)] felo de se, or felonious homicide of a man's self; [(2)] murder, which is the killing of another with malice aforethought, either express or implied; and [(3)] manslaughter, which is the killing of another without premeditation or malice aforethought." The same distinction was reflected in the results of the decided cases for two centuries. Almost without exception, arguments which attempted to treat felo de se as a type of murder, based on the inaccurate and loose language of "self-murder", were consistently rejected by the courts. In 1660, in R v Ward, the crime of felo de se was held not to be murder and therefore capable of being pardoned. The view which prevailed in that case was that "felo de se and murder are distinct things, and distinctly treated of by all authors". The same point was made in Tombes v Ethrington. In 1832, in R v Russell, the Crown argued that a prisoner could be tried as an accessory before the fact of "self-murder" where the self-killing was in the course of a felonious act. Nine judges of the Court held that although the prisoner was an accessory before the fact of the offence of felo de se, neither by the preceding common law nor under the 1826 statute, which conferred jurisdiction to try felonies committed abroad, could a person be tried as an accessory to murder where the killing was a self-killing. Tindal CJ observed that another statute, the Burial of Suicide Act 1823 (4 Geo IV c 52), had used the term "felo de se" rather than the word "murder" throughout, and that this confirmed that "'murder' means murder of another only". Again, in 1862 in R v Fretwell, the question arose whether a conviction for murder could be upheld where the prisoner had, at the deceased's request, procured a poison for her to attempt an abortion. The Chief Justice explained that it was not necessary to decide whether the deceased was felo de se. It was sufficient for the Court to conclude that the prisoner was not guilty of murder. It was not necessary for the Court to consider whether the prisoner could have been convicted of felo de se by a derivative liability for aiding and abetting. Conviction of felo de se was not a conviction for murder. Yet again, in 1862 in R v Burgess, the Court held that an attempt to commit suicide is not an offence within the terms of a statute creating the offence of an "attempt to commit murder". The issue arose by a case stated to determine whether the Court of Quarter Sessions had jurisdiction in a case of attempted self-killing. If the accused's attempt to kill herself were attempted murder within the statute then the Court of Quarter Sessions would have no jurisdiction. The Crown argued that the attempted self-killing was not attempted murder and that it remained only a common law misdemeanour within the jurisdiction of the Court of Quarter Sessions. The entirety of the Crown submissions by Poland (later Sir Harry Poland QC) were to the effect that the crime of felo de se is separate from murder. His submissions began with the assertion that these two crimes were treated separately in all the text books. He concluded his submissions with an aside that Sir John Jervis had said, in his A Practical Treatise on the Office and Duties of Coroners, that it was not necessary to use the word "murdravit" in an inquisition for suicide. In a further passage from that work, Sir John Jervis explained that although an inquisition for felo de se should conclude with the words "sic seipsum murdravit" (and so murdered himself), there had been inquisitions where these words had not been added, including the decision in Hales v Petit. The final submission by Poland about coronial inquests did not assist the Crown case. As Sir John Jervis had observed, and as the reporter of Toomes v Etherington expressed the point, the omission of the concluding words of "self-murder" in a coronial inquest was explicable because, unlike the killing of another person, there were no "different degrees subject to different punishments". This point was noticed by the Court during argument. In response to the submission concerning inquisitions by coroners, Pollock CB postulated that the absence of a reference to murder in an inquisition for suicide was because there was no offence of "self-manslaughter". To this, Williams J added that there are "no degrees in self-destruction" so that if "a man feloniously kill himself, it must be self-murder". His Honour also added that the presentment in Toomes v Etherington was equivalent to an allegation of self-murder. Poland, for the Crown, accepted this explanation of the lack of need for words of "murder" in coronial inquests but reiterated his principal submission by referring to a number of cases where felo de se was treated separately from murder. The Court reserved its decision for a week, indicating the importance of the issue. The reasons for decision of the five judges were delivered by Pollock CB. His Lordship concluded that an attempt to commit suicide was not an attempt to commit murder within the Offences against the Person Act 1861 (24 & 25 Vict c 100). However, like the submissions of Poland for the Crown, his Lordship's reasoning was not based upon a mere matter of statutory construction. He said: "[A]ttempting to commit suicide is not attempting to commit murder within that statute. If it were, it would follow that any one attempting to commit suicide by wounding himself must be indicted for the offence of wounding with intent to commit murder, which until very recently was punishable with death. There is a vast difference between inflicting a wound on another and inflicting a wound on oneself with that intent." The distinction between these three different types of homicide, namely (i) self-killing; (ii) manslaughter (of another); and (iii) murder (of another), was therefore well known in 1883 when the progenitor to the Crimes Act was enacted. Although some cases were difficult to classify, the distinction was one which was constantly drawn. Two examples given in Hawkins' A Treatise of the Pleas of the Crown illustrate the careful distinction that was drawn between self-killing and murder in difficult cases. The first example was where one person "kills another upon [the other's] desire or command". The killing of another in these circumstances was treated as murder and "the person killed [was] not looked upon as a felo de se". In contrast, the second example was where one person induces another to buy a poison which they both drink with the intention of killing themselves. The purchaser of the poison does not die but, nevertheless, since he was not the inducer, the purchaser was not a murderer and the deceased was treated as a felo de se. In between these two examples was the difficult case where two people reached a joint agreement to commit suicide. If only one died, the agreement of the survivor was sometimes held to be sufficient to treat the survivor as having a derivative liability as "principal in the second degree". As we explain later, the derivative liability was not a primary liability. It was a derived liability for the crime committed by the person who was primarily liable. As we have explained, that crime could only have been felo de se even if, on occasion, it was loosely, and inaccurately, described as "self-murder". The categories were reiterated in Kenny's criminal law text in 1902. In that text Kenny again distinguished between, on the one hand, (i) felo de se, that is, "a suicide that takes place under such conditions as to be criminal", and, on the other hand, two other categories, being (ii) manslaughter, that is, "killing another person unlawfully, yet under conditions not so heinous as to render the act a murder", or (iii) murder, involving the killing of another person including with the "distinctive attribute" of malice aforethought. In relation to the killing of another which amounted to murder, the difficulty with the expression "malice aforethought" was that "malice" was misleading, and "aforethought" was false. The expression was an unfortunate description of six different species of mens rea which were sufficient to establish murder. Although the mens rea for murder, and particularly for felony murder, was amended in s 9 of the Criminal Law Amendment Act 1883 and s 18 of the Crimes Act, these sections remained concerned only with murder and manslaughter. They were not concerned with felo de se (self-killing). Other statutes, enacted at the same time, dealt with self-killing. The species of murder which was felony murder was controversial. As Kenny explained, "if a thief gives a man a push with intent to steal his watch, and the man falls to the ground and is killed by the fall − or if a man assaults a woman, with intent to ravish her, and she, having a weak heart, dies in the struggle − such a homicide would ... be murder". The 1883 Act preserved felony murder but confined it to capital offences or those punishable by life imprisonment. Although qualifying the common law, this was based upon a common law conception of murder. As explained by the two Draftsmen of the 1883 Act: "The accidental taking of life, by a person committing (or about to commit) a felony of any kind, is by the Common Law murder. Under the ninth section [later s 18] it will not amount to that crime, unless the felony was a capital one, or punishable by penal servitude for life." The terms of s 18 of the Crimes Act are not concerned with self-killing Section 18 of the Crimes Act provides as follows: "Murder and manslaughter defined (1) (a) Murder shall be taken to have been committed where the act of the accused, or thing by him or her omitted to be done, causing the death charged, was done or omitted with reckless indifference to human life, or with intent to kill or inflict grievous bodily harm upon some person, or done in an attempt to commit, or during or immediately after the commission, by the accused, or some accomplice with him or her, of a crime punishable by imprisonment for life or for 25 years. (b) Every other punishable homicide shall be taken to be manslaughter. (2) (a) No act or omission which was not malicious, or for which the accused had lawful cause or excuse, shall be within this section. (b) No punishment or forfeiture shall be incurred by any person who kills another by misfortune only." It would be a strong thing to hold that, under the common law in 1883 upon which this section was based, the appellant was guilty of murder when no case can be found which held that a person in the position of the appellant, ie a person whose accomplice unintentionally killed himself in the course of carrying out a joint criminal enterprise, was guilty of murder. Consistently with its origins in the common law, the text and context of s 18 also confirm that the concern of the section was not with the killing of oneself. First, when it was enacted s 18 was immediately followed by the penalty in s 19, which provided that the person who commits murder "shall be liable to suffer death". That plainly indicated that the "murder" the section was concerned with was the killing of another person. Secondly, the instance of murder involving "intent to kill or inflict grievous bodily harm upon some person" (emphasis added) contemplated that "some person" was some person other than the person causing the death. Indeed, on a literal reading of s 18(1) the "person" whose death is caused is differentiated from the "accused" and the "accomplice". The language in which s 18(1) is cast does not contemplate that the accomplice of the accused might be the person whose death was caused by the accused or the accomplice. Thirdly, s 18(2)(b) is an express acknowledgement that the punishment or forfeiture in relation to killing, other than by misfortune, is concerned with the killing, by murder or manslaughter, of "another". The short point is that the murder "taken to have been committed" and "[e]very other punishable homicide" taken to be manslaughter to which s 18 refers require the killing by one person of another. Section 18 is not concerned with the circumstance of a person who kills himself or herself intentionally. Nor is it concerned with a person who kills himself or herself accidentally. It follows that the offence of murder is not committed where a person kills himself or herself in an attempt to commit, or during or immediately after the commission of, a relevant crime. Nor is the offence of manslaughter committed when a person kills himself or herself in some other way. Section 18 did not create such new offences. Nor could the section be engaged, and such offences created, by attributing to another person an act which caused a self-killing. Attribution Our conclusion that s 18(1) was not engaged by an act of the deceased lighting the ring burner is sufficient to allow the appeal. However, we record our agreement with the assumption of the parties that the usual rules of attribution in criminal law, sometimes described as "joint enterprise liability", apply. On the assumption (which was not in dispute on this appeal) that those rules apply to s 18(1), an act done by one participant in the course of effecting a common criminal purpose, which was incidental to that purpose, can be attributed to the other participant under s 18 of the Crimes Act. This makes the act of the other participant an act for which the accused is personally responsible. In relation to murder, the attribution of an act causing personal responsibility for the other participant is, in summary terms, an "act of the accused" within the meaning of s 18(1)(a) although, plainly, attribution does not mean that the actual act is committed by the accused. In R v Surridge, Jordan CJ explained, in the context of felony murder, that where the act to be attributed is an act in the course of a common criminal purpose: "[I]t is necessary, in order that the person who is an accomplice only may be guilty of murder, that it should have been within the common purpose of both that ... a crime [punishable by imprisonment for life or for 25 years] should be committed, and the cause of the death must have been something done by the other in an attempt to commit or during or immediately after the commission of that ... crime." It is not necessary on this appeal to speculate about the different circumstances in which acts might fall within (so as to attribute), or outside (so as not to allow attribution), the scope of something done during the commission, or immediately after the commission, of a crime. It suffices to illustrate this point with an example based on the circumstances of this case. On the assumption that the deceased lit the ring burner, that act would have been an act within the scope of the commission of a crime. If the deceased's act of lighting the ring burner had led to an explosion which killed a child standing outside the house then the appellant could have been charged under s 18(1) with felony murder for the death of the child. It was within the common purpose of both the appellant and the deceased that the crime of manufacture of a large commercial quantity of methylamphetamine (punishable by imprisonment for life) should be committed, and the cause of the child's death was an act done by the deceased during the commission of that crime. There should not be anything surprising in the notion of attributing the acts of one person to others with a common criminal purpose where the person's acts are in the course of, or incidental to, carrying out a common criminal purpose. The same principle applies in civil cases, where, apart from cases of employment or agency, "to constitute joint tortfeasors two or more persons must act in concert in committing the tort". The important point is that it is the acts which are attributed from one person (the actor) to another who shares the common purpose and, by attribution, becomes personally responsible for the acts. It is not the liability of the actor which is attributed. Nor is it the actus