Evidence - Criminal trial - Accomplice inculpating accused - Whether accomplice warning necessary - Nature of warning.

CANBERRA, 1993, December 8; BRISBANE, 1994, June 30

MASON CJ AND McHUGH J The facts and issues in this matter are set out in the judgment of Toohey J. Except to the extent necessary to explain our reasons, it is unnecessary to refer to them.

2. The learned trial judge held that he had a discretion to discharge the jury if he believed that "there was a real danger that the position of the accused had been or might have been prejudiced" by the conduct of the juror in arranging for flowers to be given to the deceased's mother ((1) Webb and Hay (1992) 64 A Crim R 38 at 70.). The learned judge relied on English authority in formulating the "real danger" test ((2) Sawyer (1980) 71 Cr App R 283 at 285; Reg. v. Spencer [1987] UKHL 2; (1987) AC 128 at 144.). However, we are of opinion that this was too stringent a test. In our opinion, the test that his Honour should have applied was whether, despite the warning that he proposed to give to the jury, the circumstances of the incident would still give a fair-minded and informed observer a reasonable apprehension of a lack of impartiality on the part of the juror.

3. When it is alleged that a judge has been or might be actuated by bias, this Court has held that the proper test is whether fair-minded people might reasonably apprehend or suspect that the judge has prejudged or might prejudge the case ((3) Reg. v. Commonwealth Conciliation and Arbitration Commission; Ex parte Angliss Group (1969) 122 CLR 546 at 553-554; Reg. v. Watson; Ex parte Armstrong [1976] HCA 39; (1976) 136 CLR 248 at 261-262, 264, 267; Re Judge Leckie; Ex parte Felman (1977) 52 ALJR 155 at 158; 18 ALR 93 at 97-98; Re Shaw; Ex parte Shaw (1980) 55 ALJR 12 at 14, 16; 32 ALR 47 at 50-51, 54; Livesey v. New South Wales Bar Association [1983] HCA 17; (1983) 151 CLR 288 at 293-294, 300; Re J.R.L.; Ex parte C.J.L. [1986] HCA 39; (1986) 161 CLR 342 at 349, 351, 359, 368 and 371; Vakauta v. Kelly [1989] HCA 44; (1989) 167 CLR 568 at 575, 584; Grassby v. The Queen [1989] HCA 45; (1989) 168 CLR 1 at 20.). The Court has applied the same test to a Commissioner of the Australian Industrial Relations Commission ((4) Re Media, Entertainment and Arts Alliance; Ex parte Hoyts Corporation Pty. Ltd. [1993] HCA 41; (1994) 68 ALJR 179 at 182; [1994] HCA 66; 119 ALR 206 at 210.) and to a member of the Australian Broadcasting Tribunal ((5) Laws v. Australian Broadcasting Tribunal [1990] HCA 31; (1990) 170 CLR 70 at 87, 92, 102.). The Court has specifically rejected the real likelihood of bias test ((6) Watson (1976) 136 CLR at 261-262.). The principle behind the reasonable apprehension or suspicion test is that it is of "fundamental importance that justice should not only be done, but should manifestly and undoubtedly be seen to be done" ((7) R. v. Sussex Justices; Ex parte McCarthy (1924) 1 KB 256 at 259 per Lord Hewart CJ; Re J.R.L.; Ex parte C.J.L. (1986) 161 CLR at 351-352.). Although the role of the juror is not the same as that of the judge, a commissioner or a member of a quasi-judicial tribunal, we do not think that the difference between the role of the juror and the role of those persons warrants any different test for alleged bias.

4. Moreover, in determining whether the conduct of a juror gives rise to a fear of bias, Australian courts have frequently applied the reasonable suspicion test. For example, the Full Court of the Supreme Court of Victoria adopted the reasonable suspicion test when the question arose whether a conversation between a juror and an outsider after the commencement of jury deliberations in a criminal trial affected perceptions as to the impartiality of the verdict ((8) Reg. v. Hodgkinson [1954] VicLawRp 23; (1954) VLR 140 at 144; Reg. v. Chaouk [1986] VicRp 70; (1986) VR 707 at 712, 717.). Similarly in a civil case, the same Court said that, in a case of contact between a juror and a third party before deliberations, "no reasonable ground of suspicion should be allowed to arise as to the fairness of that tribunal" ((9) Trewartha v. Confidence Extended Co. NL [1906] VicLawRp 48; (1906) VLR 285 at 288.). In Duff v. The Queen ((10) [1979] FCA 83; (1979) 28 ALR 663.), the Full Court of the Federal Court applied the reasonable suspicion test in a criminal appeal in upholding a trial judge's decision not to discharge a jury after the judge's associate had spoken with a juror at a party. Their Honours said ((11) ibid. at 681.):

