Perry v The Queen

[1982] HCA 75

HIGH COURT OF AUSTRALIA

Gibbs C.J., Murphy, Wilson and Brennan JJ. The Hon. Mr Justice Aickin died before reasons for judgment were published.

PERRY v. THE QUEEN

(1982) 150 CLR 580

16 December 1982

Evidence

Evidence—Criminal trial—Admissibility—Similar facts—Charge of attempt to murder husband by arsenic poisoning—Husband suffering from arsenic poisoning in 1978—Death of first husband of arsenic poisoning in 1961—Death of brother of arsenic poisoning in 1962—Death of de facto husband from overdose of barbiturates in 1970—Evidence of earlier deaths.

Decisions

1982, December 16.
The following written reasons for judgment were published:-
GIBBS C.J. The applicant, Mrs. Perry, was tried on indictment before the Supreme Court of South Australia on two charges of attempting to murder Mr. Perry, who is her third husband. The case for the prosecution was that from about 1 July 1978 until about 30 November 1978, and again from about 1 February 1979 until about 31 October 1979, the applicant administered poison to her husband with intent to kill him. She was convicted on both counts. She appealed to the Court of Criminal Appeal on a number of grounds, but her appeal was dismissed. She then applied to this Court for special leave to appeal. The main ground on which the application was made (and the only ground which the Court permitted to be argued in full) was that the learned trial judge had wrongly admitted evidence which, according to the Crown, showed that three other persons with whom the applicant had had a close relationship had died of poisoning. This Court granted special leave, allowed the appeal, set aside the conviction and ordered a new trial but announced that the reasons for the decision would be given later. I now give my reasons for joining in the decision. (at p582)

2. There was clear evidence that during part at least of the periods in question the applicant's husband, Mr. Perry, became seriously ill as a result of poisoning by lead and arsenic, probably in the form of lead arsenate. The applicant, who was living with Mr. Perry, had the opportunity to administer the poison to him. She stood to benefit financially from his death, under a number of policies of insurance. It is unnecessary to review the details of the evidence against the applicant, other than that whose admissibility is in question, and undesirable to comment on its weight, since a new trial has been ordered. The defence was that the poison had been accidentally ingested by Mr. Perry over a period of time and that the applicant had not played any part in causing him to take it. Mr. Perry himself gave evidence for the defence and asserted the innocence of the applicant. The theory advanced by the defence was that Mr. Perry must have ingested the poison while he was working on the renovation of some old musical instruments, especially an orchestrelle (a sort of large organ), which contained the remains of rat or insect poison, one of whose components was arsenic, and contained also lead dust from crumbling lead pipes in the instruments. (at p582)

3. The evidence to which objection was taken concerned the deaths of Albert Otto Haag, the second husband of the applicant, Francis William Montgomerie, her brother, and John Alfred Jameson, otherwise known as Jim Duncan, a man with whom the applicant was living in a de facto relationship. There was evidence that Constable Haag died on 13 March 1961 of acute arsenic poisoning. He had previously been ill in December 1960 and January 1961, and there was evidence that this illness might have been due to arsenical poisoning. The applicant stood to benefit from insurance policies on his life. The defence suggested that the death was due either to accident, possibly caused by the fact that Constable Haag ate some corn which had been sprayed with weed killer of a kind which contained arsenic; alternatively, suicide was suggested. The applicant had some knowledge of poisons, had bought the weed killer which Constable Haag had used for spraying, and had an opportunity to administer it to her husband. There was evidence of domestic trouble between the spouses and evidence that the applicant had made false statements concerning her knowledge of the insurance policies and the state of the domestic relationship. An inquest was held, but no charge was laid against the applicant. (at p583)

4. Mr. Montgomerie died on 9 April 1962, also from acute arsenical poisoning. He was an alcoholic, and was given to moodiness and depression; he had twice been admitted to a psychiatric hospital. He had been living with a Mrs. Walker, who had left him on 7th April 1962, after he had attempted to strangle her. On the following day, 8 April, the applicant took her mother to Montgomerie's house, so that she might look after him. The applicant said that Montgomerie was then incoherent, as he usually was after drinking heavily. She left her mother there for the day but picked her up that evening. On 9 April she telephoned Montgomerie, and receiving no response, went to his house, where she found him dead. The applicant may have been the last person to see Montgomerie alive, and the first to find him dead. Some bottles were found on a bedside table, including a bottle containing a small quantity of wine and some arsenic. A post mortem examination showed arsenic in the body. There was evidence that a sister of the applicant had, a few days before, bought some weed killer, the nature of which was not established by the evidence, and that the container, with a little of the weed killer in it, had been placed by the applicant on a shelf at her mother's house. This, it was said, showed that the applicant had access to the poison. The defence argued that the facts pointed clearly to suicide. The Crown argued that if Montgomerie had committed suicide one would have expected to find the bottle or tin from which the arsenic had been obtained and put into the wine bottle; none was found, but there was evidence that old bottles at Montgomerie's house had on the morning of the death been cleaned up and thrown away by the applicant and her sister, Mrs. Henderson. There was also evidence that the applicant, when questioned by the police, gave her name as Hulse, which was the name of her first husband. Many of the facts relied on by the Crown were contested. Montgomerie had no insurance policies and the applicant did not stand to gain from his death. The only motive for murder suggested by the Crown was a desire to rid the family, and particularly the applicant's mother, of a tiresome burden. (at p584)

5. Mr. Duncan died on 21 March 1970 from an overdose of barbiturates. He had commenced to live with the applicant about the end of 1967. Insurance policies on his life had been arranged by the applicant during 1968 and she received payment under them after his death. He had swallowed about twenty tablets of barbiturates; it is most unlikely that the applicant could have forced him to swallow them, and although it was suggested by the Crown that perhaps the applicant had persuaded him to take them, there was nothing to support that suggestion. Duncan was a heavy drinker, and in bad health, and one possibility was that he had committed suicide. However, the Crown particularly relied on evidence which, it was submitted, showed that Duncan had suffered from arsenical poisoning over a considerable period before his death. Duncan had had operations for haemorrhoids in 1944 and 1957. By the beginning of 1968 he was seeking medical treatment for anal trouble, and underwent another operation for haemorrhoids during that year. From about the end of that year he was complaining of some loss of bowel control and pains in the lower abdomen, and by the end of 1969 and the beginning of 1970, of diarrhoea, incontinence, vomiting, pain, and other symptoms, including enlargement of the breasts. None of the many doctors who examined him diagnosed heavy metal poisoning, and no test was made, before his death or post mortem, for the presence of arsenic. A witness for the Crown, Dr. Manock, said that Duncan's symptoms, as described to him, were consistent with, but not specific for, the ingestion of lead arsenate. However, he saw no signs of heavy metal poisoning when he conducted an autopsy. A witness for the defence, Dr. Lander, thought it unlikely that Duncan was suffering from heavy metal poisoning, and said that some symptoms were contra-indicative of such poisoning. It was conceded by the Crown that the evidence relating to Duncan, if viewed in isolation, was insufficient to enable a jury to find that he had suffered from heavy metal poisoning. (at p584)