reus of some notional crime without a mental element that might be committed by the actor. These points were established in the decision of the majority of this Court in Osland. In Osland, Mrs Osland and her son David were tried for the murder of Mr Osland. The prosecution case was that the blows causing death were struck by David but that Mrs Osland acted in concert with him. The jury convicted Mrs Osland but were unable to reach a verdict in relation to David. One question on the appeal was whether those verdicts were inconsistent. The majority of the High Court (McHugh, Kirby and Callinan JJ) held that they were not. As McHugh J (with whom Kirby J and Callinan J agreed on this point) explained, the liability of persons as accessories before the fact to murder (ie persons not present at the commission of the crime) was a form of derivative liability. So too was the liability derivative for persons who were "merely present, encouraging but not participating physically, or whose acts were not a substantial cause of death". But where two or more persons act in concert then any liability is primary. The acts of one are attributed to the others because they reached an understanding or arrangement that together they would commit a crime and the acts were performed in furtherance of that understanding or arrangement. In Osland, McHugh J concluded that it was likely that the jury had found beyond reasonable doubt that Mrs Osland and David killed the deceased in accordance with a common understanding or arrangement but that one or more jurors were not satisfied that the Crown had negatived David's claim of justification based on self-defence. This meant that David's act was not proved to lack justification. It was not proved to be a criminal act. Nevertheless, the reason why the verdicts in Osland were not inconsistent was that it was David's acts which were attributed to Mrs Osland. David's acts did not need to be criminal acts. They did not even need to be wrongful. A critical passage of the reasoning of McHugh J was as follows: "It is more accurate to describe the person, who escapes liability in a concert case where the other person is convicted, as a non-responsible ... agent. No doubt there are cases where the person who does the harm‑causing act is innocent in a moral sense. For example, the accused may have induced a child of tender years to do the act which constitutes the actus reus of the crime, or imported drugs via an airline carrier. In that case, the agent is innocent of any wrong doing and the accused is regarded as a principal in the first degree. The acts of the innocent person are attributed to the accused who is guilty of the crime because the latter has the necessary mens rea. The fact that the innocent agent is not guilty of the crime is of no relevance." When McHugh J spoke of the child of tender years doing the act which "constitutes the actus reus of the crime" he was referring to the actus reus of the crime which was committed by the accused person, not some notional crime which might have been committed if the young child had been an adult. The same is true of the case which his Honour cited concerning the importation of drugs. Justice McHugh cited the decision in White v Ridley in the context of attribution based on acts in concert, although the rule of attribution in that case was that an agent's acts are attributable to a principal. However, the key point was that the acts were attributed although the airline agent was wholly innocent and committed no crime. As Gibbs J had said earlier, "it is well settled at common law that a person who commits a crime by the use of an innocent agent is himself liable as a principal offender". This conclusion is also consistent with the discussion by McHugh J of a proposition of a majority of the Full Court of the Supreme Court of Victoria in R v Demirian. That case concerned a circumstance where a person accidentally killed himself by detonating a bomb in the course of commission of a crime with an accused accomplice. The accused was convicted of murder. That conviction was quashed by the Full Court. The proposition to which McHugh J referred was that even if the accused was present at the scene of the explosion and was acting in concert with the deceased, the accused could not have been convicted as a principal in the first degree (ie by attribution of acts). Justice McHugh said that on the facts of Demirian the proposition may be correct. The accused would not be guilty of murder by attribution of the act of the deceased person. But McHugh J rejected, as contrary to a long line of cases, the "general proposition" that those acting in concert at the scene of a crime were not principals in the first degree. The conclusion that joint enterprise liability involves attributing only the acts of the participants who share a common purpose is consistent with the controversial English decisions in Bourne, R v Austin, and R v Cogan. However, there is difficulty with some of the reasoning in those cases. The difficulty, not confined to the criminal law, arises due to a failure to separate clearly a liability which is primary, and a liability which is derivative. Liability which is primary can involve attribution of the acts of another. But the liability remains personal to the accused. Liability which is derivative depends upon attribution to the accused of the liability of another. If the other is not liable then the accused cannot be liable. In Bourne, a husband was found guilty of aiding and abetting his wife to commit buggery with a dog. On appeal to the Court of Criminal Appeal it was submitted that the husband could not commit an offence as an aider and abetter unless his wife was also guilty of an offence. It was submitted that the wife would have had a defence of duress and therefore did not commit an offence. The Court of Criminal Appeal dismissed the appeal without calling on the Crown. In brief reasons, the Court explained that even assuming that the wife had a defence of duress, the husband was liable as a principal in the second degree. The difficulty with this reasoning is that, as McHugh J noted in Osland, the liability of a principal in the second degree is derivative, not primary. The husband's guilt as a principal in the second degree required the wife to be guilty. As Glanville Williams observed: "The notion that, where the actus reus of felony is committed without mens rea, there is a felony for collateral purposes, is one without precedent in the long history of the criminal law." This conclusion does not mean that no offence was committed in Bourne. The husband could have been indicted as a principal on the basis, as explained by McHugh J in Osland, that the husband and wife were involved in a joint criminal enterprise. The acts of the wife were attributed to the husband. It did not matter if the wife was not liable for any crime, because it was not the liability that was attributed. Nor did it matter that the husband could not physically have performed the precise act of buggery, because the notion of attribution of an act does not involve a fiction that the act was undertaken by the husband. In company law, for example, it is never said that acts cannot be attributed to a company merely because the company cannot physically perform the acts itself. "To say that a company cannot do something means only that there is no one whose doing of that act would, under the applicable rules of attribution, count as an act of the company". The same explanation applies to R v Austin. In that case, the four appellants were convicted of assisting a father to take his child from the mother. They had planned and executed the taking of the child by the father. The Court of Appeal held that although the father had committed the offence of "child stealing" using force to take away the child, he could have claimed "a right to possession of the child", which would have excused him from commission of the offence under a proviso to s 56 of the Offences against the Person Act 1861. The Court of Appeal upheld the conviction of the appellants for aiding and abetting the commission of the offence even though the father had committed no offence. In Osland, McHugh J treated this case as one in which the appellants were principals in the first degree, that is, persons who were primarily liable for the forcible acts of "child stealing" committed by the father. On that basis, the father's excuse under the proviso was irrelevant. The decision in R v Cogan is also possibly explicable in these terms, although the reasoning in the case is again difficult to justify. In that case, Leak was charged with aiding and abetting Cogan to rape Leak's wife. The jury convicted Cogan of rape, and convicted Leak of aiding and abetting the rape. The jury also returned a special verdict that Cogan had believed that Leak's wife was consenting but had no reasonable grounds for that belief. Cogan's appeal against conviction was allowed and his conviction was quashed, based upon the decision of R v Morgan. Leak appealed on the basis that he could not be guilty of aiding and abetting Cogan since Cogan's conviction had been quashed. The Court of Appeal upheld Leak's conviction, saying that it "would be an affront to justice and to the common sense of ordinary folk" if a person could not be convicted of aiding and abetting an offence "merely because the person alleged to have been aided and abetted was not or could not be convicted". As Gaudron and Gummow JJ observed in Osland, the reasoning of the Court of Appeal in R v Cogan has been described as "demonstrably unsound" and "contrary to principle". It is, at least, difficult to understand how liability can be attributed for an offence which had not been proved. However, in Osland, McHugh J referred to the obiter dictum of the Court of Appeal in R v Cogan where the Court suggested that Leak could have been indicted as a principal offender. The basis for such an indictment would be the attribution of the acts of Cogan, rather than Cogan's liability, based upon a joint criminal enterprise. The Court of Appeal dismissed an objection to this view on the basis of the "presumption", then current in England, that a husband could not rape his wife. The Court of Appeal said that there "is no such presumption when a man procures a drunken friend to do the physical act for him". Whether or not this exception to the abhorrent fiction was correct, the obiter dictum rests upon the correct assumption that acts can be attributed even if the actor is not liable for any offence. In summary, the decision of the majority of this Court in Osland resolved much confusion that had existed in the context of the primary liability of an accused person based upon the attribution of acts done in the course of a joint criminal enterprise. That decision was, and continues to be, authority for the proposition that joint criminal liability involves the attribution of acts. The attribution of acts means that one person will be personally responsible for the acts of another. The decision in Osland does not involve attribution of liability for either the whole of a crime or part of a notional crime. Conclusion The appeal should be allowed and orders made as proposed by Bell and Nettle JJ. BELL AND NETTLE JJ. The appellant was tried in the Supreme Court of New South Wales on one count of manufacturing a large commercial quantity of a prohibited drug, namely 6.7 kilograms of methylamphetamine (Count 1); one count of murder (Count 2a); in the alternative, one count of unlawfully causing the death of Zhi Min Lan ("the deceased") (Count 2b); and four offences relating to the unlawful possession of firearms. The Crown alleged that the appellant committed the offence charged by Count 1 by participating with the deceased in a joint criminal enterprise to manufacture the methylamphetamine. In relation to Counts 2a and 2b, the Crown alleged that, although the evidence could not exclude the possibility that the deceased was killed accidently as a result of his own act, the appellant was guilty of his murder, or alternatively manslaughter, pursuant to s 18(1) of the Crimes Act 1900 (NSW) by reason that the act which caused the deceased's death was committed in the course of the joint criminal enterprise to manufacture the methylamphetamine, an offence punishable by imprisonment for life, and was therefore an act of the appellant as a participant in that enterprise. At the conclusion of the Crown case, the trial judge (Hamill J) directed the jury to acquit the appellant of Counts 2a and 2b. On appeal by the Crown pursuant to s 107(2) of the Crimes (Appeal and Review) Act 2001 (NSW), the Court of Criminal Appeal (Simpson JA, R A Hulme and Bellew JJ agreeing) held that the directed verdicts of acquittal should be quashed and that there should be a new trial on those counts. The question for decision in this appeal is whether the trial judge was correct to direct the jury to acquit the appellant of the counts of murder and manslaughter charged by Counts 2a and 2b. For the reasons which follow, the question should be answered affirmatively and the appeal should be allowed. Relevant statutory provisions To the extent that is relevant, ss 24 and 33 of the Drug Misuse and Trafficking Act 1985 (NSW) provide as follows: "24 Manufacture and production of prohibited drugs ... (2) A person who manufactures or produces, or who knowingly takes part in the manufacture or production of, an amount of a prohibited drug which is not less than the commercial quantity applicable to the prohibited drug is guilty of an offence. ... 