party would not by itself give rise to a reasonable

suspicion as to the fairness of the trial. Nor could the

ensuing conversation give rise to a reasonable suspicion of

an improper extra-curial communication with a juror once

the circumstances and the terms of the communication were

disclosed. The disclosure was prompt, complete, satisfying

5. New Zealand courts also seem to apply the reasonable suspicion test where bias on the part of a juror is alleged. In Reg. v. Papadopoulos (No.2) ((12) (1979) 1 NZLR 629 at 634.), the New Zealand Court of Appeal applied the test whether there "is reasonable ground for suspecting that the verdict may have been influenced by bias on the part of the foreman towards the prosecution" in a case where the foreman of the jury worked in the same government department as two prosecution witnesses. The Court of Appeal applied the same test in Reg. v. McCallum and Woodhouse ((13) (1988) 3 CRNZ 376.) where the accused were charged with manufacturing morphine. The foreman of the jury probably knew of the drug dependency of the girlfriend of one of the accused. In the course of its judgment, the Court of Appeal referred ((14) ibid. at 379.) to one of its unreported judgments ((15) Reg. v. Sannd, 2 March 1988.) where the Court had expressed the test to be "whether there was a reasonable suspicion or a real danger that the accused's position had been prejudiced". In McCallum and Woodhouse, the Court said ((16) (1988) 3 CRNZ at 379.) that it did not read this statement as a different approach from that which had been expressed in Papadopoulos. In Reg. v. Te Pou ((17) (1992) 1 NZLR 522 at 527.), the Court of Appeal formulated the test as being whether there was "a reasonable suspicion or real danger of bias". However, the Court did so after referring to McCallum and Woodhouse without criticism. It seems likely, therefore, that the New Zealand Court of Appeal is still effectively applying the reasonable suspicion standard and not some higher standard in determining allegations of bias.

6. However, English courts have rejected the reasonable suspicion test. In Sawyer ((18) (1980) 71 Cr App R at 285-286.), the Court of Appeal said that the correct test was "whether there was any danger from anything done or said that the jury might have been prejudiced against the appellant". This test was approved by the House of Lords in Reg. v. Spencer ((19) [1987] UKHL 2; (1987) AC 128 at 144.). More recently, in Reg. v. Gough ((20) [1993] UKHL 1; (1993) AC 646.), Lord Goff of Chieveley, after examining the authorities in detail, reformulated the real danger test. He expressed bewilderment ((21) ibid. at 659.) at the different tests found in the authorities. Lord Goff rejected the need to distinguish between juries and judges, and formulated the test to be applied, where bias is alleged, as follows ((22) ibid. at 670.):

"(H)aving ascertained the relevant circumstances, the

court should ask itself whether, having regard to those

circumstances, there was a real danger of bias on the part

of the relevant member of the tribunal in question, in

the sense that he might unfairly regard (or have unfairly

regarded) with favour, or disfavour, the case of a party to

the issue under consideration".

Lord Goff felt that it was unnecessary "to have recourse to a test based on mere suspicion, or even reasonable suspicion" ((23) ibid. at 668.). He also thought that the concept of the reasonable person was inapplicable because the court acted as the reasonable person and inquired into the circumstances about which the reasonable "observer" in the courtroom would not necessarily have any knowledge ((24) ibid. at 670.). Lord Goff said that he had adopted the real danger test instead of the real likelihood test "to ensure that the court is thinking in terms of possibility rather than probability of bias" ((25) ibid.).

7. In Canada, the approach of the Courts to the question of juror bias has not been uniform. In Reg. v. Cameron ((26) (1991) 64 CCC (3d) 96 at 102.), the Ontario Court of Appeal said that bias could be found on one of two bases. First, where there has been conduct sufficient to "taint the administration of justice". Secondly, where "actual prejudice was occasioned to the accused" ((27) ibid.). In determining whether "the events in question are so serious as to affect the administration of justice", the Court said that "the focus turns upon the justice system and the miscarriage of justice occurs whenever the confidence of the public in the system is shaken" ((28) ibid.). In Reg. v. Lessard ((29) (1992) 74 CCC (3d) 552.), however, the Quebec Court of Appeal applied a different test when associates of the accused attempted to bribe a juror and the trial judge discharged the juror but continued the trial with the rest of the jury. Baudouin JA, giving the judgment of the Court, said ((30) ibid. at 563.):

"One must take the jurors for reasonable people and I am

convinced, in light of all of the circumstances and the

contents of the warnings, that (the trial judge) properly

exercised his discretion in considering that there was not,

in the present case, a 'real danger' of the distortion of

the system and therefore a possibility of negative influence

towards the accused."

8. In Reg. v. Horne ((31) (1987) 35 CCC (3d) 427.), the Alberta Court of Appeal applied another test. The trial judge had refused to order a new trial when a police officer, who was a Crown witness, had had a conversation with three jurors during an adjournment. The Court of Appeal said that the issue is "What is the real risk of a verdict being rendered which is influenced by unproven facts?" ((32) ibid. at 432.) The Court said ((33) ibid. at 434.) that cases of jury irregularity differ. If there was no real prejudice to the accused or the Crown, there was no inflexible rule that the jury should be discharged "in sole deference to the image of the proceeding". On the other hand, where "tainted evidence" had reached a deliberating jury, a real danger of prejudice had to be presumed. Where the extent of the prejudice is readily measurable, however, and the trial judge finds "that there is no real danger of the loss of a fair trial of the issues on the admissible evidence, the result differs" ((34) ibid.).

9. In considering the merits of the test to be applied in a case where a juror is alleged to be biased, it is important to keep in mind that the appearance as well as the fact of impartiality is necessary to retain confidence in the administration of justice. Both the parties to the case and the general public must be satisfied that justice has not only been done but that it has been seen to be done. Of the various tests used to determine an allegation of bias, the reasonable apprehension test of bias is by far the most appropriate for protecting the appearance of impartiality. The test of "reasonable likelihood" or "real danger" of bias tends to emphasise the court's view of the facts. In that context, the trial judge's acceptance of explanations becomes of primary importance. Those two tests tend to place inadequate emphasis on the public perception of the irregular incident.