6. The principles governing the admissibility of evidence of similar facts in criminal cases were quite recently restated in Markby v. The Queen (1978) 140 CLR 108, at pp 116-117 . They derive from the well known statement of Lord Herschell L.C. in Makin v. Attorney-General (N.S.W.) (1894) AC 57, at p 65 . The prosecution cannot adduce evidence tending to show that the accused has been guilty of criminal acts other than those with which he is charged if the evidence shows only that he had a propensity to commit crime, or crime of a particular kind, or that he was the sort of person likely to have committed the crime charged. On the other hand, if the evidence is relevant in some other way, it is admissible even though it reveals that the accused was disposed or likely to commit the sort of crime with which he is charged. Cases since Makin v. Attorney-General (N.S.W.) have established that even if the evidence is legally admissible, the trial judge has a discretion to exclude it if its prejudicial effect is likely to be out of proportion to its true evidential value. (at p585)

7. Evidence that an accused person has a propensity to commit crimes of the sort with which he is charged, or is the sort of person who is likely to commit such crimes, would ordinarily be regarded as relevant to the question whether he did commit the offence in question. Such evidence is excluded, not because it is irrelevant, but because it is likely to be unfairly prejudicial to the accused. A jury might attach too much importance to it. It would, however, be a mistake to think that because the reason for the rule is that the evidence would be unfairly prejudicial, the rule itself does no more than require the judge to exercise a discretion, and to weigh the prejudicial effect of the evidence against its probative value. Many cases, including Reg. v. Boardman (1975) AC 421 , Harris v. Director of Public Prosecutions (1952) AC 694 , and Markby v. The Queen, to name only a few, have since recognized that the principles laid down in Makin v. Attorney-General (N.S.W.) are authoritative. A trial judge must decide as a matter of law whether the evidence is admissible, and it is only if he decides that it answers the test of admissibility that he need consider whether he should exclude it in the exercise of his discretion. The importance of the distinction between the two steps which the trial judge is required to take can be seen if the case goes on appeal, since the grounds on which a court of criminal appeal will intervene will be different depending on whether inadmissible evidence has been received or there is a challenge to the exercise of the judge's discretion. (at p585)

8. It is true that in deciding whether the evidence is admissible, questions of degree arise and that the judgment to be made is to some extent discretionary. It was said in Harris v. Director of Public Prosecutions (1952) AC, at p 710 , that evidence of similar facts, to be admissible, must have "a really material bearing" on the issues to be decided. The judgments in Reg. v. Boardman show that there are degrees of relevance, and I respectfully agree with the statement of Lord Wilberforce that the evidence to be admissible must have "a strong degree of probative force": Reg. v. Boardman (1955) AC, at p 444; see also at pp 439, 452-453, 456 . In Reg. v. Chee (1980) VR 303, at p 308 , the Full Court of the Supreme Court of Victoria held that it was enough that the similar fact evidence should have some probative force, by which no doubt was meant probative force other than merely by way of showing that the accused had a propensity to commit the sort of crime with which he was charged or was the sort of person who was likely to commit such a crime. It was there said that it was not a condition of admissibility that the evidence should have a high degree of probative force. With all respect, it is not right to treat evidence which tends to show the commission by the accused of other criminal acts in the same way as any other circumstantial evidence. In the first place, as I have already said, a jury might place too much weight on the fact that the accused had a criminal tendency. Secondly, evidence of this kind will often raise difficult and doubtful questions as to whether the accused had in fact been guilty of other criminal acts, and may distract the attention of the jury from the vital issues in the case. It is therefore not enough that the evidence should be only technically relevant (otherwise than as showing a propensity); it must be really material; it must have strong probative force. (at p586)

9. There are various ways in which evidence of similar facts may tend to show, otherwise than via propensity, that the accused committed the act which forms the subject of the charge. Some examples may be given, although they are of course not intended to be exhaustive. Prior acts of sexual intercourse may establish a relationship whose existence is relevant to the question whether the accused committed incest: R. v. Ball (1911) AC 47 . The fact that a person engages in a series of acts which form part of a continuous business operation, and which would ordinarily be accompanied by the payment of money, may be relevant to show that money was paid on a particular occasion: Martin v. Osborne (1936) 55 CLR 367 . The fact that an accused has, in the past, committed crimes in a particular, unusual, manner may be relevant to show that he was the person who committed the crime in question, when that was committed in the same characteristic manner: Reg. v. Straffen (1952) 2 QB 911 . In some cases the frequency with which a particular set of circumstances has occurred may, having regard to ordinary human experience, make it unreasonable to suppose that they have occurred other than by design. Makin v. Attorney-General (N.S.W.) and R. v. Smith (1915) 11 CrAppR 229 (the case of the brides in the bath) are cases of this kind. In cases of the last two kinds it may be important to consider whether there is a "striking similarity" between the similar facts sought to be proved and the facts in issue. (at p587)

10. Clearly on principle it is not admissible, on a charge of murder or attempted murder by poisoning, to give evidence that the accused has poisoned other persons, where that evidence shows no more than that the accused is a poisoner - one who has a tendency to poison others. However, where a number of poisonings have occurred, and the victims have all been associated with the accused person, the evidence of the other poisonings may be admissible to support the inference that the accused was responsible for the death in issue, because it would be contrary to ordinary experience that a series of poisonings, caused by accident or suicide, would occur by coincidence in the circle of persons with whom the accused was associated. In such cases, the evidence is admissible for reasons similar to those which governed the admissibility of the evidence in Makin v. Attorney-General (N.S.W.) and R. v. Smith, although it must be said that the reasons given in the judgments have not always explained the admissibility of the evidence in this way. A leading case is Reg. v. Geering (1849) 18 LJMC 215 , a case which has been mentioned with approval in the House of Lords, the Judicial Committee and this Court. Geering was charged with the murder of her husband by arsenical poisoning, and evidence was admitted that within the seven months following her husband's death three of her sons also suffered from arsenical poisoning; two of them died. All of the victims lived in the same house as the accused, who prepared their food. Pollock C.B. said that the evidence was admissible to confirm that the death of the husband was occasioned by arsenic; it was further argued that the evidence was admissible to prove that the death was not accidental, and the headnote states that the decision rested on that ground also. It appears that there was direct evidence that the husband was poisoned by arsenic but it is not clear from the report whether there was any dispute as to the cause of the husband's death. If the cause of death was only technically in issue, the evidence should not, in my opinion, have been admitted to confirm that the husband's death was occasioned by arsenic. To use the evidence for that purpose, in those circumstances, would have been to credit the accused with a fancy defence in order to rebut it with a damning piece of prejudice, to use the words of Lord Sumner in Thompson v. The King (1918) AC 221, at p 232 . In Martin v. Osborne the judgments of Dixon J. (1936) 55 CLR, at p 376 and Evatt J. (1936) 55 CLR, at pp 384-385 show that it is the improbability that a number of deaths would occur in similar circumstances merely by coincidence that gives the evidence its probative force in such cases. Dixon J. said (1936) 55 CLR, at p 376 :

"If four close relatives of a woman, dwelling in her house and eating meals prepared by her, die of arsenical poisoning one after another within a few months, the inference that she wilfully administered the poison has more support than if one death only occurred in such circumstances (R. v. Geering)."