33 Penalties for offences involving commercial quantities or cultivation for a commercial purpose (1) This section applies to the following offences: (a) an offence under section 23(1A) or (2), 24(2) or 25(2) or (2A), ... (2) The penalty for an offence is: (a) except as provided by paragraph (b), a fine of 3,500 penalty units or imprisonment for 20 years, or both, ... (3) Despite subsection (2), if the court is satisfied that the offence involved not less than the large commercial quantity of the prohibited plant or prohibited drug concerned, the penalty for the offence is: (a) except as provided by paragraph (b), a fine of 5,000 penalty units or imprisonment for life, or both, or (b) where the offence relates to cannabis plant or cannabis leaf, a fine of 5,000 penalty units or imprisonment for 20 years, or both. (4) In this section: large commercial quantity, in relation to a prohibited plant or prohibited drug, means the number or amount, if any, specified opposite the plant or drug in Column 5 of Schedule 1." Section 18 of the Crimes Act defines murder, including that known as constructive or "felony" murder, thus: "Murder and manslaughter defined (1) (a) Murder shall be taken to have been committed where the act of the accused, or thing by him or her omitted to be done, causing the death charged, was done or omitted with reckless indifference to human life, or with intent to kill or inflict grievous bodily harm upon some person, or done in an attempt to commit, or during or immediately after the commission, by the accused, or some accomplice with him or her, of a crime punishable by imprisonment for life or for 25 years. (b) Every other punishable homicide shall be taken to be manslaughter. (2) (a) No act or omission which was not malicious, or for which the accused had lawful cause or excuse, shall be within this section. (b) No punishment or forfeiture shall be incurred by any person who kills another by misfortune only." The trial The evidence adduced at trial established that, on 4 January 2013, emergency services officers were called by a neighbour to attend a fire that had broken out in residential premises owned by the appellant in the Sydney suburb of Ryde. Upon the officers' arrival, the appellant attempted to block their entry into the premises. After gaining entry, the officers treated the deceased and the appellant for injuries suffered in the fire. The deceased's injuries were severe and he died in hospital on 14 January 2013 without regaining consciousness. Police found that a process of evaporation and purification had been set up at the premises to allow for the refinement of raw methylamphetamine which had been manufactured at another location. It appeared to be a serious commercial venture. The bathroom was equipped with a gas burner, liquid petroleum gas bottle and large cooking pot. The kitchen was equipped to be used in the same way. Various quantities of methylamphetamine, totalling more than six kilograms, were found in differing degrees of purity, indicating that the process of refinement had been ongoing. There were also funnels, sieves, buckets, latex gloves, thermometers, a vacuum flask and pump, and other equipment which converted the bathroom into an "ad hoc meth lab", as well as evidence, in the form of empty tins and bottles, that more than 70 litres of acetone had been kept at the premises. This equipment and evidence suggested that the premises were used for the singular purpose of refining methylamphetamine. On further searches of the premises, police found three pistols, a prohibited weapon and cash in the sum of $328,000, in circumstances indicating that the cash belonged to the deceased. Police also discovered a sum of $16,900 in cash and 15 grams of methylamphetamine in the appellant's locked bedroom in her home in Hurstville. From those facts and circumstances, it was concluded that the appellant was paid for her involvement in the manufacture of methylamphetamine at the Ryde premises. The process of refining the methylamphetamine involved dissolving a solute (containing raw methylamphetamine) in a solvent (acetone) over a low heat. Acetone is flammable and, when heated, the liquid generates flammable vapours. Once the concentration of flammable vapours reaches a certain level, the vapours may explode if exposed to a source of ignition. A spark from an electrical appliance or a naked flame can operate as a source of ignition. An ardent concentration of vapours is likely to be reached more quickly when the evaporation process is undertaken in a confined space. The Crown case on Count 1, relating to the manufacture of the methylamphetamine, was based on evidence of the appellant's ownership of the Ryde premises; her purchase of eight litres of acetone from a hardware store on 1 January 2013; her presence at the premises at the time of the fire; and what was said to be a consciousness of guilt demonstrated by her trying to bar entry by emergency services officers to the premises. The appellant also gave evidence that she had taken two gas bottles, similar to the ones found in the bathroom, to a service station and filled them, and that she had stirred a pot of methylamphetamine in the kitchen early in the morning of 4 January 2013. The Crown contended that, because of the amount of items associated with the manufacture of drugs located at the premises, anyone who entered the premises would have appreciated that they were being used for the manufacture of drugs and, for that reason, the appellant must have understood that the premises were being used for that purpose. This evidence was said to establish, at least, that the appellant was involved in the manufacture of methylamphetamine by permitting it to occur at premises she owned. The Crown case on Counts 2a and 2b, murder and manslaughter respectively, was that the fire, and thus the deceased's death, was caused by the lighting of the gas ring burner in the small and inadequately ventilated bathroom in circumstances which created an objectively appreciable risk of serious injury. The Crown did not allege, however, and could not prove, that it was the appellant who lit the gas ring burner. The Crown argued instead that, because the appellant was a participant with the deceased in a joint criminal enterprise to manufacture a large commercial quantity of methylamphetamine – being an offence punishable by life imprisonment – the appellant was criminally liable, for the purposes of s 18(1) of the Crimes Act, for all acts committed in the course of carrying out that enterprise. It followed in the Crown's contention that, although it could have been the deceased who lit the gas ring burner, the appellant was criminally liable for the consequence of it having been so lit and thus for the deceased's death. At the close of the Crown case, the appellant moved for directed verdicts of not guilty in respect of Counts 2a and 2b. For reasons which the trial judge published on 2 December 2014, his Honour acceded to that application and so directed the jury. Thereafter, the trial continued to conclusion on 9 December 2014 when the jury brought in verdicts of guilty on each of the remaining counts. The trial judge's reasons for the directed verdicts The trial judge considered that there was a flaw in the Crown's argument arising from a misunderstanding of the true effect of the doctrine of common purpose. As his Honour observed, in McAuliffe v The Queen this Court explained that if one or other of the parties to an understanding or arrangement comprising a joint criminal enterprise to commit a crime does, or they do between them, in accordance with the continuing understanding or arrangement, all those things which are necessary to constitute the crime, they are all equally guilty of the crime regardless of the part played by each participant in its commission. The trial judge was persuaded by a submission of the appellant's counsel that any criminal liability that the appellant may have had for the death of the deceased according to the doctrine of common purpose would be "derivative" and, since the deceased had not committed a crime by killing himself, the appellant could not be convicted of murder or manslaughter as a principal in the second degree. As his Honour explained: "Properly analysed, if the [appellant] is liable for murder, she is liable as a principal in the second degree. ... The deceased could not be convicted of his own murder. The offence of suicide was long ago abolished and [in any event] there is no evidence capable of establishing that he acted with the requisite specific intent." Thus, the trial judge concluded: "I do not accept that the combination of principles of common purpose and constructive murder work together to make [the appellant] liable to conviction for murder in the circumstances of the present case. Whether the situation may be different where the deceased person was not the one who committed the act or where the death was of an innocent victim is not necessary to decide in the circumstances of this case." His Honour further concluded that there was insufficient evidence to support a verdict of guilty in relation to the alternative charge of manslaughter. Proceedings in the Court of Criminal Appeal In the Court of Criminal Appeal, Simpson JA, with whom R A Hulme and Bellew JJ agreed, held that the trial judge erred in treating the liability of a passive participant for acts committed by a co-participant in the course of a joint criminal enterprise as derivative, and by focussing on whether the deceased's death was within the scope of the joint criminal enterprise or contemplated by the participants. Her Honour stated that: "The correct question is whether the ignition of the ring burner was within that scope [of the enterprise] or contemplation; if it was, both participants were responsible for it, and liable for its consequences." It followed, her Honour held, that the trial judge was incorrect to direct a verdict of acquittal on the count of murder and, because the lighting of the gas ring burner was an unlawful and dangerous act for which the appellant was criminally liable, the trial judge was also incorrect to direct a verdict of acquittal on the count of manslaughter. Constructive murder Section 18 of the Crimes Act is poorly drafted and, for that reason, difficult to construe. It assists, however, to break down the provision into its constituent parts. For present purposes, the relevant part of s 18(1)(a) is that which defines constructive murder as having been committed by an accused "where the act of the accused ... causing the death charged, was done ... during ... the commission, by the accused, or some accomplice with him or her, of a crime punishable by imprisonment for life or for 25 years". There are two points to note about that definition. First, unlike the common law felony murder rule from which the definition of constructive murder derives, and unlike most statutory formulations enacted in other States, the act which causes the death charged need not be an act of violence. It is sufficient, for the purpose of s 18(1)(a), that the act causing death be done during the commission of a crime punishable by imprisonment for life or 25 years. Secondly, although the natural and ordinary meaning of the terms of s 18 is that the act which causes the death charged must be "the act of the accused" (even though the crime punishable by imprisonment for life or 25 years' imprisonment, during which the act causing death was done, may be committed either by the accused or by some accomplice with the accused), it has long been held that, upon its proper construction, s 18 relevantly imports common law rules of complicity. Thus, an accused may be found guilty of murder even if he or she did not commit the act which caused the death charged provided the act was committed by an accomplice of the accused in the course of carrying out a joint criminal enterprise to which both were parties. In the result, it is the common law doctrine of joint criminal enterprise liability which is determinative of the outcome of this appeal. Joint criminal enterprise liability At its base, the common law doctrine of joint criminal enterprise liability, or "common purpose" or "concert" as it may equally be called, is that: "if two or more persons reach an understanding or arrangement that together they will commit a crime and then, while that understanding or arrangement is still on foot and has not been called off, they are both present at the scene of the crime and one or other of them does, or they do between them, in accordance with their understanding or arrangement, all the things that are necessary to constitute the crime, they are all equally guilty of that crime regardless of what part each played in its commission. In such cases they are said to have been acting in concert in committing the crime." (emphasis added) The doctrine also has a further dimension which extends to any other crime committed by a party to the understanding or arrangement in the course of carrying out the understanding or arrangement if that other crime is within the scope of the understanding or arrangement. Thus, as was explained by Stephen J in Johns v The Queen: "The criminal responsibility here under discussion is not that relating to the crime which is the prime object of a criminal venture. As to that crime, one who, while not actually physically present and participating in its commission, nevertheless knows what is contemplated, and both approves of it and in some way encourages it thereby becomes an accessory before the fact ... His knowledge, coupled with his actions, involves him in complicity in that crime. But if, in carrying out that contemplated crime, another crime is committed there arises the question of the complicity of those not directly engaged in its commission. The concept of common purpose provides the measure of complicity, the scope of that common purpose determining whether the accessory before the fact to the original crime is also to share in complicity in the other crime. If the scope of the purpose common both to the principal offender and to the accessory is found to include the other crime, the accessory will be fixed with criminal responsibility for it." There is then also a third dimension of joint criminal enterprise liability usually called "extended common purpose" or "extended concert" which was considered by this Court in McAuliffe and more recently in Miller v The Queen, which extends to crimes that, although not within the scope of the understanding or arrangement, are foreseen as possibly being committed in the course of carrying out the understanding or arrangement, and are then committed by one of the participants when carrying out the understanding or arrangement. The doctrine of extended common purpose is not in issue in this appeal and, for present purposes, need not be considered further. Nonetheless, it should be observed that the doctrine of common purpose and the doctrine of extended common purpose are at one in attributing criminal liability to one participant for a crime committed by another participant in the course of carrying out their joint criminal enterprise. In this matter, it is apparent that the Court of Criminal Appeal proceeded on the basis that the doctrine of joint criminal enterprise liability renders an accused liable for all acts within the scope of the enterprise committed by a co‑participant in the course of carrying out the enterprise, whether or not those acts amount to a crime. Hence, as Simpson JA reasoned: "The Crown case on murder was that [the deceased's] death was caused by an act of the [appellant], in the course of the commission, or attempted commission, of an offence of manufacturing a large commercial quantity of methylamphetamine. However, as has been noted above, because the Crown was unable to nominate any act or event that caused the ignition of the ring burner, it was unable to nominate any specific act of the [appellant] that caused [the deceased's] death. The Crown therefore relied upon principles of law relating to joint criminal enterprise, particularly those with respect to fixing one participant with criminal liability for the acts (or omissions), within the scope of their agreement, of another participant, or other participants. ... It was, thus, the Crown case that, because the [appellant] and [the deceased] were engaged in a joint criminal enterprise, the [appellant] was equally responsible for the act of ignition whichever of the two actually did it; that, in effect, if it were [the deceased's] act that caused the ignition of the ring burner that act was, on the principles stated in Johns and McAuliffe, the [appellant's] act. Since the [appellant] was thus responsible for the act causing death, and it was an act done in the course of the commission of an offence punishable by imprisonment for life, the [appellant] was guilty of the murder of [the deceased]. ... There was no real dispute that the Crown could make out a case that the [appellant] and [the deceased] were engaged in a joint criminal enterprise to manufacture a large commercial quantity of a prohibited drug. There could have been no real dispute that each bore criminal liability for all of the acts of the other that were within the scope of that joint criminal enterprise, or were contemplated by it. Plainly, the act of lighting the ring burner was such an act. Whichever of the [appellant] and [the deceased] did that act, the other was equally liable for it." (emphasis added) With respect, however, that is not so. Although it is not infrequently, and in a sense not inaccurately, stated in the authorities that a participant in a joint criminal enterprise is criminally liable for acts committed by a co-participant in the course of carrying out the enterprise, a careful examination of those authorities shows that such references are invariably to acts that are identified, expressly or by necessary implication, as comprising the actus reus of a crime. And logically it could not be otherwise, given, as has been seen, that the essence of joint criminal enterprise liability is that two or more participants in a joint criminal enterprise who between them do all the things that are necessary to constitute a crime are equally liable for the acts