10. We do not think that it is possible to reconcile the decision in Gough with the decisions of this Court. In Gough, the House of Lords specifically rejected the reasonable suspicion test and the cases and judgments which had applied it in favour of a modified version of the reasonable likelihood test. In Watson, faced with the same conflict in the cases between the two tests, this Court preferred the reasonable suspicion or apprehension test. That test has been applied in this Court on no less than eight subsequent occasions. In the light of the decisions of this Court which hold that the reasonable apprehension or suspicion test is the correct test for determining a case of alleged bias against a judge, it is not possible to use the "real danger" test as the general test for bias without rejecting the authority of those decisions.

11. Moreover, nothing in the two speeches in the House of Lords in Gough contains any new insight that makes us think that we should re-examine a principle and a line of cases to which this Court has consistently adhered for the last eighteen years. On the contrary, there is a strong reason why we should continue to prefer the reasoning in our own cases to that of the House of Lords. In Gough, the House of Lords rejected the need to take account of the public perception of an incident which raises an issue of bias except in the case of a pecuniary interest. Behind this reasoning is the assumption that public confidence in the administration of justice will be maintained because the public will accept the conclusions of the judge. But the premise on which the decisions in this Court are based is that public confidence in the administration of justice is more likely to be maintained if the Court adopts a test that reflects the reaction of the ordinary reasonable member of the public to the irregularity in question. References to the reasonable apprehension of the "lay observer" ((35) Vakauta (1989) 167 CLR at 573, 574.), the "fair-minded observer" ((36) Livesey (1983) 151 CLR at 300; Laws (1990) 170 CLR at 87.), the "fair-minded, informed lay observer" ((37) Laws (1990) 170 CLR at 92.), "fair-minded people" ((38) Watson (1976) 136 CLR at 263.), the "reasonable or fair-minded observer" ((39) Vakauta (1989) 167 CLR at 585.), the "parties or the public" ((40) Ex parte Hoyts Corporation Pty. Ltd. (1994) 68 ALJR at 182; 119 ALR at 210.), and the "reasonable person" ((41) Vakauta (1989) 167 CLR at 576.) abound in the decisions of this Court and other courts in this country. They indicate that it is the court's view of the public's view, not the court's own view, which is determinative. If public confidence in the administration of justice is to be maintained, the approach that is taken by fair-minded and informed members of the public cannot be ignored. Indeed, as Toohey J pointed out in Vakauta ((42) ibid. at 585.) in considering whether an allegation of bias on the part of a judge has been made out, the public perception of the judiciary is not advanced by attributing to a fair-minded member of the public a knowledge of the law and the judicial process which ordinary experience suggests is not the case. That does not mean that the trial judge's opinions and findings are irrelevant. The fair-minded and informed observer would place great weight on the judge's view of the facts. Indeed, in many cases the fair-minded observer would be bound to evaluate the incident in terms of the judge's findings.

12. A further reason for rejecting the Gough formulation is that, where the conduct of a juror is in issue, it will often be difficult to determine objectively whether the incident has affected or might affect the impartiality of the juror and whether directions to the jury were or will be adequate to protect the parties from the effect of the irregular incident. To place confidence in a test based on the assumption that an investigation will reveal all the facts of the incident may lead to a miscarriage of justice. In our experience, the investigation of such incidents during the course of the trial is not exhaustive. Ordinarily, the judge simply asks the juror for an explanation. However, a juror involved in an irregular incident may feel defensive about his or her role. Understandably, the juror may seek to put the best light on the matter. Seldom, if ever, is there a detailed cross-examination of the juror by counsel or by the judge in such a case. Indeed, many counsel would consider it unwise to cross-examine the juror while the possibility existed that the trial would continue with that juror. One can never be certain, therefore, whether all the circumstances have been elicited by the trial judge. If real danger of bias was the governing criterion, the judge might reach a conclusion opposite to that which he or she might have reached if all the facts were known. The reasonable apprehension test, on the other hand, allows a margin for error in evaluating the facts as elicited. It concentrates not on whether there is a danger of bias as an objective fact, but whether a fair-minded and informed person might apprehend or suspect that bias existed.

13. Furthermore, if the reasonable apprehension test remains the test for alleged bias on the part of a judge, as we think it should, it is not easy to see why a different test should be applied to a juror. In criminal trials in particular, the jury's function is of great public importance. It is certainly no less important than that of the judge sitting alone in a civil trial, a commissioner determining an industrial dispute or a member of a statutory tribunal inquiring into conduct in an industry which it supervises. The public is entitled to expect that issues tried by juries as well as judges and other public office holders should be decided by a tribunal free of prejudice and without bias. It is true that, unlike the judge and persons exercising quasi-judicial functions, the juror is subject to the directions of a third party - the trial judge. In considering whether a reasonable apprehension of bias exists, it is therefore necessary to consider the likely effect of the judge's directions (if any) as well as the irregularity in question. But that difference does not seem to us to be sufficient to distinguish the test for juror bias from the test for judges and persons who exercise quasi-judicial functions.

14. It follows that the test to be applied in this country for determining whether an irregular incident involving a juror warrants or warranted the discharge of the juror or, in some cases, the jury is whether the incident is such that, notwithstanding the proposed or actual warning of the trial judge, it gives rise to a reasonable apprehension or suspicion on the part of a fair-minded and informed member of the public that the juror or jury has not discharged or will not discharge its task impartially.