Another case in which evidence of this kind was clearly admissible on the same principle is Reg. v. Grills (1954) 73 WN (NSW) 303 . In that case there was evidence of a series of thallium poisonings in a small group of people, and that the accused was the only person associated with them all. The frequency with which the symptoms of poisoning followed upon the association of the victim with the accused excluded as unreasonable any explanation of the death of the person with whose murder she was charged other than that she had administered the poison (1954) 73 WN (NSW), at p 304 . Other cases are not so easy to explain. In R. v. Armstrong (1922) 2 KB 555 , the accused was charged with the murder of his wife by arsenical poisoning. The defence was suicide. Arsenic was found in his possession but he claimed that it was for use as a weed killer. Evidence was admitted that eight months after the death of his wife he attempted to poison another person with arsenic. It was said that the fact that the accused used arsenic for a similar deadly purpose was evidence from which the jury might infer that the poison was not in his possession for an innocent purpose (1922) 2 KB, at p 566 . With all respect, the manner in which the poison was used eight months later hardly seems relevant to the question for what purpose it had been kept at the time of the murder. The evidence would no doubt have been relevant had it been suggested that the wife's death was accidental, but that was not suggested. However, it would not be profitable to discuss the facts of the various cases in which evidence of poisonings has been either admitted or rejected, for one cannot decide the question whether evidence is admissible in one case simply by comparing the facts of that case with those of another. What is necessary is to apply the principles to the circumstances of the case under consideration. (at p589)

11. I have already said that evidence tendered to prove the similar fact which it is sought to admit in accordance with these principles must have a strong degree of probative force. It is not enough that it merely raises a suspicion that the accused may have been guilty of the similar misconduct alleged or of the crime charged. Two leading cases in which the evidence was rejected because it was not sufficiently cogent are Noor Mohamed v. The King (1949) AC 182 and Harris v. Director of Public Prosecutions (1952) AC 694 . In the former case the accused was charged with having murdered his mistress with cyanide. Evidence that a previous mistress of the accused had died by cyanide poisoning was rejected. In each case it had been shown that ill-feeling existed between the accused and the woman and that the accused was in possession of cyanide which, however, he used in his trade as a goldsmith. I need not refer in detail to the facts of the case, which are fully examined in the judgment of Owen J. in Reg. v. Fletcher (1953) 53 SR (NSW) 70, at pp 77-79 . Owen J. would have said that there was no evidence that the accused caused the first of his mistresses to take poison. However, the Judicial Committee, which accepted that it could be inferred that the appellant had tricked the first woman into taking poison, nevertheless held that the evidence should have been rejected because it was of no real substance apart from showing that the accused was a person who from his past conduct was likely to commit murder (1949) AC, at pp 192-193 . Similarly in Harris v. Director of Public Prosecutions, a conviction on a count of larceny was quashed when the evidence given on other counts of larceny did not connect the accused with the other thefts. (at p589)

12. It is for a reason similar to those which led to the decisions in those two cases that I consider that the evidence in relation to the deaths of Duncan and Montgomerie was inadmissible. It was not established that Duncan had ingested any poison (except of course the barbiturates, with whose consumption the applicant was not shown to have had anything to do). It was necessary for the Crown to rely on the other instances of poisoning, including those the subject of the present charges, in an attempt to provide a basis for an inference that Duncan had suffered from arsenical poisoning. In other words, it was necessary to assume the guilt of the applicant of the offences of which she was charged in order to render admissible the evidence regarding the death of Duncan. Such a line of reasoning is obviously objectionable. (at p590)

13. The case of Montgomerie creates more difficulty. It is clear that Montgomerie died by arsenical poisoning and that the applicant had had an opportunity to administer arsenic to him. However, Montgomerie was not living with the applicant, she was not preparing his meals and many other persons had an equal opportunity to put the arsenic in his bottle of wine, assuming that he did not do so himself. The applicant did not in any way stand to benefit from the death, and she had no adequate motive to poison her brother; others, it appears, may have had stronger motives than she. The fact that she gave a false name (her name when she was previously married) may be explained by a desire to escape from publicity since the death of Constable Haag had occurred only about a year before, but in any case it is only a circumstance of suspicion. The evidence linking the applicant with the death of Montgomerie does no more than raise a suspicion that she poisoned him, and viewed alone it would hardly even raise a suspicion. There is no striking similarity between the circumstances of the death of Montgomerie and those of the poisoning of Mr. Perry. On the whole it seems to me that the evidence relating to Montgomerie is not sufficiently cogent to be admitted. (at p590)

14. The evidence relating to the death of Constable Haag was, however, in my opinion admissible. There was proved a clear connexion between the applicant and the death of Constable Haag by arsenical poisoning and she had in that case, as in the case of Mr. Perry, a motive to kill him. There is indeed a striking similarity between the two cases. It is true that a considerable time had elapsed between the death of Constable Haag in 1961 and the poisoning of Mr. Perry in 1978 and that if the evidence of the other two poisonings is rejected there is only one other instance of poisoning besides those charged. It is unnecessary to consider whether two similar facts only (including that the subject of the charge) are enough to constitute a system, since evidence of this kind is not properly described as evidence of a system. There have been a number of cases (including Reg. v. Fletcher (1953) 53 SR (NSW) 70 ) in which evidence of one similar act of poisoning has been held admissible. On principle I cannot see why the evidence regarding the death of Constable Haag should not be admissible as a matter of law, although it would be a question for the trial judge whether he should exclude it in the exercise of his discretion. (at p590)

15. Notwithstanding the clear and careful summing up by the learned trial judge and the usful examination of the questions in the Full Court, I have reached the conclusion, in this difficult case, that the evidence relating to the deaths of Montgomerie and Duncan was not admissible and should not be given on a new trial. (at p591)