which constitute the actus reus of that crime. Thus, by definition, joint criminal enterprise liability is limited to participation in acts constituting the actus reus of a crime and has nothing to say about liability for acts which are not the actus reus of a crime or are incapable of constituting the actus reus of a crime. Of course, that does not mean that the liability of one participant for the actus reus of a crime committed by another participant in the course of carrying out their joint criminal enterprise is derivative of the other participant's liability for committing the act constituting the offence. As was established in Osland v The Queen, the liability of each participant in a joint criminal enterprise for acts committed in the course of the enterprise is direct, primary liability. Rather, the foregoing observations emphasise that the purpose of the doctrine of joint criminal enterprise liability in this respect is, and is only, to attribute liability for crimes incidental to the enterprise. For that reason, it is not open under the doctrine of joint criminal enterprise liability to attribute criminal liability to one participant in a joint criminal enterprise for an act committed by another participant in the course of carrying out the enterprise unless the act is or is part of the actus reus of a crime. By way of illustration of the point, if two persons enter into an arrangement to commit an armed robbery of a bank with a dangerous weapon, and pursuant to that arrangement one participant in the enterprise is to wait in a car keeping a lookout while the co-participant goes directly across a busy street to the bank to effect the robbery, and, in crossing the street, the latter participant walks in front of a bus and is killed, although the act of walking across the street was done in furtherance of the joint criminal enterprise, the participant who waited in the car is not guilty of the murder or manslaughter of the co‑participant: the act of the co-participant in walking across the street was not the, or part of the, actus reus of the crime of murder or manslaughter. Equally, if two persons enter into an arrangement to break into a dwelling with a dangerous weapon with an intent to commit a serious indictable offence, and one participant in the enterprise falls from the roof while attempting to gain entry to the dwelling via a second-storey window and is killed, although the act of attempting to access the dwelling was done in furtherance of the joint criminal enterprise, the other participant is not guilty of the murder or manslaughter of the co-participant: the act of climbing on the roof did not amount to the actus reus of the crime of murder or manslaughter. Again, if two persons enter into an arrangement to cultivate a large commercial quantity of opium plants and one participant in the enterprise is killed as a result of some mechanical failure of the tractor used to cultivate the opium crop, the other participant is not guilty of the murder or manslaughter of the co-participant: the use of the tractor was not the actus reus of the crime of murder or manslaughter. Despite that being so, before this Court, the Crown invoked passages from the reasons for judgment of Street CJ in Johns in the New South Wales Court of Criminal Appeal, and from the judgments of Barwick CJ and Mason, Murphy and Wilson JJ in Johns in this Court, as establishing that the doctrine of joint criminal enterprise liability does impose "liability for acts and not liability for crimes". It was also contended that the judgment of McHugh J in Osland makes clear that the doctrine of joint criminal enterprise liability renders participants in a joint criminal enterprise liable for all acts within the scope of the enterprise that are committed in the course of carrying out the enterprise regardless of whether the acts constitute the actus reus or part of the actus reus of a crime. So to construe those judgments misunderstands the essential effect of them. In the passage from the judgment of Street CJ in Johns to which the Crown referred, his Honour stated: "A principal in the second degree may be held liable pursuant to the doctrine of common purpose, if the particular actus reus, whilst differing from that directly and specifically intended by the principals, was nevertheless one that the jury regard as within the contemplation of the parties as an act which might be done in the course of carrying out their primary criminal intention. This can be described, alternatively, as an act contemplated by the principals as a possible incident of the particular venture upon which they embarked". (emphasis added) It is true that Barwick CJ in this Court spoke in terms of acts, as opposed to acts comprising the actus reus of a crime. In the passage relied upon by the Crown in this appeal, his Honour stated: "The participants in a common design are liable for all acts done by any of them in the execution of the design which can be held fairly to fall within the ambit of the common design." But it needs to be understood that this passage follows immediately after Barwick CJ's approval of the judgment of this Court in Brennan v The King, which speaks unmistakably in terms of participants in a joint criminal enterprise being liable for an offence or unlawful act committed in the course of carrying out the venture. It is, therefore, necessarily implicit in Barwick CJ's subsequent reference to "all acts" that such acts are limited to those which comprise an offence. The relevant passage from the judgment of Mason, Murphy and Wilson JJ in Johns also makes clear that the doctrine of joint criminal enterprise liability is concerned with acts by which a crime is committed: "The object of the doctrine is to fix with complicity for the crime committed by the perpetrator those persons who encouraged, aided or assisted him, whether they be accessories or principals. Broadly speaking, the doctrine looks to the scope of the common purpose or design as the gravamen of complicity and criminal liability." In Osland, it was not in issue that Mrs Osland and her son David had agreed to kill Mr Frank Osland, nor that Frank Osland had been killed pursuant to that agreement. In each case the issue was whether self-defence and provocation had been negatived. The jury convicted Mrs Osland but were unable to agree on whether David should be acquitted on the ground of self-defence. Mrs Osland appealed against conviction on the ground inter alia of inconsistency of verdicts. It was held that there was no inconsistency: David's acts in delivering the fatal blows were consistent with him having acted in self-defence and in accordance with the agreement with his mother, whereas Mrs Osland was not acting in self-defence. As McHugh J explained, the liability of each party to a joint criminal enterprise for crimes committed in the course of that enterprise is direct or primary liability, not derivative or secondary liability: "they are all equally liable for the acts that constitute the actus reus of the crime" (emphasis added). Hence, although a participant whose act causes death may have a defence to the homicide, the other participant in the joint criminal enterprise cannot escape liability. David's acts which constituted the actus reus of the unlawful homicide charged against Mrs Osland were acts for which she was responsible and, unlike her son, she could not call in aid a legal justification for their commission. Admittedly, at one point in McHugh J's reasoning, his Honour stated that "it is the acts, and not the crime, of the actual perpetrator which are attributed to the person acting in concert". But, immediately after that statement, his Honour reiterated: "If the latter person has the relevant mens rea, he or she is guilty of the principal offence because the actus reus is attributed to him or her by reason of the agreement and presence at the scene. It is irrelevant that the actual perpetrator cannot be convicted of that crime because he or she has a defence such as lack of mens rea, self-defence, provocation, duress or insanity." (emphasis added) The point was further emphasised later in his Honour's reasons in observations concerning the decision of the Full Court of the Supreme Court of Victoria in R v Demirian. In that case, the accused had entered into an arrangement with an accomplice to place and detonate a bomb so as to destroy a building, but, in the course of carrying out the arrangement, the accomplice had accidentally detonated the bomb, thereby killing himself. On those facts, a majority of the Full Court held that the accused could not be convicted as a principal in the first degree in respect of the accomplice's murder, because, although acting in concert with the accomplice and, subject to proof by the prosecution, present at the scene when the accomplice detonated the bomb, the accused himself did not detonate the bomb. Their Honours took the law to be that parties acting in concert to commit a crime who were present when the crime was committed were not liable as principals unless they actually committed the act or one of the acts that constituted the actus reus of the crime; they were liable otherwise only as accessories. As was later demonstrated in Osland, however, that was not so. As McHugh J stated: "Counsel for the appellant relied on R v Demirian where a majority of the Full Court of the Supreme Court of Victoria said that, even if it could have been established that the accused in that case was present at the scene of a bomb explosion and was acting in concert with the person who exploded the bomb, the accused could not have been convicted as a principal in the first degree. Upon the facts of that case, the statement may be correct. However, it is plain that their Honours were seeking to lay down a general proposition. In my opinion it is not an accurate statement of the modern law." (emphasis added; footnote omitted) His Honour continued: "Where the parties are acting as the result of an arrangement or understanding, there is nothing contrary to the objects of the criminal law in making the parties liable for each other's acts and the case for doing so is even stronger when they are at the scene together. If any of those acting in concert but not being the actual perpetrator has the relevant mens rea, it does not seem wrong in principle or as a matter of policy to hold that person liable as a principal in the first degree. Once the parties have agreed to do the acts which constitute the actus reus of the offence and are present acting in concert when the acts are committed, the criminal liability of each should depend upon the existence or non-existence of mens rea or upon their having a lawful justification for the acts, not upon the criminal liability of the actual perpetrator. So even if the actual perpetrator of the acts is acquitted, there is no reason in principle why others acting in concert cannot be convicted of the principal offence. They are responsible for the acts (because they have agreed to them being done) and they have the mens rea which is necessary to complete the commission of the crime." (emphasis added) As each of those passages of McHugh J's judgment affirms, in order to engage the doctrine of joint criminal enterprise liability, the participants in the joint criminal enterprise must do between them all things necessary to constitute the crime; and so, in order to render a participant in a joint criminal enterprise criminally liable for an act of homicide committed by a co-participant in the course of the enterprise, the act causing death must constitute the actus reus of a crime involving the unlawful killing of the deceased. Evidently, that is why McHugh J observed that, apart from principle, upon the facts of Demirian it was possible that the accused could not have been convicted as a principal in the murder of the accomplice. At the time of Demirian, the crime of suicide had been abolished in Victoria and so the accomplice's act of killing himself was not the actus reus of a crime. The actus reus of the crime for which the appellant was liable In this case, it was accepted that the Crown could not negative the reasonable possibility that it was the deceased who lit the gas ring burner which sparked the fire that caused his death. It followed that the Crown could not exclude as a reasonable possibility that the deceased had killed himself by his own act. And, as the trial judge observed in the course of his ruling on the directed verdicts, it is no longer an offence to kill oneself in the State of New South Wales. As was earlier noted, the trial judge reasoned that the liability of a participant in a joint criminal enterprise for an offence committed by another participant in the course of the enterprise is secondary or derivative liability, and therefore, because the deceased committed no crime in killing himself, the appellant could not be held liable for the killing. And as Simpson JA observed in the Court of Criminal Appeal, that process of reasoning was erroneous because the liability of a participant in a joint criminal enterprise for a crime committed by another participant in the course of that enterprise is direct or primary liability. But, although the trial judge's process of reasoning was incorrect in that respect, his Honour's conclusion was correct. To explain why that is so it is necessary to digress. At common law, murder included self-murder, or suicide as it is now more often called, and self-murder included not only the intentional taking of one's own life but also the doing of a felonious act which unintentionally caused one's own death. Thus, in R v Russell, it was held by a majority of the Court for Crown Cases Reserved that the act of a woman in taking arsenic to procure her own miscarriage and thereby killing herself, though that self-killing was not her intention, was felo de se or self-murder. The decision in Russell was later doubted. The argument in Russell had relied upon the ancient common law conception of implied malice aforethought that, if an act were malum in se, it was no excuse that the actor did not intend all its consequences. That was at odds with subsequent 19th century developments that confined the common law felony murder rule so that an act causing death committed in the course of a felony was not considered murder unless the act involved violence or danger to some person and early 20th century cases, also involving illegal abortion procedures, in which it was held that it was not murder to cause a woman's death in the course of procuring her miscarriage if the act were done with her consent and without it being considered that the act would, or would likely, cause her death. But the authority of Russell was followed and there is no reason to doubt that it formed part of the common law in New South Wales. To a large extent, the common law of murder was replaced in New South Wales by s 9 of the Criminal Law Amendment Act 1883 (NSW), the legislative forebear of s 18(1) of the Crimes Act. Like s 18(1), s 9 defined murder by reference to various categories of acts causing death which at common law gave rise to murder. In its terms it did not include acts of self-murder but nor did it abrogate common law rules to which it was not specifically directed. Hence, it is arguable that the common law rule that it was murder to kill oneself intentionally, or unintentionally by an act committed in the course or furtherance of a felony, continued to apply in New South Wales, with provision for attempted suicide to be dealt with summarily, until the rule was abrogated by the enactment of s 31A of the Crimes Act in 1983. Certainly from that point, however, suicide or self-murder ceased to be a crime in New South Wales; and self-manslaughter was never a crime, even at common law. Accordingly, assuming it were the deceased's act of lighting the gas ring burner which caused the deceased's death, that act was not the actus