15. In the present case, the prosecutor made the trial judge aware of the incident. The learned trial judge asked the juror for an explanation of her conduct. She told the trial judge, "I didn't consider it of importance, I'm afraid I'm an impulsive person". He accepted that the contact between the juror and the mother of the deceased's fiancee was spontaneous. If his Honour had applied the correct test, it would be difficult to interfere with the exercise of his discretion to continue the trial with that juror being part of the jury. He had the opportunity to see the juror. A fair-minded person would give considerable weight to the judge's conclusion that the public ventilation of the incident - together with an appropriate warning - would nullify the inference otherwise to be drawn from the irregularity. Moreover, the decision of the judge is a discretionary judgment in the sense that it involves a value judgment. Where no error of principle is involved, an appellate court is naturally slow to substitute its opinion for the trial judge's opinion. The law reports contain many cases where the decision of the trial judge to continue a trial has been upheld, notwithstanding irregular incidents ((43) See R. v. Twiss (1918) 2 KB 853; Reg. v. Giles [1959] VicRp 76; (1959) VR 583; Reg. v. White (1969) SASR 491; Prime (1973) 57 Cr App R 632; Reg. v. Norton-Bennett (1990) 1 NZLR 559; Te Pou (1992) 1 NZLR 522.), including conversations or contact between a juror and outsiders before deliberations have commenced. However, the learned judge did not apply the correct test. Once he concluded that the action of the juror was a spontaneous, sympathetic gesture and that the jury was a diligent jury, his decision to continue the trial after applying the real danger test is unsurprising.

16. If the learned judge had applied the reasonable apprehension test, however, it is by no means certain that he would have reached the same conclusion concerning the conduct of the juror. In the course of his reasons for refusing to discharge the jury, the learned judge said ((44) Webb and Hay (1992) 64 A Crim R at 70.):

"An examination of the reported authorities on this

question suggests that something more than an impulsive act

of sympathy is required before a court will conclude that

there is some bias or misconduct which requires the

discharge either of that juror or all of the jury."

His Honour then referred to Bliss ((45) (1986) 84 Cr App R 1.), Sawyer, Spencer ((46) (1985) QB 771.) and Pennington ((47) (1985) 81 Cr App R 217.) as well as other English cases. None of the cases to which his Honour referred were factually similar to the present case and all of them applied a test which is contrary to the law of this country. Moreover, the language used by the learned trial judge in the above passage comes close to suggesting that the facts must prove actual bias or misconduct before a juror or jury will be discharged. Later on in his reasons the learned judge said that he did not think "that an expression of sympathy necessarily points to bias or prejudice or any other incapacity on the part of a juror to reach a decision with regard only to the evidence ... Even if there is a risk of bias or prejudice, I think that risk can, in this case, be met by an appropriate warning to the jury." ((48) Webb and Hay (1992) 64 A Crim R at 72.) Nothing in his Honour's judgment suggests that he gave any thought to the conclusion of a fair-minded and informed member of the public who was made aware of the incident.

17. The gesture of the juror may have been spontaneous, but a fair-minded person might fairly apprehend that it revealed a state of mind that was not compatible with the unemotional and impartial consideration of the case. One can accept the juror's own explanation of her gesture without derogating from the impact of that gesture on the minds of fair-minded people. Her conduct was not a reaction to evidence that she had just heard. It occurred after the conclusion of the evidence and the addresses of two of the three counsel and after the jury had been warned about communicating with persons associated with the trial. The incident indicated that the juror felt strongly for the plight of the mother. Her sympathy, manifested as it was by disobedience of the judge's warning, raised a serious question as to her ability to consider the evidence dispassionately and impartially.

18. An accused person and the public at large are always entitled to be concerned with the fairness of a criminal trial where a juror exhibits sympathy for a relative of the victim. But this case goes further. One or other or both accused killed the deceased. It was a savage, senseless and unprovoked murder. The judge described the photographs of the deceased taken when the body was discovered and during the post mortem examination as "quite unpleasant" ((49) ibid.). The murder was one, therefore, which was likely to excite feelings of revulsion against the person or persons responsible for a crime which defence counsel had conceded was "horrific". If the accused were to have a fair trial according to law, it was essential that the jurors perform their difficult task as unemotionally as possible. It was certainly open to the trial judge to find that the conduct of the juror gave rise to a reasonable apprehension of her inability to perform her task in a detached manner. Moreover, we think that his Honour should have concluded that the conduct of the juror did give rise to a reasonable apprehension of a lack of impartiality.

19. But a finding that the incident gave rise to a reasonable apprehension of bias is not the end of the matter. The fair-minded and informed observer would also consider the effect of the judge's warning on the juror and the judge's assessment of the character of the juror. We have already set out the passage where the learned judge said that he thought that, even if there was a risk of bias or prejudice, the case could be met with an appropriate warning. That was an opinion that a fair-minded person would not lightly reject. Further, the learned judge made findings concerning the juror and gave a warning which a fair-minded and informed person was bound to consider. His Honour found that the juror was a very attentive and diligent juror who had taken extensive notes of the evidence and addresses. He also found that she was very concerned about her conduct, conduct for which she had apologised. The warning which the learned judge gave was very detailed. Part of it was directed to the juror in question. He referred to the incident and said "that it might appear that in some way you had formed a view about the issues in this case and were displaying some kind (of), perhaps, bias towards the Crown case, or, alternatively the demonstration of your sympathy in that way might cloud your proper consideration of all of the evidence" ((50) ibid. at 73.). The judge concluded his warning to the jury by saying ((51) ibid. at 74.):

"Sift and weigh each of the witnesses, all of the witnesses

including Ms Hay without any feelings of emotion, any

feelings of sympathy. Just look at it, coldly,

dispassionately and above all, objectively and using your

common sense."