MURPHY J. Mrs. Perry was convicted on two accounts of attempted murder of her husband by arsenic poisoning. Mr. Perry had been poisoned by arsenic but this could have been accidental or due to deliberate poisoning by Mrs. Perry, Mr. Perry or some other person. There was no evidence directly implicating Mrs. Perry and no suggestion that she poisoned him unintentionally. The prosecution conceded that there was no basis for the verdicts but for inferences which it claimed should be drawn from three earlier events. (at p591)

2. 1. In March 1961 Constable Haag, an earlier husband of Mrs. Perry, died from arsenic poisoning. (at p591)

3. 2. In April 1962 Mr. Montgomerie, a brother of Mrs. Perry, died from arsenic poisoning. (at p591)

4. 3. In March 1970 Mr. Duncan with whom she had lived for some years after the death of her brother, died from an overdose of barbiturates, self-administered, but the prosecution claimed that his history of illness showed in the light of the two previous events and the poisoning of Mr. Perry that Duncan also had suffered from arsenic poisoning during the time he lived with the accused. (at p591)

5. The prosecution conceded that if each of the alleged poisonings was considered in isolation, the evidence would not justify a finding that Mrs. Perry poisoned any of them. It relied upon the reasoning in cases such as Makin v. Attorney-General (N.S.W.) (1893) 14 LR (NSW) 1; (1894) AC 57 and Martin v. Osborne (1936) 55 CLR 367 that the cumulative effect of the evidence relating to the four, justified the verdict; that the only rational explanation for the deaths of Haag and Montgomerie, the sickness of Duncan and the poisoning of Perry is that the applicant deliberately poisoned all of them with arsenic; that to hold otherwise is "an affront to common sense". (at p591)

6. In general, evidence of previous misconduct is not admissible to prove the charge in issue. There are exceptions. Evidence of previous misconduct is available to rebut a "defence that the accused unintentionally committed the act charged". That is strictly not a defence because the intent is to be established by the prosecution. As there was no suggestion of that in this case, the prosecution could not properly introduce the evidence to rebut such a defence (see Lord Sumner in Thompson v. The King (1918) AC 221 ). (at p592)

7. The supposed rigid division between prohibited use of previous criminality to show propensity and its use to establish a chain of coincidences so remarkable that it excludes the accused's innocence as "an affront to common sense", is unsatisfactory. The Australian Law Reform Commission points out:
"Traditionally, the explanation of the admissibility of 'propensity' evidence has almost invariably been in terms of the particular issue to which the evidence was relevant; for example, it has been variously admitted to show mens rea, to prove identity, to show that the accused's purpose was not innocent. The inadequacy of this sort of justification for admissibility is that it disregards the basic point that disposition evidence is objectionable not for what it proves but for the way in which it proves it. At other times similar fact evidence has been admitted on a different type of pretext, to show 'system', but this is little more than a specific and consistently practised propensity. The exact basis of admissibility of this evidence remains far from clear . . . " (Evidence Reference Research Paper No. 11, Character and Conduct (1982), p. 47).
The "principles" put forward as providing a basis upon which admission of similar fact (or "propensity") evidence can be admitted have been subjected to harsh criticism. Piragoff dissects the so-called criteria for admission or exclusion of similar fact evidence as stated by Lord Herschell in Makin v. Attorney-General (N.S.W.), which have been used in many cases to support the admission of such evidence. (See Similar Fact Evidence (1981).) Lord Herschell stated (1894) AC, at p 65 :

"It is undoubtedly not competent for the prosecution to adduce evidence tending to shew that the accused has been guilty of criminal acts other than those covered by the indictment, for the purpose of leading to the conclusion that the accused is a person likely from his criminal conduct or character to have committed the offence for which he is being tried. On the other hand, the mere fact that the evidence adduced tends to shew the commission of other crimes does not render it inadmissible if it be relevant to an issue before the jury, and it may be so relevant if it bears upon the question whether the acts alleged to constitute the crime charged in the indictment were designed or accidental, or to rebut a defence which would otherwise be open to the accused."

As Piragoff says, read together, the criteria for excluding and admitting the evidence are contradictory. See also Waight and Williams, Cases and Materials on Evidence (1981), Ch. 9, pp. 392 and following.) If, despite the exclusionary rule contained in the first sentence, evidence of propensity can be admitted where it is relevant in accordance with the second sentence, the "relevancy" standard is so broad as to render the statement meaningless. (at p593)

8. Despite protestations to the contrary, similar fact evidence has been admitted "for the purpose of leading to the conclusion that the accused is a person likely from his criminal conduct or character to have committed the offence for which he is being tried". (See Piragoff, p. 14 and Ch. 3; Lord Cross in Director of Public Prosecutions v. Boardman (1975) AC 421, at p 456 ; Hoffman, "Similar Facts After Boardman", Law Quarterly Review, vol. 91 (1975), p. 193; Sklar, "Similar Fact Evidence - Catchwords and Cartwheels", McGill Law Journal, vol. 23 (1977), p. 60.) In the present case, the prosecution alleges that the accused is an arsenic poisoner, that she has poisoned three other persons with arsenic, and that in these circumstances the arsenic poisoning of her husband is explicable rationally only by her poisoning him. Although if his poisoning is considered alone there is an obvious explanation consistent with her innocence - the abundance of arsenic in his work environment and his evidence that he unwittingly exposed himself to arsenic in the course of his work, the prosecution claim is that his evidence should be disbelieved, and that even though the poisoning could have occurred without her participation, the earlier events make this an affront to common sense. (at p593)

9. There is no universal formula for proof of guilt by circumstantial evidence. The process of reasoning relied upon by the prosecution, that is proof of guilt by association with circumstances, is theoretically acceptable. If there is a sufficient accumulation of events which, according to human experience, would not occur unless the accused were guilty, the tribunal is entitled to act on this material in arriving at a guilty verdict even though each event standing alone would not itself justify an adverse finding. The other events need not be strikingly similar, but no doubt the conclusion that their occurrence is inconsistent with the accused's innocence will more usually be reached when they are. However, this is an extremely dangerous method of determining criminal guilt. For centuries it was regularly used in England, other parts of Europe and the American colonies to convict millions of persons of the impossible crime of witchcraft. (at p593)

10. The presumption of innocence. One danger is that the presumption of innocence tends to be brushed aside. In the criminal justice system every person is taken to be innocent unless the contrary is legally proved. No one should be found guilty on appearances, suspicions, conjecture or anything but evidence establishing guilt beyond a reasonable doubt. (at p594)