reus of a crime of murder or manslaughter; or, to put it another way, the deceased and the appellant did not do between them all the things necessary to constitute a crime of murder or manslaughter. It follows that the appellant could not properly be considered liable for the deceased's death pursuant to the doctrine of joint criminal enterprise liability. It would have been a very different case, however, if a third party had been killed. The Crown contended in the course of argument before this Court that, according to the doctrine of joint criminal enterprise liability, each participant in a joint criminal enterprise is considered to be the agent of each other participant for the purpose of committing acts within the scope of the understanding or arrangement, and consequently that, although it is to be assumed that the deceased may have lit the gas ring burner which caused his death, by operation of the doctrine the appellant is to be taken to have done that act herself through the agency of the deceased. It followed, it was contended, that the act of lighting the gas ring burner which caused the deceased's death was a crime because it was an act done, or which must be taken to have been done, by the appellant in the course of manufacturing the methylamphetamine, and so was an act causing death to which s 18(1) of the Crimes Act applied. That contention should be rejected. It confuses liability for an act committed by an agent with the doing of the act itself. Agency attributes legal responsibility for an act done by an agent, within the scope of his or her authority, to the agent's principal. It does not in a physical sense transmogrify the act done by the agent into an act done by the principal. For that reason, where in the criminal law it is recognised that an accused may commit an offence by a "non-responsible" (or partially responsible) agent, the act of the agent remains in fact the act of the agent. It is the act of the agent, as was committed by the agent, and not such an act as if it had been committed by the principal, that is to be assessed to determine whether the act comprised the actus reus of an offence. As has been observed by Professor Gillies: "The courts have never pretended that the defendant incriminated pursuant to [the doctrine of innocent agency] has personally perpetrated the criminal act ... Rather, it is the offender's classification as a constructive principal which has represented the enduring element of fiction in the doctrine." The point may be illustrated by reference to three of the authorities mentioned by McHugh J in Osland in his Honour's consideration of the law relating to criminal offending by means of an innocent agent. In the first of those cases, Bourne, a husband was found guilty of bestiality as a principal in the second degree on the basis that he compelled his wife to have sexual intercourse per vaginam with a dog. The wife was not charged with an offence, and it was assumed that, if she had been charged, she would have been entitled to an acquittal on the ground of duress. But it was held that the husband was rightly convicted as a principal because: "if this woman had been charged herself with committing the offence, she could have set up the plea of duress, not as showing that no offence had been committed, but as showing that she had no mens rea because her will was overborne by threats of imprisonment [by the husband] or violence so that she would be excused from punishment. But the offence ... does not depend upon consent; it depends on the act, and if an act of buggery [sic] is committed, the felony is committed." The significance of that for present purposes is that the actus reus of the crime was the wife's act in having sexual intercourse with the dog. There was no suggestion that, because the husband had coerced the wife into performing that act, the husband was somehow to be taken as himself having had sexual intercourse with the animal. Nor, obviously, could he have done so per vaginam, that being the manner in which the offence was said to have been committed. The second case, R v Cogan, concerned a husband, Leak, who had forced his wife to have sexual intercourse with another man, Cogan. The husband was convicted of aiding and abetting the rape of his wife but Cogan was acquitted of the rape by reason that the jury were not satisfied that he knew that the wife was not consenting. Notwithstanding that the husband's appeal was against his conviction for aiding and abetting, it was observed that he could equally have been convicted as a principal and if he had been so indicted: "[i]t would have been no defence for him to submit that if Cogan was an 'innocent' agent, he was necessarily in the old terminology of the law a principal in the first degree, which was a legal impossibility as a man cannot rape his own wife during cohabitation. The law no longer concerns itself with niceties of degrees in participation in crime; but even if it did [the husband] would still be guilty. The reason a man cannot by his own physical act rape his wife during cohabitation is because the law presumes consent from the marriage ceremony ... There is no such presumption when a man procures a drunken friend to do the physical act for him." The significance of that observation for present purposes is that it was not considered, for the purpose of the husband's criminal liability as a principal, that Cogan's act of rape was in any sense converted into an act committed by the husband. Rather, to the contrary, it was Cogan's act an act of non-consensual sexual penetration of a woman by a man other than the woman's husband that comprised the actus reus of the crime, and the husband was guilty by reason of procuring that act to be committed. Had that not been so more precisely, had Cogan's act been treated as if the act of penetration had been committed by the husband himself the husband would have had a complete defence to the charge as the law then stood in England. The third case was R v Austin, concerning a father who forcibly took possession of his child from his wife. Four other men who assisted the father in locating the wife and child and in taking possession of the child were charged with aiding and abetting an offence of child stealing contrary to s 56 of the Offences against the Person Act 1861 (UK). The four men contended that, although they had deliberately aided and abetted the father, they had not committed an offence because the father had not acted unlawfully within the meaning of s 56. That section provided in substance that a person who unlawfully by force takes any child under 14 years of age with the intent of depriving any parent of the child's possession is guilty of an offence "[p]rovided, that no person who shall have claimed any right to the possession of such child ... shall be liable to be prosecuted by virtue hereof on account of the getting possession of such child". The Court of Appeal held that the effect of s 56 was that the taking of the child was "unlawful" but that the proviso operated to protect the father as a person claiming a right of possession of the child. That did not protect the father's aiders and abettors against the unlawfulness of the father's act. In the result, it was the father's unlawful, albeit protected, act which comprised the actus reus of the offence in relation to which the aiders and abettors were convicted and it was that act, not some act which they were deemed or somehow supposed to have committed themselves, for which they were held criminally liable. That stands in contrast to the situation in this case, where an act of the deceased causing his own death was not unlawful and was thus incapable of constituting the actus reus of the offence of murder with which the appellant was charged. Malice and foresight of incidental crime What has been said to this point is sufficient to resolve the appeal. Counsel for the appellant accepted that was so, but urged the Court to confine the doctrine of joint criminal enterprise liability on a broader basis relating to the foresight of any incidental crime and also to hold that, perforce of s 18(2)(a) of the Crimes Act, an act cannot fall within the scope of s 18(1) unless the act is malicious. It is unnecessary to take up those arguments. So far as foresight of incidental crimes is concerned, it is sufficient to note that what was said by Carruthers J in R v Sharah about the need to prove foresight of the possibility of an outcome has been subsequently questioned by the New South Wales Court of Criminal Appeal in Batcheldor v The Queen and was questioned by the Court of Criminal Appeal below. For this Court, the resolution of that difference can await another day. As to malice, it should be recorded that nothing advanced in argument in this appeal has dissuaded us from the view recently expressed in Aubrey v The Queen that the effect of s 18(1) is to replace the common law concept of malice aforethought with a list of matters that would previously have established malice aforethought; and, consequently, that in a case in which the Crown is able to prove an act of the kind described in s 18(1), s 18(2)(a) (which excludes from the definition in s 18(1) any act or omission which was not malicious) has no role to play. Conclusion For these reasons, the appeal should be allowed. The orders of the Court of Criminal Appeal should be set aside. In lieu thereof, it should be ordered that the appeal to the Court of Criminal Appeal be dismissed. GAGELER J. Intention as a mental element of serious crime is now generally regarded as "the most fundamental element in a rational and humane criminal code". It has not always been so. In particular, it has not always been so in relation to the crime of murder. "At common law, killing constituted murder in two classes of case which, stated broadly, were (1) where the killing was intentional, and (2) where it was done unintentionally in the course of committing certain crimes which did not necessarily involve killing". The same distinction was maintained in the two limbs of the statutory definition of murder enacted in s 9 of the Criminal Law Amendment Act 1883 (NSW) and re‑enacted in s 18(1)(a) of the Crimes Act 1900 (NSW). Murder under the first limb of s 18(1)(a) is taken to have been committed where "the act of the accused, or thing by him or her omitted to be done, causing the death charged, was done or omitted with reckless indifference to human life, or with intent to kill or inflict grievous bodily harm upon some person". That is not this case. Murder under the second limb of s 18(1)(a) is constructive murder. Murder under that limb is taken to have been committed where "the act of the accused, or thing by him or her omitted to be done, causing the death charged, was … done in an attempt to commit, or during or immediately after the commission, by the accused, or some accomplice with him or her, of a crime punishable by imprisonment for life or for 25 years". Except for its identification of the foundational crime (as enacted in 1900 as an act obviously dangerous to life or as a crime punishable by death or penal servitude for life but as amended since 1989 as a crime punishable by imprisonment for life or for 25 years), the second limb of s 18(1)(a) replicates the common law crime of felony murder as understood in 1883. The common law moved on, ultimately to hold that "an unintended killing in the course of or in connexion with a felony is murder if, but only if, the felonious conduct involved violence or danger to some person". The statute did not. This Court has recently confirmed that, by stating that "[n]o act or omission which was not malicious ... shall be within this section", s 18(2)(a) does not cut down the scope of the definition in s 18(1)(a) and does not add to that definition by making malice a freestanding element of murder. Although a reservation had earlier been tentatively expressed to the effect that malice might have some independent role in relation to constructive murder, there is no warrant in the text or history of s 18 for attempting to give s 18(2)(a) such a differential operation. In the case of murder under the first limb of s 18(1)(a), the malice inheres in the requisite intent or reckless indifference. In the case of murder under the second limb of s 18(1)(a), as was indicated by its drafters and as has been recognised judicially, malice no less than the crime itself is constructive: the malice is taken to inhere in the requisite intention to commit the foundational offence. Drawing attention to the fact that many crimes punishable by imprisonment for life or for 25 years are not crimes of violence, the New South Wales Law Reform Commission observed in 2010 that the law of constructive murder "risks over-criminalising those who are involved in a foundational offence that, inadvertently, escalates to a homicide, at least so far as accomplices to the foundational offence are concerned". This case is an extreme illustration of that observation. Constructive murder, like other constructive crime, "should be confined to what is truly unavoidable". Had I been able to see any path of reasoning to the conclusion that Lan's assumed lighting of the ring burner resulting in his own death did not make IL the constructive murderer of Lan, I would have felt justified in taking that path to allow the appeal. Regrettably, I cannot. The Court of Criminal Appeal's holding that constructive murder was an available verdict in this case was, I am convinced, the inexorable result of the statutory assimilation and perpetuation of an outmoded common law doctrine. Together with Gordon J, I would therefore dismiss IL's appeal to this Court. Mindful of the irony of finding myself compelled to reach a deeply disquieting conclusion rejected by the majority in this Court, I confine my reasons to explaining why I am unable to take either of the two alternative paths of reasoning to allowing the appeal indicated by the majority. Lan's lighting of the ring burner was an "act" of IL The starting point for each limb of the definition of murder in s 18(1)(a) is an "act of the accused, or thing by him or her omitted to be done, causing the death charged". The singular reference to an "act of the accused" has consistently been held to encompass an act of an accomplice attributed to the accused through the application of the common law doctrine of joint criminal enterprise. According to that doctrine, "the participant in a joint enterprise or common design is liable for all that occurs in the course of its execution which is of a kind which fairly falls within the ambit of the enterprise or design". More specifically, according to that doctrine, a participant bears criminal responsibility for an act subjectively contemplated by that participant "as an act which might be done in the course of carrying out the primary criminal intention – an act contemplated as a possible incident of the originally planned particular venture". The common law doctrine of joint criminal enterprise applies to render an act of an accomplice an act of the accused for the purpose of each of the two limbs of s 18(1)(a). There is a difference, however, in how the doctrine works out in its application as between the two limbs. The difference derives from the difference in the requisite characteristics of the act causing the death charged. For the purpose of the first limb, the act of the accomplice causing death will be an act of the accused only if the joint enterprise or common design in which the accused and the accomplice participated was contemplated by the accused to extend to the accomplice doing an act of that kind with reckless indifference to human life or with intent to kill or inflict grievous bodily harm upon some person. For the purpose of the second limb, the act of the accomplice causing death will sufficiently be an act of the accused if the joint enterprise or common design in which the accused and