20. Although, by her own admission, the juror was an impulsive person and had disobeyed an earlier warning not to communicate with persons associated with the case, we think that a fair-minded and informed person would not apprehend bias on her part. While her sympathy for the deceased's mother had caused her to act as she did, that sympathy had not manifested itself in any act of hostility towards the accused or of partiality to the Crown. Its significance lay in the fact that it indicated that the juror was or might be incapable of examining the evidence dispassionately and impartially. But the public ventilation of the incident, the juror's apology, the recognition of the seriousness of what she had done, the general attentiveness and diligence of the juror, and the strength and detail of the judge's second warning were countervailing factors of considerable strength. A fair-minded person would assume that the juror would do her best to follow the judge's direction to look at the evidence "coldly, dispassionately and above all, objectively and using (her) common sense". When a fair-minded observer also considered the opinion of the judge - the person on the spot - that the juror would be able to approach the issues dispassionately, we think that a fair-minded person would not have an apprehension of bias or lack of impartiality on the part of the juror. It follows that this ground of appeal fails.

21. We agree with the reasons of Toohey J for concluding that the remaining grounds of appeal also fail.

BRENNAN J It is a valid ground of objection to the continued sitting of a judge or juror in a criminal trial that a fair-minded and informed member of the public would entertain a reasonable apprehension that the judge or juror will not discharge his or her duty impartially. In this respect I agree with Mason CJ and McHugh J. Provided no objection can be taken to a juror on this ground, an accused person, if he be convicted, can have no reasonable apprehension that his defence was not impartially considered by the jury. In other words the trial will have been as fair as the court can make it - that being the duty of the court ((52) Jago v. District Court (N.S.W.) [1989] HCA 46; (1989) 168 CLR 23 at 49; Dietrich v. The Queen [1992] HCA 57; (1992) 177 CLR 292 at 323.). Lord Devlin ((53) "Judges and Lawmakers", the Fourth Chorley Lecture (1975) 39 Modern Law Review 1 at 4, reprinted in Devlin, The Judge, (1979), at 4.), extolling the virtues of the system of administering justice in the presence of the parties, said:

" This is why impartiality and the appearance of it are

the supreme judicial virtues. It is the verdict that

matters, and if it is incorrupt, it is acceptable. To be

incorrupt it must bear the stamp of a fair trial. The

judge who does not appear impartial is as useless to the

process as an umpire who allows the trial by battle to be

fouled or an augurer who tampers with the entrails."

What his Lordship said of a judge may be said of a juror.

2. The practice and procedure of the criminal court are moulded to ensure that an accused is given no reasonable grounds for apprehending that the judge and jury will not discharge their respective duties impartially. Challenges to the array and challenges to individual jurors, the physical separation of jurors from the judge, prosecutor, accused and witnesses except in the court room, the keeping of the jury together and in isolation while considering their verdict, and judicial directions to jurors to disregard evidence not regularly adduced in court and not to discuss the case with any person other than fellow jurors are some of the steps taken to ensure both the appearance and the reality of a fair trial. If any of these safeguards of a fair trial are breached, the mere breach may lead the hypothetical informed and fair-minded member of the public reasonably to apprehend that the jurors involved will not discharge their duty impartially.

3. In recent years, jurors have been allowed to separate during adjournments of a criminal trial. In earlier times when a trial was adjourned, the practice was to appoint bailiffs to keep the jury and to swear the bailiffs "neither to speak to them themselves, nor suffer any other person to speak to them touching any matter relative to this trial" ((54) R. v. Stone [1796] EngR 2292; (1796) 6 TR 527 at 531 [1796] EngR 2292; (101 ER 684 at 686).). That practice continued well into the 20th century in this country. When the practice changed and jurors were allowed to separate before considering their verdict, a judicial warning was given to the separating jurors in terms corresponding with the keepers' oath. But, in respect of a jury which retires to consider its verdict, the strict practice of keeping the jury together and in isolation from outside contacts was generally continued. Thus, in Reg. v. Chaouk ((55) [1986] VicRp 70; (1986) VR 707.), a conviction was set aside and a new trial ordered where three of the jury, after the jury had retired to consider its verdict, had been transported by taxi, unaccompanied by a keeper, to their overnight accommodation. The possibility of a prejudicial communication during the journey by taxi could not be excluded. In Chaouk, the irregularity in procedure was sufficient by itself to warrant the setting aside of a conviction, although there was no other ground shown for apprehending that the jury had not reached its verdict impartially. Of course, a stricter approach to an irregularity is taken when it affects the isolation of a jury from external contacts while the verdict is being considered than when it affects the isolation of a juror from external contacts at an earlier stage of the trial ((56) R. v. Twiss (1918) 2 KB 853 at 858-859 distinguishing R. v. Ketteridge (1915) 1 KB 467.).