11. In Mrs. Perry's case there is a very great temptation in weighing the evidence and more particularly in deciding admissibility, to ignore the presumption of innocence and to replace it with a presumption of guilt. The allegation that a number of the accused's relatives died or suffered from arsenic poisoning immediately conjures up a highly suspicious prejudicial atmosphere in which the presumption of innocence tends to be replaced with a presumption of guilt. The presumption of innocence and the strict standard of proof required in criminal cases tend to be indirectly and subtly undermined from the outset by reference to a sequence of events which according to common human experience would not occur unless the accused were guilty. By a backdoor this tends to reintroduce the standard rejected in Thomas v. The Queen (1960) 102 CLR 584 that guilt should be judged "in an ordinary common sense manner and in the way you would consider the more serious matters which come up for consideration and decision in your lives". It is very easy to assume that in common experience a person is hardly ever associated with poisonings of four close relatives, and that if such an association occurs it is so remarkable that it is unlikely to be innocent. Common assumptions about improbability of sequences are often wrong. A suggested sequence, series or pattern of events is often incorrectly regarded as so extremely improbable as to be incredible. However highly improbable, as well as merely improbable, sequences and combinations are constantly occurring. In random tossing the occurrence of a run of ten consecutive heads or tails is generally regarded as highly improbable. But this will occur on the average once in every 512 tosses, and the lesser sequences more frequently (2 runs of 9; 4 runs of 8; 8 runs of 7). If one randomly tosses a coin 257 times, more likely than not there will be a sequence of ten heads or tails. Although it is extremely improbable that any particular ticket will win a large lottery, it is certain that one will. (at p594)

12. Circular reasoning. There is a danger of admitting and using evidence of collateral circumstances (the death of Haag, or of Montgomerie, or the sickness of Duncan) which not only standing alone, but taken together with the other evidence (including any which independently tends to establish the central issue), is insufficient to exclude a rational explanation of that circumstance consistent with the accused's innocence, and then using the supposed guilt in relation to that circumstance as proof of others and the central issue. For example, on the prosecution's theory the supposed poisoning of Duncan depended on acceptance that Haag, Montgomerie and Perry were poisoned by the applicant, but the proof of these poisonings depended on the acceptance that she poisoned Duncan. Legal proof is not limited to proof by syllogisms nor to other methods which depend upon precision. The solution of most legal problems are those of approximate reasoning, which deals with indeterminate or "fuzzy" factors. For this generation, even the simplest of mathematical formulae are unsuitable tools for use in ordinary trials. But some practical working rules can be adopted for a criminal case like the present. Evidence of any circumstance such as any of the alleged poisonings of Duncan, Haag or Montgomerie should be discarded when it appears on consideration of the whole of the evidence that there is reasonable doubt about the accused's culpability in relation to that circumstance. Otherwise the stage is being set for a miscarriage of justice. (at p595)

13. Improper prejudice. In the method adopted here there is an extreme danger of improper prejudice to the accused. Usually an accused has to answer (at least on very serious charges) in relation to one alleged event, but here the accused was "put to answer" for a good part of her life. In effect she was tried on charges of murder of Haag and Montgomerie, and of attempted murder of Duncan as well as the two counts of attempted murder of Mr. Perry. Many of the allegations related to events years ago and extended in the case of Mr. Duncan over some years. This would put any accused in considerable difficulty. (at p595)

14. Mr.Duncan's sickness and death. The evidence was not fit to be taken into consideration. The immediate cause of Duncan's death was barbiturate poisoning while he was in severe depression. He held himself responsible for the death of his best friend in a car accident; lost his job through alcoholism; was refused a war pension and was bankrupt. The prosecution ultimately conceded that Duncan killed himself while depressed, but argued "what matters most is the evidence that the accused was slowly poisoning him with arsenic". In his summing up, the trial judge said "Plainly no-one thought of lead or arsenic as even a possibility until this case began." (at p595)

15. Mr. Duncan had a long history of alcoholic poisoning and bouts of depression which continued during his de facto marriage with Mrs. Perry (from about the mid nineteen sixties to his death in 1970). Before becoming acquainted with Mrs. Perry, Duncan's health was poor. Evidence was given by several doctors, confirmed by Repatriation files, that his "chronic ill health", involving severe bouts of diahorrea, sometimes accompanied by vomiting, extended back to World War II. He had operations for haemorrhoids in 1944 and 1957. He suffered from incontinence at times. Surgical explorations or operations were carried out on his haemorrhoids between 1968 and 1970. A medical report in 1970 found he was suffering from a personality disorder of an hysterical type, and that the physical symptoms were "probably a manifestation of that disorder". With depression came bouts of weeping, which he told at least one doctor had been a habit of his own father. Although psychiatric treatment was advised, Duncan refused it. The prosecution tendered "expert" evidence that his illness during his time with Mrs. Perry was explicable by arsenic poisoning. (at p596)

16. In the process of reasoning adopted in this case, the questions of weight and admissibility of evidence, of collateral circumstances (such as Duncan's alleged arsenic poisoning) are not separate as they usually are. Evidence may be admitted provisionally irrespective of weight. Ultimately however admissibility here depends on weight, because the evidence is available for consideration and therefore admissible only if, when taken with the other evidence, it could justify a finding that the accused poisoned Duncan. The evidence being circumstantial, it is admissible only if (in the light of all the evidence, not merely about Duncan but about the others including Perry) there is no rational explanation of Duncan's sickness consistent with the accused's innocence (that is of poisoning him). However there was powerful evidence in the shape of the transaminase test, as well as that of an expert pathologist, which overwhelmingly discredited the notion of arsenic poisoning. The dangers in the method of reasoning employed are illustrated thus: suppose the evidence were not available of the transaminase test, or of Mr. Duncan's medical history prior to the accused living with him; undoubtedly the weight and therefore the claim for admissibility of the evidence of his sickness would have been strengthened. (at p596)

17. It is disturbing that, especially when the accused was suffering inevitable prejudice from the multiplicity of allegations of poisoning, the prosecution persisted at the trial in relying upon Duncan's sickness as showing that Mrs. Perry poisoned him with arsenic. The Crown Solicitor, Mr. Prior Q.C., who appeared on this application, did not prosecute at the trial. (at p596)

18. The evidence about Mr. Duncan was left to the jury on the basis that if they accepted the prosecution's medical evidence, it was strikingly probative of the accused's guilt. In this Court, the prosecution conceded that standing alone, the evidence about Duncan was incapable of the inference that he at any time suffered from arsenic poisoning. The possibilities that she poisoned Haag, Montgomerie and Perry were not capable of justifying a finding that she poisoned Duncan. The evidence about Duncan was inadmissible. It follows that the conviction must be set aside. (at p597)

19. The prosecution's theory of guilt was that the only explanation of the events was that the accused poisoned all (including Duncan). It does not follow that the resolution is "guilty of all, or guilty of none". In theory a verdict of guilty would be justified if there were no other rational explanation of the two deaths (Haag and Montgomerie) and the poisoning of Mr. Perry than Mrs. Perry's guilt. Therefore it is appropriate to consider those deaths. (at p597)