the accomplice participated was contemplated by the accused to extend to the accomplice doing an act of that kind as an incident of committing the foundational crime. That difference in how the doctrine of joint criminal enterprise works out in its application as between the two limbs of s 18(1)(a) is reflected in the reasoning in R v Sharah, which has been consistently followed, and which was not challenged in this case. I do not think Sharah to be inconsistent with the earlier explanation in R v Surridge. As the circumstances in Sharah and Surridge both illustrate, there is no reason in principle why the two limbs cannot both be engaged in the same circumstances. The entirety of the explanation in Surridge was given in the context of explaining how both limbs were capable of applying where the accused and the accomplices had the common purpose of robbing a man using "any violence, without any limitation as to its quantum, which might be necessary to overcome his resistance". Surridge was accordingly a case in which the mental element identified in Sharah as necessary to render an act of an accomplice an act of the accused for the purpose of each limb of s 18(1)(a) was satisfied. The nature and extent of the criminal responsibility attributed by operation of the common law doctrine of joint criminal enterprise has long been obscure, but was squarely addressed in Osland v The Queen. On my understanding of the reasoning of the majority in that case, the effect of the operation of the doctrine is to attribute to the accused primary (as distinct from derivative) criminal responsibility for the physical act of the accomplice, and to do so whether or not the act of the accomplice was one which the accused was physically capable of performing and whether or not the act of the accomplice amounted to an element of a crime committed by the accomplice. The irrelevance of whether the act of the accomplice was one which the accused was physically capable of performing was, I think, implicit in the majority's acceptance of the outcomes in the English cases of R v Cogan and R v Austin and its treatment of those outcomes as explicable "only on the basis that the person acting in concert with the actual perpetrator has attributed to him or her the acts of the actual perpetrator". The irrelevance of whether or not the act of the accomplice amounted to an element of a crime committed by the accomplice was, on my reading, the basis of the actual holding in Osland that there was no inconsistency in the jury in that case having convicted the accused of the murder of her husband while having failed to agree whether her son acted in self-defence or under provocation in performing the act which caused death. The irrelevance of whether the act of the accomplice amounted to an element of a crime committed by the accomplice was emphasised by the majority's express rejection of a statement in the Full Court of the Supreme Court of Victoria in R v Demirian which had been made in the course of examining the potential for an accused who had conspired with another man to blow up a building to be liable at common law for the murder of that man when the bomb exploded accidentally, killing him. What had been said in Demirian was that the accused could not have been convicted of murder as a principal even if the accused was present at the scene of the bomb explosion and was acting in concert with the man who exploded the bomb and who was killed by it. That statement was said in Osland not to be an accurate statement of modern law. Osland was unchallenged in this case. My understanding of Osland accords with the explanation given by Kiefel CJ, Keane and Edelman JJ. That understanding prevents me from acceding to the view of Bell and Nettle JJ that Lan's lighting of the ring burner was not an "act" of IL if IL contemplated Lan lighting the ring burner as an incident of executing their joint enterprise of manufacturing methylamphetamine. For completeness, I note that Osland concerned only the doctrine of joint criminal enterprise and not the doctrine of extended joint criminal enterprise recently affirmed by majority in Miller v The Queen. This case too concerns only joint criminal enterprise. Whether criminal responsibility attributed by operation of the doctrine of extended joint criminal enterprise is primary or derivative and how, if at all, the doctrine of extended joint criminal enterprise might intersect with constructive murder are questions which do not now arise for consideration. Lan's killing of himself is no answer to constructive murder Were the text of s 18 to be read without reference to historical context, the first limb of s 18(1)(a) might be read as drawing a distinction between (1) the accused, being the person whose act or omission has caused death, and (2) the person whose death has been caused. Implicit in that two-fold distinction might be thought to be that the accused and the deceased must be different persons, with the result that a person who intentionally kills himself or herself could never commit murder as defined by the first limb. That reading might be thought to be confirmed by the statement in s 18(2)(b) that "[n]o punishment or forfeiture shall be incurred by any person who kills another by misfortune only" and by the prescription of the punishment for murder, originally in s 19 but now in s 19A, being a punishment which can only be imposed on a person who is alive. The second limb might similarly be read as distinguishing between (1) the accused, (2) the person whose death has been caused, and (3) an accomplice with the accused in the foundational crime. Implicit in that three-fold distinction, it might be thought, is a requirement that the deceased be neither the accused nor any accomplice with the accused in the commission of the foundational crime. One result of that reading would be that a person who unintentionally killed himself or herself in the course of attempting to commit a foundational crime could not become the constructive murderer of himself or herself. Another result would be that a person who unintentionally killed an accomplice in the course of attempting to commit the foundational crime could not become the constructive murderer of the accomplice. The difficulty with that reading of s 18(1)(a) is that it runs counter to the accepted view that the statutory definition replicated the definition of murder as understood at common law in 1883, other than to the extent of removing the common law requirement for malice aforethought and altering the identification of the foundational crime for constructive murder. The contrast in language between ss 18(2)(a) and 18(2)(b), together with the relationship between ss 18(1)(a) and 18(2)(b), is to be understood against the background that whether a person had committed murder and whether a person was able to be tried and punished for murder were treated at common law as distinct questions. Murder at common law encompassed self-murder, and self-murder at common law could be constituted either by intentional self-killing or by unintentional self-killing in the course of the commission or attempted commission of another felony. That murder (at that time defined as "the killing [of] a man with malice prepense") included the intentional killing of a man by himself ("felonia de se") was authoritatively determined at common law in 1562. The casuistic reasoning then used to support the conclusion that intentional self-killing was self-murder soon entered into popular culture. That murder included as well the unintentional killing of a woman by herself in the course of attempting to commit another felony was accepted by text writers and was judicially confirmed in 1832. To an assertion made in argument in 1862 that "[t]he crime of felo de se is treated of apart from murder in all the text books", the judicial response was immediate and blunt: "There are no degrees in self-destruction. If a man feloniously kill himself, it must be self-murder." The man who feloniously killed himself obviously could not be prosecuted for his own murder. But others could. A person present at a self-killing who aided and abetted could be a principal in the second degree to murder. That was true even of the survivor of a suicide pact. Equally, a person not present at a self-killing who aided and abetted could be an accessory before the fact to murder. An accessory before the fact to self-murder was in practice shielded from prosecution at common law through the operation of the general common law rule that an accessory before the fact could not be tried unless the principal felon had first been convicted. The statutory abrogation of that general common law rule was held in 1857 to have the consequence that a person could thereafter be tried for and convicted of being an accessory before the fact to self-murder by another person, including a self-murder constituted by an unintentional self-killing in the course of committing another felony. Moreover, it was by applying the general common law rule that an attempt to commit a felony was a misdemeanour that attempted suicide came to be held in 1854 to be a misdemeanour at common law. Attempted suicide came in that way to be recognised as a common law misdemeanour not as a result of "deliberate penal policy" but rather as the result of "mechanical legal logic": attempted suicide was a misdemeanour for no reason other than that it was an attempt to commit the common law felony of self-murder. The common law misdemeanour of attempted suicide was held in 1862 to be unaffected by the creation in England in 1861 of the statutory felony of "attempt to commit murder". That holding was not on the basis that suicide was not self-murder. An argument to that effect, as already noted, was specifically rejected. The holding was rather on the basis that on a purposive construction the general statutory reference to an attempt to commit murder did not encompass the specific case of an attempt to commit self-murder; the statute did not turn what had previously been a common law misdemeanour punishable by imprisonment into a statutory offence punishable by death. Writing in 1883, the year of enactment of the Criminal Law Amendment Act, Sir James Fitzjames Stephen was accordingly able to state without equivocation or qualification that "[s]uicide is by the law of England regarded as a murder committed by a man on himself" and that "[s]uicide is held to be murder so fully, that every one who aids or abets suicide is guilty of murder". Stephen added the illustration of a suicide pact. "If, for instance, two lovers try to drown themselves together, and one is drowned and the other escapes, the survivor is guilty of murder". The illustration alluded to a case determined by the opinion of nine of the common law judges of England in 1823, which had been relied on in the Supreme Court of New South Wales in 1836. The statutory abolition of the coronial verdict of "felo-de-se" in New South Wales in 1876 was expressed not to affect the law with respect to attempts to commit suicide, and did nothing to remove self-murder from the felony of murder at common law. Within the Crimes Act, as enacted in 1900, the application of the definition of murder in s 18(1)(a) to self-killing needed to be understood in light of s 345, which provided that "[e]very principal in the second degree in any felony ... shall be liable to the same punishment as the principal in the first degree", and also in light of s 346, which went on to provide that "[e]very accessory before the fact to any such felony may be indicted, convicted, and sentenced, either before or after the trial of the principal felon, or together with such felon, or indicted, convicted, and sentenced, as a principal in the felony, and shall be liable in either case to the same punishment as the principal felon, whether the principal felon has been tried or not, or is amenable to justice or not". The application of the definition in s 18(1)(a) to self-killing also needed to be understood in light of the inclusion by s 477(a) of "attempting to commit suicide" within the list of offences punishable summarily under s 476. The offence of attempting to commit murder created by s 30, read consistently with the prior judicial interpretation of its English progenitor of 1861, did not extend to attempting to commit self-murder. There then being no statutory offence of attempting to commit suicide, s 477(a) could only have been drafted on the assumption of the continuing existence of that offence as a common law misdemeanour. The logic which led to attempting to commit suicide continuing to be treated as a common law misdemeanour required that actually committing suicide continue to be treated either as a common law or statutory felony. Borrowing language from the Suicide Act 1961 (UK), the Crimes Act was amended in 1983 to insert s 31A, which provides that "[t]he rule of law that it is a crime for a person to commit, or to attempt to commit, suicide is abrogated". Sections 31B and 31C were also then inserted, the latter to create the new and distinct statutory offence of aiding or abetting the suicide or attempted suicide of another person, and the former to provide that the survivor of a suicide pact is not to be guilty of murder or manslaughter but may be guilty of that new and distinct offence. The rule of law that it was a crime in New South Wales for a person to commit suicide, to which s 31A referred and which it abrogated, cannot be treated as having been a remnant of the common law which somehow survived outside the statutory definition of murder which had existed since 1883. The rule was rather the consequence of s 18(1)(a) being read consistently with the definition of murder at common law so as to permit of the possibility that the accused, being the person whose act or omission caused death, and the person whose death has been caused, might be one and the same person. Although I am aware of no case in which the point was considered, I can see no reason why suicide before 1983 did not constitute murder within s 18(1)(a). Nor can I see any reason why the survivor of a suicide pact would not have been liable to murder, if not as a principal in the first degree by reason of having been a participant in a joint criminal enterprise, then as a principal in the second degree under s 345 read with s 18(1)(a). Similarly, I can see no reason why a person who knowingly assisted without being present at a suicide would not have been liable as an accessory before the fact to murder under s 346 read with s 18(1)(a). Section 31A operated from 1983 to carve suicide out of the definition of murder in s 18(1)(a). So operating, s 31A had the effect of removing suicide as the basis for the common law crime of attempted suicide (which was for good measure also expressly abrogated by s 31A) and as the basis for accessorial liability under s 346. Suicide within the meaning of s 31A undoubtedly encompasses intentional self-killing previously amounting to intentional self-murder within the first limb of s 18(1)(a). Less clear is whether suicide within the meaning of s 31A encompasses unintentional self-killing previously amounting to constructive self-murder within the second limb of s 18(1)(a). Telling in favour of the view that s 31A encompasses constructive self-murder is that the term suicide appears to have been used historically to refer to any felonious self-killing, an example of which given as late as 1929 being that of death of a duellist occurring as a result of the duellist's own pistol exploding in the course of the duel. Telling against the view that s 31A encompasses constructive self-murder is that, by the time of the incorporation of the term into the language of the Suicide Act, felony murder and with it felony self-murder had been abolished in the United Kingdom. There may accordingly be room for doubt as to whether an act by which a person unintentionally causes his or