4. However, if a juror and a non-juror engage in a conversation during the course of a trial before the jury retires and refer to the case on which the juror is sitting, the reference can sometimes be the cause of the trial miscarrying. In Reg. v. Spencer ((57) [1987] UKHL 2; (1987) AC 128.) a juror, who had been discharged because he was apparently biased against the accused, drove three of the continuing jury members in his car on their journey home, the trip taking half an hour. In quashing the verdict of guilty Lord Ackner, with whose speech the other members of the House of Lords agreed, endorsed a test of a "real danger" that the accused's position was prejudiced ((58) ibid. at 144.). This is not the test to be applied in this country, where the question whether a juror or the jury should be excused or discharged is answered by reference to the test of reasonable apprehension of bias. In Spencer, Lord Ackner found the verdict to be unsafe and unsatisfactory because he had a "lurking doubt, that justice may not have been done" ((59) ibid. at 146.). Again, this test is not applicable here. Nevertheless, his Lordship's approach indicates that an apprehension that a juror or the jury might not deal with the case impartially may be derived from the occurrence of an irregularity where the irregularity infringes a practice designed to ensure both the appearance and the reality of a fair trial.

5. The basic rule of practice to be applied in a criminal trial was stated by Holroyd ACJ, speaking for the Full Court in Trewartha v. Confidence Extended Co. N.L. ((60) [1906] VicLawRp 48; (1906) VLR 285 at 288-289.):

"It is highly desirable, and it has always been so

considered, that not only should justice be administered

purely and without any actual bias on the one side or the

other on the part of the tribunal which hears the case, but

further that no reasonable ground of suspicion should be

allowed to arise as to the fairness of that tribunal. It

is very desirable, in my opinion, that during a trial by

jury none of the jury should converse with anybody except

their fellow-jurymen on the subject of the case."

In the present case, that rule had been emphasized by Debelle J, who presided at the Mt Gambier sittings of the Supreme Court, in an address to the assembled jury panel before any jury was sworn. The actual address was not recorded but it was given in accordance with custom and the customary form includes this admonition:

"I warn you against speaking to, or having any conversation

with anyone other than your fellow jurors while you are

engaged in a trial. Do not speak to counsel, do not speak

to any witness, do not speak to any other member of the

public in and about the courtroom. Of course there is

no reason why you shouldn't discuss the matter with your

colleagues during the course of your trial. You should

discuss with them any matters that you may have heard

during the trial, provided they are serving on the same

jury as that on which you are then engaged, but don't speak

to anybody else about the trial or the evidence.

The design of this building is such that it is very easy

for you to come into contact with other jurors not involved

in the case, members of the public, witnesses, friends of

persons involved in the case, any manner of persons

attending the court. You must be careful not to get into

conversations with anyone, apart from jurors sitting in

the case with you. You must not allow them to get into

conversation with you."

This admonition was breached by the juror involved in the flower incident which is described by Toohey J in his reasons for judgment.

6. The rule is not absolute in the sense that any breach necessarily results in a trial that is deemed to be unfair. The circumstances might show that there are no reasonable grounds for apprehending unfairness ((61) See, for example, Duff v. The Queen [1979] FCA 83; (1979) 39 FLR 315 at 336- 338; [1979] FCA 83; 28 ALR 663 at 679-681; Reg. v. White (1969) SASR 491.). As Isaacs J said in an intervention in argument in David Syme and Co. v. Swinburne ((62) [1909] HCA 92; (1909) 10 CLR 43 at 47.):

"If after examination of the facts the Court thinks that

there is a suspicion of unfairness there should be a new

trial, but not if the suspicion is wiped away."

7. Unfortunately, I think that the suspicion of unfairness was not wiped away in this case. In my view, the conduct of the juror who made the gift of flowers to the mother of the victim and who thereby breached Debelle J's admonition gave reasonable grounds to apprehend that she might not give impartial consideration to the respective cases of the appellants. I state my reasons for this view by reference to the circumstances that appear to me to be significant.

8. The deceased Patrick had been killed by an injury inflicted in the course of a brutal assault upon him. The brutality and force of the assault were not disputed. Webb had given statements to the police in which he blamed Hay for inflicting the fatal blow. Hay had given evidence in which she had, by implication, attributed the infliction of the fatal blow to Webb. Each of the accused sought to escape conviction by attributing criminal responsibility solely to the other. But, as Debelle J directed the jury, both could be convicted either as joint participants in the fulfilment of a common purpose or on the footing that one had aided and abetted the other in the commission of the crime. It was critical to the acquittal of either accused that the other be found solely responsible for Patrick's death. Having regard to the fact that both had assaulted Patrick, the defences called for a most discriminating analysis of the evidence by the jury. It was the juror's impartiality in making this analysis that was in question.

9. At the time of the flower incident, the trial had been proceeding for more than a month. All the evidence had been given. The members of the jury were in possession of all the facts. The Crown Prosecutor and Counsel for Webb had completed their final addresses. Counsel for Hay was reaching the end of his final address. Debelle J was about to sum up. At that stage of a criminal trial, even of a long criminal trial like the trial of Hay and Webb, there is a heightened sense of drama. The jurors, in whose hands the fate of the accused will shortly be left, become the focus of attention. It was at this stage that the juror, impetuously and generously, made her gesture of sympathy with the mother of the victim. Patrick's mother had been in court during the long trial. The hypothetical member of the public might reasonably have apprehended that she was there to see justice done to the two people who had been involved in assaulting her son before his death. The juror's gesture of sympathy might reasonably have been regarded as a gesture of solidarity with the mother of the victim. That gesture of solidarity might reasonably appear - particularly where each accused was seeking a discriminating consideration of his or her case - to make it difficult for the juror to acquit either of the accused who had been involved in the attack.