20. Mr. Montgomerie's death. Frank Montgomerie was aged 39 at his death. He had a long history of illness (beginning in childhood) and chronic alcoholism, which led to his discharge from the army after World War II on medical grounds. He spent time in military and civilian mental hospitals, suffering delirium, depression and outbursts of physical violence. His alcoholism included episodes such as being found drunk and naked in the daytime at a public beach in the late 1950's, and returning home on many occasions with his arm in a sling, his hand wrapped up, or a patch over one eye. His nerves were bad and at any upset he "went to pieces". When his first wife left him, he attacked and shot her mother in the hand, the bullet lodging in his brother-in-law's skull. For this, he was convicted and sentenced to three years' imprisonment. During one spell of two to three weeks in a Sydney hospital when he was living with his first wife and children, away from the rest of his family which was based in Melbourne, he complained to his wife of severe stomach pains: the lining of his stomach was badly burned from ant poison, apparently deliberately taken by him. In early 1950 after sexually molesting his six-year-old daughter, Mr. Montgomerie spent several months in a lock-up ward in a mental hospital. (at p597)

21. Shortly before his death he was in financial difficulties. His television and radio repair business had collapsed; he was insolvent; he had borrowed a considerable sum of money from his brother. His de facto spouse told him she was leaving. He became depressed, then distraught, attempted to strangle her, and attacked her son. He had often said "If you leave me, I'll die. I can't live without you." It is an appalling history of physical and mental ruin, of violence toward others and of self destructive behaviour. His body was found lying beside a note signed by another of his sisters, Jean, saying "Frank, Marj has left. She said she couldn't take anymore. If you want anything, ring me." There was an almost empty wine bottle beside him with some arsenic compound in it. His death was attributed to arsenic poisoning. The amount of arsenic in his body was consistent with the minimum lethal dose which was sufficient because of his debilitation. (at p598)

22. As an amateur photographer who developed his own photographs he was familiar with and had easy access to chemicals. An inquest was held, to which Mrs. Perry was never called as a witness and was never approached as a possible witness. In his summing up at Mrs. Perry's trial for attempted murder of Mr. Perry, the judge stated "There is no evidence that the accused (Mrs. Perry) stood to gain financially from her brother's death, and no evidence of any ill feeling between them." It is unnecessary to detail the circumstances of Mr. Montgomerie's death, except to state that there was not a scrap of evidence to sustain a conclusion that the accused poisoned him. The evidence (and this includes the evidence about Haag and Perry as well as Montgomerie) is not merely incapable of excluding the hypothesis of innocence of Montgomerie's death; it is far more likely than not that the accused is innocent. The evidence presents a likelihood that Mr. Montgomerie died by self-poisoning either by suicide, or accidentally in an attempt to gain sympathy to cause the return of his de facto wife. (at p598)

23. Constable Haag. He died of acute arsenical poisoning. There was ample evidence providing a rational explanation of this consistent with Mrs. Perry's innocence. He used arsenic weed killer in gardening on the week-end of his death. There was some tenuous evidence suggesting earlier arsenic poisoning, but this was explicable as resulting from many other causes than poisoning. The trial judge observed "It is clear that the arsenic concentrations in (Haag's) hair were low, indeed they were within normal limits for random members of the public." (at p598)

24. After his death there was a thorough police investigation of all the circumstances, including extensive interrogation of the accused. An inquest was held, resulting in a finding of accidental death. It was not then suggested that Mrs. Perry was responsible for the death of Constable Haag. (at p598)

25. Again, this evidence is admissible only if, taken together with the other evidence, it could justify a finding that the accused poisoned him. The evidence being circumstantial, it must be capable of excluding any rational hypothesis other than guilt. The evidence was not in this category. The evidence against Mrs. Perry considered in the light of the other evidence against her could not establish guilt beyond reasonable doubt of poisoning Haag; it does not "affront common sense" to conclude that she is innocent of poisoning him. Rather, the evidence (including all the other admissible evidence in the case) easily admits of the rational hypothesis that Haag died by accident (or possibly by suicide) in which the accused was not inculpated. (at p599)

26. The exclusion of the evidence relating to the other alleged poisonings destroys the basis upon which the prosecution sought to establish the guilt of the poisonings actually charged. (at p599)

27. The case is not in the same universe of discourse as R. v. Smith (1915) 11 Cr App R 229 ("the brides in the bath case") or other similar fact cases. In the brides in the bath case the facts disclosed a clear pattern of involvement, on the part of the accused, with the very similar deaths, in rapid succession, of three women whom he had not long before married, each of whom had made a will in his favour. As Lord Maugham has concluded, "No reasonable man could believe it possible that Smith had successively married three women, persuaded them to make wills in his favour, bought three suitable baths, placed them in rooms which could not be locked, taken each wife to a doctor and suggested to him that she suffered from epileptic fits, and had then been so unlucky that each of the three had had some kind of fit in the bath and been drowned". (See Glanville Williams, The Proof of Guilt, 3rd ed. (1963), p. 230.) (at p599)

28. Mrs. Perry had association with many family members. Three died in unusual circumstances of poisoning. Other members of Mrs. Perry's family had association with these three; some had a far more intimate relationship with one (Mr. Montgomerie) than she. In two cases there was an insurance policy in Mrs. Perry's favour - however in the third there was not. There was an insurance policy in the case of Mr. Perry. In the deaths of Haag, Montgomerie and Duncan there was a history of chronic alcoholism and other severe ill-health, which could not be attributed to Mrs. Perry. With Mr. Perry there was no evidence of alcoholism, but there was a history of motor bike accidents, including severe injury to Mr. Perry's facial structure and nasal passages, which led to symptoms such as rhinitis. The prosecution's expert witness attributed this to arsenic or lead poisoning - although the condition had existed years before Mr. Perry made Mrs. Perry's acquaintance and had been the subject of a published medical article on facial reconstruction. (at p599)

29. Forensic Evidence. The evidence particularly in relation to Duncan, but also of the other alleged poisonings including that of Mr. Perry, reveals an appalling departure from acceptable standards of forensic science in the investigation of this case and in the evidence presented on behalf of the prosecution. As the Director of the Australian Institute of Criminology has said:

"The modern forensic scientist is not always happy with his apportioned role as an adjunct to the police forces. He too has been affected by the modern emphasis on the balance which needs to be maintained between law enforcement and human rights. More than that, the forensic scientist as a scientist is jealous of the independence of his discipline. He prefers to serve the court independently and not any system of law enforcement. He wants his services to be truly forensic or juridical and less partisan." (W. Clifford, The Relationship Between Criminology and the Forensic Sciences, p. 5).