her own death in the course of committing a foundational crime now constitutes constructive self-murder. This is not the case in which to resolve that doubt. That is because the criminal responsibility of IL for the constructive murder of Lan does not depend on whether Lan was a constructive murderer of himself. The act of Lan in lighting the ring burner was for the purpose of s 18(1)(a) the act of IL. The prosecution case against IL is not one of constructive self-murder, but one of constructive murder of an accomplice with the accused in the foundational crime. Clear enough is that suicide within the meaning of s 31A does not encompass the killing of one person by another person. Before 1983, there was no reason why an act attributed to an accused which caused the unintentional death of an accomplice with the accused in the foundational crime did not result in the accused becoming the constructive murderer of the accomplice in the same way as the accused would have become a constructive murderer if the same act had caused the unintentional death of a third person. Since 1983, nothing has changed. For these reasons, I am unable to agree with Kiefel CJ, Keane and Edelman JJ that the appeal in relation to constructive murder can be allowed on the basis that Lan's lighting of the ring burner resulted only in the death of Lan. Lan's killing of himself is also no answer to manslaughter Given that Kiefel CJ, Keane and Edelman JJ and Bell and Nettle JJ do not address manslaughter as a discrete topic, the explanation for my dissent on that topic can be brief. Unlike s 18(1)(a), which defines murder, s 18(1)(b) merely provides that every punishable homicide other than murder is taken to be manslaughter. The elements of the offence of manslaughter are supplied by the common law. According to the taxonomy of the common law, manslaughter can be either voluntary manslaughter or involuntary manslaughter, and involuntary manslaughter can be either manslaughter by criminal negligence or manslaughter by an unlawful and dangerous act carrying with it an appreciable risk of serious injury. The prosecution case in manslaughter against IL was limited to the last of those sub-categories: involuntary manslaughter by an unlawful and dangerous act constituted by lighting the ring burner carrying with it an appreciable risk of serious injury. Lighting the ring burner was plainly not unlawful in itself. Lighting the ring burner was unlawful because it was an incident of executing the joint enterprise of manufacturing methylamphetamine. It was open to the jury to conclude that, to the knowledge of Lan and of IL, the act of lighting the ring burner was dangerous and carried an appreciable risk of serious injury. The common law never recognised involuntary manslaughter to extend to self-manslaughter. Lan's criminal responsibility for his own unlawful and dangerous act could not have resulted in Lan being guilty of his own manslaughter. But that is no answer to the case against IL, which turns on IL having primary criminal responsibility for Lan's act. The common law doctrine of joint criminal enterprise, which applies to attribute responsibility for the act of Lan to IL for the purpose of constructive murder, applies also to attribute responsibility for that act to IL for the purpose of involuntary manslaughter. If the joint enterprise in which IL and Lan participated was contemplated by IL to encompass Lan lighting the ring burner as an incident of manufacturing methylamphetamine, the verdict of manslaughter was available as an alternative to murder. GORDON J. Felony murder has a long history. In general terms, the common law felony murder rule provides that a person who causes the death of another in the course of committing another offence of a specified category (the "foundational offence") is guilty of murder regardless of whether that person had the mental element otherwise required for the offence of murder. From at least the 1860s, the felony murder rule has been the subject of strident criticism. Despite those concerns about the felony murder rule, New South Wales adopted the common law felony murder rule in an amended form by the enactment of s 9 of the Criminal Law Amendment Act 1883 (NSW). As Windeyer J said in Ryan v The Queen: "Sir Alfred Stephen and Alexander Oliver, Parliamentary Draftsmen, said in their commentary on the Act, and were it would seem right in saying, that the effect of the New South Wales definition was to banish the expression 'malice aforethought', but not otherwise to alter the common law, except in one respect. 'In one particular', they said, 'there is an important difference. The accidental taking of life, by a person committing (or about to commit) a felony of any kind, is by the common law murder. Under the ninth section it will not amount to that crime unless the felony was a capital one or punishable by penal servitude for life'". (emphasis added) Section 9 of the 1883 Act was re-enacted as s 18(1) of the Crimes Act 1900 (NSW). In its statutory form, the felony murder rule is commonly referred to as "constructive murder". Notwithstanding continued criticisms of both felony murder and constructive murder, s 18, headed "Murder and manslaughter defined", continues to include constructive murder as one method of establishing the offence of murder. Constructive murder is not a separate offence. Section 18 of the Crimes Act Section 18 relevantly provides that: "(1) (a) Murder shall be taken to have been committed where the act of the accused, … causing the death charged, was done … with reckless indifference to human life, or with intent to kill or inflict grievous bodily harm upon some person, or done in an attempt to commit, or during or immediately after the commission, by the accused, or some accomplice with him or her, of a crime punishable by imprisonment for life or for 25 years. (b) Every other punishable homicide shall be taken to be manslaughter. (2) (a) No act or omission which was not malicious, or for which the accused had lawful cause or excuse, shall be within this section. …" (emphasis added) Although this appeal is concerned with the italicised words in s 18(1)(a) – constructive murder – the whole section is important. For present purposes, s 18(1)(a) comprises a number of distinct but interrelated elements. First, it requires an "act of the accused … causing the death charged". That element is common to each method of establishing murder. That element has the same meaning and operation for each method. Second, that element – "the act of the accused … causing the death charged" – must take one of two forms. It must be an act done: (1) with reckless indifference to human life or with intent to kill or inflict grievous bodily harm; or (2) in an attempt to commit, or during or immediately after the commission by the accused or some accomplice of, a crime punishable by imprisonment for life or for 25 years (the foundational offence) – constructive murder. The distinction drawn in s 18(1)(a) between the two categories of murder (which is consistent with the common law and its approach to liability for murder) was identified in 1942 by Jordan CJ in R v Surridge. First category In relation to the first category, Jordan CJ in Surridge stated that the special mention of the accomplice in the second category did not exclude accomplices from liability for murder in the first category because of the common law principles of complicity. That point is illustrated by R v Grand and Jones. Mr Grand, along with another person, was convicted of murder for a death that occurred during a burglary as a result of a gunshot. The jury were directed that "if they thought that the persons who committed the burglary went there resolved, if necessary, to resist at all hazards any interference with them in the execution of their purpose, and the constable was killed by one of them, it was not necessary to determine which of them fired the fatal shot". In the course of upholding the conviction, Stephen ACJ said: "I do not think that s 18 was intended to alter the Common Law, that where two persons go out with the common intention of committing a crime and of resisting any opposition by force, and death ensues as the result of the use of force, then both of those engaged are guilty of murder." In the circumstances described by Stephen ACJ, "both of those engaged are guilty of murder" because "the killing was within the common purpose". Grand was a case in the first category because, at the time, burglary was not capable of being a foundational offence for the purpose of constructive murder. The case does not directly address constructive murder. It does, however, illustrate the meaning to be given to "the act of the accused ... causing the death charged" in s 18(1)(a). It is not always necessary to show that the accused applied the fatal force. Second category In relation to the second category, Jordan CJ in Surridge relevantly stated that for an accomplice to be guilty of constructive murder, all that is required under s 18(1)(a) is that it was within the common purpose of the parties that an offence capable of being a foundational offence be committed and that "the cause of the death [ie, the act(s)] must have been something done by the other in an attempt to commit or during or immediately after the commission of that … crime". Here, the Crown did not run a case against IL based on the first category. As Simpson JA noted in the Court of Criminal Appeal of the Supreme Court of New South Wales, the Crown could not "realistically proceed" on the basis of the first category. The Crown's case against IL was based solely on the second category. On the approach to the second category explained by Jordan CJ in Surridge, IL could be found guilty of constructive murder. The Crown could make out a case that it was within the common purpose of IL and Mr Lan to commit an offence capable of being a foundational offence, namely the manufacture of a large commercial quantity of a prohibited drug, and that the act "causing the death charged" – the act of lighting the ring burner – was something done by the other (Mr Lan) during the commission of the foundational offence. Complicity and constructive murder under s 18 The application of the constructive murder limb of s 18(1)(a) is harsh. But the statutory rule and its common law antecedent have persisted, notwithstanding that they have been criticised for over 150 years. Common law complicity may be understood as infringing on, and in some instances taking the place of, constructive murder. But while constructive murder remains on the statute books, the provision must be applied where it is sought to be relied upon. The outcome may be thought unattractive, but that feeling of unease cannot be met by ignoring the terms of the provision. And the further developments of the common law of complicity do not require a departure from Surridge in relation to either category. Rather, those developments assist in explaining how "the act of the accused ... causing the death charged" in s 18(1)(a) encompasses acts done by another. Osland v The Queen is not a constructive murder case, but it is one example of liability for murder where the accused did not personally apply the fatal force. In that case, the deceased was killed in accordance with a common understanding or arrangement between the accused and another person, with the other person striking the fatal blow or blows in the presence of the accused. The accused, who did not personally do the physical act or acts, was convicted of murder even though the person who did do the physical act or acts was acquitted. That outcome was possible because, as McHugh J (with whom Kirby J and Callinan J agreed) explained, "it is the acts constituting the actus reus, and not the crime, of the actual offender which are attributed to the other party. The liability is direct or primary, not derivative". The distinction between primary and derivative liability may sometimes be important. The person who does the act might have a defence that diminishes or absolves that person of criminal responsibility. But that does not mean the other party to the enterprise cannot be convicted of the more serious offence. As McHugh J said: "Once the parties have agreed to do the acts which constitute the actus reus of the offence and are present acting in concert when the acts are committed, the criminal liability of each should depend upon the existence or non-existence of mens rea or upon their having a lawful justification for the acts, not upon the criminal liability of the actual perpetrator." And as was explained by the Court of Criminal Appeal of the Supreme Court of New South Wales in Tangye, quoted with approval by McHugh J in Osland, "where two or more persons carry out a joint criminal enterprise, each is responsible for the acts of the other or others in carrying out that enterprise" (emphasis added). Each person is responsible for the acts of the other or others "because they have agreed to them being done". For the constructive murder limb of s 18(1)(a), where it is established that the accused was a party to a joint criminal enterprise to commit an offence capable of being a foundational offence, the accused is responsible for the acts of the other party to the agreement that were done in carrying out that enterprise. An act of the other party done in those circumstances can be "the act of the accused" for the purpose of s 18(1)(a). It follows that, if the act of the other party causes the death charged and the act is done in an attempt to commit, or during or immediately after the commission of, the foundational offence, then the accused will be liable for constructive murder. As Jordan CJ correctly explained in Surridge, nothing more is required. As explained above, those elements were capable of being satisfied in this case: (1) "There was no real dispute that the Crown could make out a case that [IL] and Mr Lan were engaged in a joint criminal enterprise to manufacture a large commercial quantity of a prohibited drug". (2) The act of lighting the ring burner was done by Mr Lan in carrying out the enterprise. That act caused "the death charged" and was done during the commission of the foundational offence. (3) IL was responsible for that act as it was an act done in carrying out the joint criminal enterprise, such that it could be "the act of the accused" for the purpose of s 18(1)(a). Distinguishing between an act and the actus reus of a crime is not useful when considering the application of s 18(1)(a). Introducing a distinction of that kind departs from the statutory words used in the provision. It is a departure because introducing a distinction of that kind necessarily attributes a different meaning to the phrase "the act of the accused ... causing the death charged" for the purpose of constructive murder from the meaning of the phrase for the purpose of the first category of murder. On that approach, it is not doubted that for the first category, the acts of the parties to a joint criminal enterprise that, between them, comprise the actus reus of an offence within the scope of the agreement can be relied on to establish murder. But the same approach leads to the result that, for the second category, the acts of the parties to a joint criminal enterprise that, between them, comprise the actus reus of an offence within the scope of the agreement and which is a foundational offence cannot be relied on to establish constructive murder. There is no basis for that distinction and one has not been identified. Moreover, observing that the common law principles of complicity do not apply to render IL guilty of murder within the first category does not address the issue of how the common law principles of complicity inform (as they do) what is the relevant "act of the accused ... causing the death charged" for the purpose of the second category – constructive murder. In the present case, that question (what is the act of the accused causing death) is to be answered by applying the common law principles of complicity to the foundational offence. The consequence of applying those principles is that the accused (IL) is responsible for all of the steps taken by Mr Lan towards manufacturing the methylamphetamine, including the lighting of the burner. Once that is recognised, the question whether Mr Lan could have been liable for homicide does not arise. First, as noted above, Osland makes it clear that IL's liability for constructive murder would be primary, not derivative. Thus, Mr Lan's liability or otherwise for murder is irrelevant. Second, consideration of the question whether Mr Lan could have been liable for homicide distracts attention from the relevant statutory question. The statutory question is not answered by observing that s 18(1)(a) requires the death of a person other than the person who did the act. Once it is understood that Mr Lan's act of lighting the burner was "the act of the accused" – that is, the act of IL – then that act did cause the death of another. Constructive murder is not a separate offence; it is one method of establishing the offence of murder under s 18(1)(a). Observations in other cases involving the offence of murder or manslaughter under different statutory regimes and different facts must be approached with caution. They are in different terms. And observing that this method of establishing murder under s 18(1)(a) – constructive murder – leads to harsh and potentially absurd results does no more than highlight the continued difficulties of this method of establishing murder remaining on the statute books. Nevertheless, the liability of accomplices for constructive murder is not at large. As with any charge of murder under s 18(1)(a), the act must be an act of the accused, the act must be voluntary and the act must "caus[e] the death charged". Each of these may be a live issue, where, for example, a person who is a party to a joint criminal enterprise, in the course of carrying out that enterprise, is hit by a bus while crossing the road or slips and falls from a height, or where machinery malfunctions. But in this appeal, neither voluntariness nor causation was in issue. This view of the operation of the constructive murder limb of s 18(1)(a) is consistent with the common law rule of felony murder and with the approach adopted in cases both before and after Surridge. Put another way, subsequent cases in New South Wales do not require a departure from Surridge. In particular, subject to a line of authority beginning with R v Sharah, which proceeds on a wrong premise and to which it will be necessary to return, the cases do not stand for the proposition that where a foundational offence has been committed, liability for constructive murder depends upon anything more than establishing (1) an agreement between the parties to commit an offence capable of being a foundational offence and (2) that the act causing the death was done in an attempt to commit, or during or immediately after the commission of, that foundational offence. Before dealing with some of the New South Wales cases, it is useful to say something about the common law principles that had developed in relation to felony murder. It is to be recalled that s 18 was intended to encompass those common law principles. And, as explained in these reasons, that is what the words of the section achieve. The common law position was summarised in R v Solomon in the following terms: "By the common law if the victim of robbery, which is a felony involving violence, be killed in the course of the robbery all parties to the robbery are guilty of murder. The probability or possibility that homicide would or would not be done is irrelevant. The fact that the homicide occurred independently of the exercise of the will of one of the accomplices would not exonerate him." The common law approach was followed by the Full Court of the Supreme Court of South Australia in R v R, a case concerning the common law felony murder rule in that State. In that case, an accomplice in a joint criminal enterprise to commit an armed robbery contended that the criminal liability of a principal in the second degree, or of an accessory before the fact, to a felony for a murder committed in the course of the felony should be determined according to the common purpose rule laid down in Johns v The Queen. In other words, the jury should have been directed that the accomplice was guilty of murder only if it was in his contemplation that the principal would or might inflict a fatal stab wound. The contention was rejected. King CJ, with whom the other four members of the Court agreed, observed: "[T]he argument on grounds of policy is really an attack on the felony murder rule itself. If the policy is accepted that the actual perpetrator should be liable for the unintended consequences of his actions in the course of the felony because in engaging in a violent or dangerous felony he must accept responsibility for what occurs in the course of that felony, even though unintended, there appears to be no reason of policy why other participants in the felony should not also have to accept the same responsibility." The common law rule was replaced in South Australia in 1994 by a constructive murder provision analogous to s 18(1)(a). That section provides: "A person who commits an intentional act of violence while acting in the course or furtherance of a major indictable offence punishable by imprisonment for ten years or more (other than abortion), and thus causes the death of another, is guilty of murder." (note omitted) That provision was considered by this Court in Arulthilakan v The Queen. In directions to the jury the trial judge said: "As to joint enterprise, if you are satisfied that all of the accused had the common purpose that they would roll or rob Hillam, and for the purpose of their joint enterprise they would be armed with knives and a billiard ball, that they would use the knives and billiard ball if necessary to achieve their purpose in the course of the attempted armed robbery, the knives or billiard ball would be used to threaten or intimidate the victims. I realise that sounds very similar to the concept of joint enterprise in relation to common law murder, but it is different in that it is not necessary that any of the accused had an intention to cause death, or to cause grievous bodily harm, or contemplated as part of the joint enterprise the possibility that the use of the knives could result in an intentional inflicting of grievous bodily harm. That is the difference between them." (emphasis added) That aspect of the directions was not challenged on appeal and there was no suggestion in Arulthilakan (or in this appeal) that it was deficient. Subject to one qualification, the cases decided in New South Wales about the application of s 18(1)(a) are consistent with the conclusions expressed earlier in these reasons. As already noted, Grand, decided in 1903, was not a constructive murder case. It was a common purpose case where the agreement extended to "resist[ing] at all hazards any interference with [the parties to the agreement] in the execution of their purpose". The acts done in furtherance of committing the agreed offence were attributable to both participants; the acts would be "the act of the accused" for both participants under the first category of murder in s 18(1)(a). Sharah was a constructive murder case. During the course of an armed robbery with wounding, an offender killed a person and wounded another person. Mr Sharah was charged as an accomplice to the armed robbery with wounding and with both categories of murder. In relation to Mr Sharah's culpability for the foundational offence of armed robbery with wounding, the Court of Criminal Appeal said that the prosecution had to prove that: (1) there was a common purpose between Mr Sharah and his co-offender in company to rob while his co‑offender was, to Mr Sharah's knowledge, armed with an offensive weapon; (2) during the course of the armed robbery, the co-offender wounded the victim of the armed robbery; and (3) Mr Sharah contemplated that, in carrying out the common unlawful purpose of armed robbery, such wounding might occur. In relation to what the prosecution had to establish for constructive murder in Sharah, the Court held that the prosecution had to prove that: (1) there was a common purpose between Mr Sharah and his co-offender in company to rob while his co‑offender was, to Mr Sharah's knowledge, armed with an offensive weapon; (2) during the course of the armed robbery, the co-offender wounded the victim of the armed robbery, and, during the course of such armed robbery with wounding or immediately thereafter, the co-offender discharged the weapon causing the death of another person; and (3) Mr Sharah had in mind the contingency that his co‑offender would discharge the weapon during or immediately after the armed robbery with wounding, whether or not the weapon was fired intentionally and whether or not in furtherance of the common unlawful purpose. As is apparent, the Court imported a third element – an additional limited mental or fault element – for constructive murder in the case of Mr Sharah, an accomplice. As the New South Wales Law Reform Commission noted: "The historical basis for [the third element] is unclear. It may be that it was thought appropriate to draw, by analogy, on the approach that had been developed, in relation to joint criminal enterprise liability; or perhaps, that the case was seen as one to which that form of liability applied." Sharah has been followed in New South Wales but it should not be: to the extent that it is authority for the proposition that it is necessary to establish the third element for the purpose of constructive murder, it is wrong. There is no foundation for the inclusion of that element. As the New South Wales Law Reform Commission correctly noted, until Sharah the constructive murder limb of s 18(1)(a) had, in relation to both the principal and the accomplice, required only that the act or omission causing death be connected (in the sense already explained) with the acts forming part of the foundational offence. If the parties have agreed to commit the foundational offence, neither felony murder at common law nor constructive murder under s 18(1)(a) required or requires any additional foresight or contemplation on the part of the accomplice. Malice In this case, the Court of Criminal Appeal allowed the Crown's appeal and quashed the directed verdicts of acquittal on counts 2a and 2b. In considering whether to exercise its discretion not to order a new trial, the Court of Criminal Appeal considered a contention that a new trial on the count of murder (count 2a) should not be ordered because the Crown could not prove that the act causing death was done maliciously as required by s 18(2)(a). The Court of Criminal Appeal concluded that the lighting of the ring burner during the commission of the foundational offence was malicious within the meaning of s 18(2)(a); in addition, it could properly be regarded as reckless within the meaning of s 5 of the Crimes Act with the consequence that it could be taken to have been done maliciously even if it was not otherwise done with malice. As I am in dissent in this appeal, it is sufficient if I set out my conclusions as follows. First, s 18(1) did not depart from the common law, except for altering the type of felony necessary for felony murder. Section 18(1) was a statutory reformulation of the element of malice in the crime of murder; it "replaced the common law concept of malice aforethought with a list of matters that would previously have established malice aforethought". Second, for constructive murder, once the mental element for the foundational offence is established to the requisite standard, malice is also established. As Toohey and Gaudron JJ said in Royall v The Queen, "in the case of the murder-felony rule, the commission of the felony satisfies any requirement of malice". That is, for the purposes of constructive murder, s 18(2)(a) will be satisfied by proof of the foundational offence. As the plurality in R v Lavender stated, the effect of s 18(1)(a) is that certain forms of punishable homicide are taken to be murder and thereby satisfy s 18(2)(a). If the act is not one of those in s 18(1)(a), then it is manslaughter. Manslaughter The second count charged IL with the alternative charge of manslaughter on the basis that the death of Mr Lan was caused by an unlawful and dangerous act – the lighting of the ring burner. On the analysis of the constructive murder limb of s 18(1)(a) explained in these reasons, the alternative charge of manslaughter was unnecessary in the circumstances of this case. It was unnecessary because the act relied upon was done in the course of committing an offence capable of being a foundational offence for a constructive murder charge, and the act was done by one of the parties to the joint criminal enterprise to commit that offence. That is not to say that an alternative verdict of manslaughter will never be available in a case based on constructive murder. For example, "[t]here may be an issue of fact to be resolved by the jury as to whether the act causing death occurred immediately after the commission of the foundational offence, or whether such act was too remote in time. If the jury were not satisfied that the relevant act occurred sufficiently proximately to the commission of the foundational offence, it would, nevertheless, be open to them in an appropriate case, to convict the accused of manslaughter on the basis that the act which caused death was an unlawful and dangerous act". And, of course, if the offence the subject of the joint criminal enterprise cannot be a foundational offence, manslaughter may be the only option available to the Crown. As the plurality said in Lavender, s 18 defines murder; it does not define manslaughter. It provides that murder is punishable homicide when it involves one of the elements in s 18(1)(a). Or, put differently, the section refers to manslaughter but only in excluding from the category of murder any form of punishable homicide "which does not satisfy s 18(1)(a)". In any application of s 18, there must be a death. The section then asks whether a person is criminally responsible for that death and, if so, which classification of responsibility (if any) attaches to that death. There are several paths to criminal responsibility. As the section states, it is murder when it involves one of the elements in s 18(1)(a) – the positive definition of murder, divisible into the two categories explained earlier. If the person is not criminally liable for the offence of murder by one of the several paths provided for by s 18(1)(a), then the section directs the reader to determine whether the person is criminally responsible for the death in manslaughter – the negative definition of murder. As the plurality said in Lavender, the reader must go to the common law of manslaughter to answer that question. Constructive murder – the second category of murder – sits uncomfortably within that structure because a person can be found guilty of murder even if the person has not intended to kill or inflict grievous bodily harm or has not acted with reckless indifference to human life. Further, in conjunction with the principles of joint criminal enterprise, constructive murder can operate where the person has not applied the fatal force. But observing that a person can be held criminally responsible for murder even though they did not have the requisite state of mind for murder in the first category, and did not apply the fatal force, does not create a unique legal conundrum. The same consequence follows from the application of the principles of extended joint criminal enterprise to the first category of murder. Order For those reasons, I would dismiss the appeal.