10. Sympathy for Patrick's mother was, one might assume, an emotion felt and shared by most of those who knew anything of the crime, particularly those who had heard the evidence at the trial. The fact that a person of the least sensitivity would have felt sympathy in those circumstances is not a factor which deprives the juror's gesture of its significance to the fairness of the trial. To the contrary, sympathy for Mrs Patrick posed the very risk of unfairness against which Debelle J had cautioned the jury in his initial admonitory address:

" I stress the importance of impartiality and objectivity

in the course of your deliberations. You must not be

affected by feelings of sympathy or prejudice one way or

the other. Any such feelings must be completely put to one

side."

Despite this admonition, the strength of the juror's sympathy for Mrs Patrick demonstrated that she had allowed herself to "be affected by feelings of sympathy".

11. It is not surprising that each of the accused, on learning of the juror's gift of flowers, protested through his or her counsel against having his or her fate decided by a jury of which that juror was a member. How was it possible to dispel the reasonable apprehension that the case of each accused would not be decided impartially? True it was that the juror frankly admitted the error in her conduct. But, as Lord Goff of Chieveley pointed out in Reg. v. Gough ((63) [1993] UKHL 1; (1993) AC 646 at 659. See also Chaouk (1986) VR at 713 per Kaye J):

"there are difficulties about exploring the actual state of

mind of a justice or juryman. In the case of both, such

an inquiry has been thought to be undesirable; and in the

case of the juryman in particular, there has long been an

inhibition against, so to speak, entering the jury room and

finding out what any particular juryman actually thought at

the time of decision. But there is also the simple fact

that bias is such an insidious thing that, even though

a person may in good faith believe that he was acting

impartially, his mind may unconsciously be affected by bias

... In any event, there is an overriding public interest

that there should be confidence in the integrity of the

administration of justice, which is always associated

with the statement of Lord Hewart CJ in Rex v. Sussex

Justices, Ex parte McCarthy ((64) (1924) 1 KB 256 at 259.),

that it is 'of fundamental importance that justice should

not only be

done, but should manifestly and undoubtedly be seen to be

done.'"

It was not practically open to counsel for either accused to cross-examine the juror as to her state of mind. Once Debelle J had refused their application to discharge the jury, an accused's only hope of an impartial consideration of his or her case depended on the juror's disregard of the sympathy with Mrs Patrick that the juror had so recently demonstrated.

12. His Honour gave the jury a further and emphatic warning to "sift and weigh ... all of the evidence ... in a dispassionate manner". Although I would readily acknowledge that a warning can eliminate or virtually eliminate unfairness derived from external attempts to influence a jury ((65) Jago (1989) 168 CLR at 49.), it is somewhat artificial to assume that either the juror's admission and apology or the judge's warning removed the grounds for apprehending that the juror would be unable to consider the defence cases impartially. The suspicion of partiality was not "wiped away". The flower incident was, in my opinion, an irregularity of a fundamental kind which vitiated the conviction ((66) Wilde v. The Queen [1988] HCA 6; (1988) 164 CLR 365 at 373, 375.). I would therefore allow the appeal of both appellants.

13. This is not a majority view and I must therefore consider the further ground of Hay's appeal relating to the "accomplice direction".

14. Hay gave evidence at the trial; Webb did not. Hay's evidence covered many topics, including intoxication and her alleged state of mental confusion. Although she admitted to assaulting Patrick, her defence was that, as she did not strike the fatal blow, that blow must have been struck by Webb. During the summing up, Debelle J gave the following direction:

" You will bear in mind that any person in the position of

the accused, and I am speaking quite generally now, any

person in the position of the accused will obviously be

under a strong temptation to consider his or her own

interest exclusively and, if need be, to play down his or

her own part in the matter, if need be, at the expense of

the co-accused. So you must bear in mind the possibility

of that kind of distortion in a trial of this kind even to

the point of deliberately false evidence. However, it is

necessary for me to say more on the subject than that

general observation about evidence that any co-accused

might give in a trial in which more than one person is

jointly charged before a jury.

There is another matter I wish to say in relation to the

co-accused and that is what is often called an accomplice

warning but I think I will deal with that tomorrow."

As Hay was the only accused who gave evidence, this direction could have affected the evaluation of her evidence alone. Her counsel objected. The objection was validly taken, for the direction was contrary to the principle stated in Robinson v. The Queen (No.2) ((67) [1991] HCA 38; (1991) 65 ALJR 644 at 646; [1991] HCA 38; 102 ALR 493 at 495.). In Robinson, the Court said:

"If (the presumption of innocence) is to have any real

effect in a criminal trial, the jury must act on the basis

that the accused is presumed innocent of the acts which

are the subject of the indictment until they are satisfied

beyond reasonable doubt that he or she is guilty of those

acts. To hold that, despite the plea of not guilty, any

evidence of the accused denying those acts is to be the

subject of close scrutiny because of his or her interest in

the outcome of the case is to undermine the benefit which

that presumption gives to an accused person."