The prosecution's evidence fell far short of this ideal. If the expert assistance available to the prosecution in this case is typical, then the interests of justice demand an improvement in investigation and interpretation of data and presentation to the court by witnesses who are substantially and not merely nominally experts in the subject which calls for expertise. (at p600)

30. Standard of proof of criminality other than that charged. In a case such as this, evidence of any other alleged poisoning must be excluded unless taken with all the other evidence it could justify a finding beyond reasonable doubt that the accused was guilty in relation to that poisoning. The trial judge recognised this in directing that none of the evidence in relation to Haag, Montgomerie or Duncan was to be used unless the jury were satisfied with respect to these other men that it proved that they were poisoned and that the accused poisoned them. Even if the evidence were admissible, provided that the evidence as a whole is substantially probative of guilt in relation to that poisoning, then the degree of probability of guilt in relation to the other alleged poisonings is not sufficiently high to exclude, as irrational or an affront to common sense, the hypothesis of innocence on the charges (the other evidence of which it is conceded could not justify conviction). (at p600)

31. All of the evidence about Haag, Montgomerie and Duncan should have been excluded; its consideration by the jury in determining whether the applicant attempted to murder her husband was a miscarriage of justice. For these reasons, I agreed with the order of the Court granting special leave to appeal, allowing the appeal against conviction in each case and ordering a new trial. An order for a new trial leaves it open to the prosecution to proceed if they think there is sufficient admissible evidence; it is not a direction by this Court to proceed to a new trial. The Court was not asked to enter judgment of acquittal, but if it had been and if a complete review of the evidence had disclosed no more against the applicant than was revealed on the appeal, judgment of acquittal should have been entered. (at p600)

WILSON J. The applicant was found guilty by a jury in the Supreme Court of South Australia on an indictment charging her with two offences of attempting to murder her husband. Her appeal to the Court of Criminal Appeal was dismissed. She then sought special leave to appeal to this Court. After hearing full argument, the Court ordered that special leave be granted, and the appeal allowed. The conviction was set aside and a new trial was ordered. The Court intimated that its reasons would be delivered at a later date. (at p601)

2. Two distinct arguments were advanced in support of the appeal. The first and principal ground was the wrongful admission of evidence of other alleged poisonings with which the applicant was said to be connected. The second matter alleged a miscarriage of justice arising from the failure of the Crown Prosecutor to cross-examine Mr. Perry, a witness called for the defence, in respect of testimony given by that witness which the jury was thereafter invited to disbelieve: cf. Browne v. Dunn (1893) 6 R 67 . (at p601)

3. This latter matter may be disposed of summarily. Mitchell J., with whom King C.J. and White J. agreed, while accepting that the Crown Prosecutor should have examined Mr. Perry on the issues in question, was nevertheless of the opinion that in all the circumstances no miscarriage of justice had occurred. There is no serious contest either as to the existence of the Browne v. Dunn principle or its application to the present case. The matter does not call for any further attention by this Court. (at p601)

4. The substantial issue is the alleged wrongful admission of so-called similar fact evidence. (at p601)

5. During 1978 and 1979 Mr. Perry, the applicant's husband, suffered from chronic arsenic and lead poisoning. He was very seriously ill in November 1978 and again in September 1979, and a charge of attempted murder was based on each of these episodes. The Crown case alleged that Mrs. Perry was systematically poisoning her husband by causing him to ingest arsenic, and the allegation was supported by evidence both of opportunity and motive. The defence consisted of an outright denial of the charges and evidence was led suggesting the accidental ingestion of the poison. (at p601)

6. In the course of the Crown case the learned trial judge admitted evidence of three earlier happenings with which the applicant had some connexion. The first concerned the illness and death of a former husband, Constable Haag. At the time of his death they were living in Melbourne. He became ill in December 1960 and died in March 1961. He was 35 years of age. He died of acute arsenical poisoning. There were insurance policies by reason of which the applicant stood to benefit financially from his death. During his marriage Constable Haag had completed a police course in poisons, and his wife had assisted him in that course. There were other matters revealed in the evidence touching the death of Constable Haag which if accepted by the jury would have supported a conclusion that the applicant was implicated in his demise. (at p602)

7. The second happening occurred in April 1962 when one Montgomerie, the brother of the applicant, died from acute arsenic poisoning. At the time of his death he was in a very poor state of health and had just been abandoned by his de facto wife. He was a chronic alcoholic. Beside the bed on which he was found dead was a wine bottle containing a small quantity of wine and some arsenic. There was some evidence of opportunity in that the applicant may have had access to arsenic and was the last person to see him alive and the first to find him dead. There was no evidence of any motive in this case, although there is the suggestion, mentioned along with others in the summing up of the learned trial judge, that she may have wished to rid the family of an embarrassment and a burden. (at p602)

8. The third episode relied on by the Crown occurred in the late sixties when a man named Duncan was living with the applicant in her home in a de facto relationship. Cohabitation commenced towards the end of 1967 and they lived together until Duncan died in March 1970 from an overdose of sleeping tablets. The post-mortem examination revealed that he had consumed approximately twenty barbiturate capsules. Although the question whether his death was due to suicide, accident or murder was agitated during the trial, the real thrust of the Crown case with regard to Duncan was that the applicant administered arsenic to him intermittently during the two years preceding his death. There was a serious deterioration in his state of health during this period. He suffered from diarrhoea, incontinence, abdominal pain, swollen breasts and vomiting accompanied by a general debilitation. The medical picture is confused by evidence of excessive drinking and a long-standing condition of haemorrhoids. Despite intensive medical and hospital attention, no satisfactory diagnosis of his condition and its cause was made. The Crown relied on the opinion of a pathologist, Dr. Mantock, to the effect that neither alcoholism nor the haemorrhoids provided a complete explanation of the medical picture, whereas lead arsenate poisoning did. Dr. Mantock conducted a post-mortem examination of the body of the deceased, and did not detect any sign of arsenic poisoning. Thereafter the cremation of the body precluded any subsequent examination. An expert witness called for the defence expressed the opinion that lead arsenate poisoning was a possible but unlikely cause, but of course this evidence is material to the weight of the impugned evidence rather than to its admissibility. The applicant stood to gain financially from Duncan's death, by reason of insurance policies which were taken out in the middle of 1968. This was sometime after the deterioration of his health commenced. (at p603)

9. The principles on which evidence of similar facts is admissible in criminal cases were expounded by Gibbs A.C.J. (as he then was), with the concurrence of Stephen, Jacobs and Aickin JJ., in Markby v. The Queen (1978) 140 CLR 108, at pp 116-117 . After citing the authoritative statement of Lord Herschell L.C. in Makin v. Attorney-General (N.S.W.) (1894) AC 57, at p 65 and making some observations based on Harris v. Director of Public Prosecutions (1952) AC 694 , his Honour continued (1978) 140 CLR, at p 117 :
"It is often difficult to decide whether a particular piece of evidence is or is not admissible within these principles. However when in doubt a judge should remember that the admission of similar fact evidence is the exception rather than the rule. To be admissible the evidence must have 'a strong degree of probative force' (per Lord Wilberforce in Reg. v. Boardman (1975) AC at p 444 ), or 'a really material bearing on the issues to be decided' (per Lord Morris of Borth-y-Gest (1975) AC, at p 439 , citing Harris v. Director of Public Prosecutions (1952) AC, at p 710 ), it may not be going too far to say that it will be admissible only if it is 'so very relevant that to exclude it would be an affront to common sense' (see per Lord Cross in Reg. v. Boardman (1975) AC, at p 456 ; and see per Lord Hailsham of St. Marylebone (1975) AC, at pp 452-453 ). The question is thus one of degree, and in answering it the judge must apply his experience and common sense. Although the judgment which the judge is required to make is to some extent discretionary, the rule of exclusion is a rule of law and not of discretion, and the principle allowing the admission of the evidence remains subject to the discretionary power to exclude it, even if legally admissible, where its prejudicial effect outweighs its probative value. In applying the test of admissibility to which I have just referred, practical assistance, in many cases, will be obtained by considering whether there is a 'striking similarity' between the similar facts and the facts in issue (see Reg. v. Boardman (1975) AC, at pp 439, 441, 442, 452, 454, 462 )." (at p603)

10. Recently, in the case of Reg. v. Chee (1980) VR 303 , the Full Court of the Supreme Court of Victoria had occasion to consider the test of admissibility in relation to similar fact evidence. After referring to Boardman and Markby, their Honours said (1980) VR, at p 308 :

"As at present advised, we are of opinion that it is not a condition of the admissibility of similar fact evidence, either in general or in cases where the evidence is tendered in relation to a particular kind of issue or issues, that the evidence shall have, not merely probative force . . ., but a high degree of probative force. It is difficult to see why, in principle, a stronger degree of probative force should be necessary for the reception of evidence of similar facts than for the reception of other circumstantial evidence."

With great respect to their Honours, and notwithstanding the impressive list of earlier cases which are cited in support of their viewpoint, it seems to me that Boardman, coupled with the approval accorded to it by this Court in Markby, has determined the issue against them. Such evidence differs from other circumstantial evidence because, notwithstanding that it may have some relevance, its admission is the exception rather than the rule. The view has been expressed that the effect of Boardman is to interpret the classic statement of Lord Herschell in Makin not in terms of a rule of exclusion with certain categories wherein exceptions may be found, nor in terms of mere propensity or something more, but as requiring the court to compare, in the context of each case, the probative value of the evidence with the risk of prejudice it conveys. To be admissible, the evidence must have a sufficiently high degree of probative force to outweigh the inevitable prejudice flowing from the nature of the evidence. It is a question of degree in each case, since probative value will vary with the nature of the similar fact evidence itself, the issues, and the other evidence in the case. There is an impressive body of academic writing which takes this view of Boardman, including L. H. Hoffmann, "Similar Facts after Boardman", Law Quarterly Review, vol. 91 (1975), p. 193; Ronald B. Sklar, "Similar Fact Evidence - Catchwords and Cartwheels", McGill Law Journal, vol. 23 (1977), p. 60; C.R. Williams, "The Problem of Similar Fact Evidence", Dalhousie Law Jo., vol. 5 (1979), p. 281; Waight and Williams, Cases and Materials on Evidence (1980), p. 392; Eggleston, Evidence, Proof and Probability (1978), p. 73, Cross on Evidence, 5th ed. (1979), pp. 375- 378. In my opinion, there is much to be said for this reconciliation of the twin principles stated by Lord Herschell, although there is no reason to treat any particular formulation as exclusive. In some cases, the categorization of exceptions will continue to be helpful. (at p604)

11. The exclusion of the evidence relating to Duncan reduces the number of proved occasions when members of Mrs. Perry's family had ingested arsenic from four to three, and the probative force of the circumstantial evidence is thereby reduced. Moreover, the probative force of the evidence relating to Montgomerie, standing alone, is much less than the force of the evidence relating to Haag. But it would be a mistake to consider the evidence relating to Montgomerie and the evidence relating to Haag as separate pieces of evidence which much be admitted or rejected in isolation from each other. Such probative force as they have depends largely upon the factor that in both cases death was caused by arsenic and that in those cases and in the case of Mr. Perry, the jury would be entitled to find that Mrs. Perry had an opportunity of administering the arsenic ingested. It is the occurrence of three arsenical poisonings in Mrs. Perry's family which gives probative force to the challenged evidence. Its cogency and weight will be a matter for the jury at the end of the trial, but in my view the evidence relating to Haag and Montgomerie is of sufficient force to warrant its admission. The wrongful admission of the evidence relating to Duncan is the sole ground which I would assign for the order made by this Court granting special leave to appeal, allowing the appeal against conviction in each case and ordering a new trial. (at p613)

12. The respective reasons for judgment now published reveal that this Court is evenly divided as to the admissibility of the evidence relating to Montgomerie. (at p613)

13. In the light of this division, it would not be right for the Crown on a retrial to press for the admission of the evidence relating to Montgomerie. If that evidence is not admitted, the evidence relating to Haag, standing alone, is greatly weakened. As a new trial has been ordered, it will be a matter for the trial judge to consider, in the light of the judgments of this Court, whether the evidence relating to Haag should be admitted. That evidence is consistent with Haag ingesting arsenic accidentally, and the trial judge will be required to determine whether it should be excluded, as being more akin to the evidence excluded in Noor Mohamed v. The King (1949) AC 182 than to the evidence admitted in Reg. v. Fletcher (1953) 53 SR (NSW) 70 . (at p613)

14. Whether the evidence relating to Haag is admitted or excluded, it will be necessary for the trial judge also to determine whether the whole of the evidence is sufficient to support a verdict of guilty or whether the jury should be directed that it would be dangerous or unsafe to return a verdict of guilty (cf. Criminal Law Consolidation Act 1935-1981 (S.A.), s. 353(1); Ratten v. The Queen (1974) 131 CLR 510, at p 515 ). No doubt the Crown, before deciding to proceed with the new trial, will give consideration to the same questions. (at p613)

Orders

Application for special leave to appeal granted.

Appeal allowed.

Order that the judgment of the Court of Criminal Appeal be set aside and in lieu thereof order that the appeal to that Court be allowed, the convictions quashed and a new trial ordered.