The objection by counsel for Hay was taken at the end of a day and argument on the point was left for the next day. The flower incident intervened. After it was dealt with, Debelle J heard argument about the direction he had given. His Honour agreed to redirect the jury. The redirection was as follows:

"What I said (towards the close last night) was a general

warning only. I was not intending, in any way, to suggest

that Ms Hay was a suspect witness. I did not, in any

way, seek to suggest that you subject her evidence to any

different kind of scrutiny from that which you would apply

to the evidence of any other witness. My purpose was to

lead into the topic I'm now about to discuss.

The topic on which the law does require me to give a

warning, that is to say, give a warning about the evidence

of a witness who is an accomplice. I emphasize indeed what

I have just been saying and what I said at an early stage

in the course of my summing up, that you should examine and

test the evidence of Ms Hay in the same way as you test

the evidence of other witnesses. You should deal with her

evidence just as you would deal with any evidence of any

other witness."

However, his Honour did subsequently give the jury an accomplice warning in the following terms:

" I must warn you, that it is dangerous to convict a

person in a case like this, if the only evidence against

him or her is the uncorroborated evidence of an accomplice.

That does not mean that you cannot convict or take his

evidence into account without corroboration. The evidence

of such a witness is always to be scrutinized with care and

if there is no corroboration, with special care. If, in

this case, you find there is no corroboration, you should

bear in mind the warning I have given to you and you should

not place any reliance upon anything that Ms Hay says in

the witness box in implicating Mr Webb, unless you are

convinced the evidence is reliable."

In explaining why an accomplice warning was necessary the trial judge said:

"While the Crown says that each of these two accused is

guilty of murder, it also relies, to some extent, in

proving that, on certain parts of the evidence that Ms Hay

gave in this court. Because of that, it is necessary for

me to give you a special warning. It has been found from

experience that there are certain classes of witnesses,

whether they are accused persons or not, whose evidence is

inherently suspect for one reason or another. One type of

witness who falls into that category is the accomplice,

that is to say, the person who knowingly assists another to

conduct a crime. On one view of the evidence against each

of the accused in this case, he or she can be regarded as

having been an accomplice to the murder of Lance Patrick."

15. Debelle J attempted to make it clear that by giving the accomplice warning he was not expressing any view about the guilt or innocence of the two accused. It was only if either of the two accused "was present and gave assistance or encouragement to the other in the commission of the murder, then that accused falls into the class of suspect witnesses ... that is to say, an accomplice." However, the effect of these directions was to advise the jury that Hay's evidence might be "inherently suspect" and should be "scrutinized with care and if there is no corroboration, with special care" if she was "present and gave assistance or encouragement to (Webb) in the commission of the murder". That direction, so far as it affected the jury's evaluation of Hay's evidence in her own case, ran contrary to the Robinson principle. At the least, it undid the redirection which his Honour had given the jury that Ms Hay was not to be treated as a "suspect witness". Her responsibility for the crime should not have been determined under the influence of a direction devaluing her evidence.

16. Of course the accomplice direction was intended to protect Webb against the jury's acting on Hay's inculpation of Webb without considering the possibility that Hay's self-interest might account for the evidence she had given. But the jury could hardly have failed to consider Hay's motives. It was obvious that Hay was inculpating Webb by exculpating herself: she sought to place on Webb sole responsibility for the murder. The purpose of an accomplice warning is to alert the jury to the possibility that an accomplice's evidence inculpating an accused might be given in order to exculpate the accomplice or to serve some other purpose of the accomplice. In the circumstances of this case, an accomplice warning was a warning of the obvious.

17. Having regard to the authorities canvassed by Toohey J, I respectfully agree that when an accused gives evidence implicating another accused the question whether an accomplice warning should be given and, if so, in what terms, cannot be answered without reference to the unique circumstances of the case. But the Court of Criminal Appeal of New South Wales was surely right to say in Reg. v. Henning ((68) Unreported, 11 May 1990 at 49.):

"For whereas the standard form of corroboration warning,

with all its complexities, may well be inappropriate or

even undesirable in relation to an accused who gives

evidence inculpating a co-accused, there is one matter

which must be stressed in all such cases where a warning is

given. It is essential in the interest of the accused who

gives the evidence that the warning should be restricted in

terms to those parts of the evidence which inculpate any

co-accused. It must be made clear to the jury that the

warning is to be applied only when they are considering the

case against the co-accused. It must not be left open to

them to believe that the warning might attach to the

accused's evidence in his own case."

This solution to the problem is logically attractive but if it were a general rule that an accomplice direction should be given subject to the qualification stated by the Court of Criminal Appeal in Henning, confusion would often be engendered in the minds of jurors. Confusion would be especially likely when the same part of an accused witness' testimony exculpates the accused witness and inculpates the co-accused. The jury would then be directed to treat that evidence in one way in deciding the guilt or innocence of the accused witness and in another way when deciding the guilt or innocence of the co-accused inculpated by the evidence. I respectfully agree with Lord Taylor of Gosforth CJ in Reg. v. Cheema in saying ((69) (1994) 1 WLR 147 at 157.):

"the complication involved in requiring a judge to give

full corroboration directions in respect of co-defendants

implicating each other, would be likely to confuse

and bewilder a jury. Especially if there are several

defendants, the difficulty of giving the full warning in

relation to each, and identifying which pieces of evidence

are capable of corroborating each of them, would create a

minefield of difficulties."

18. There is no rule of law that a warning should be given when the alleged accomplice who testifies is one of the accused. In Cheema ((70) ibid. at 156.) Lord Taylor CJ, who was speaking for the Court of Appeal, said:

" The effect of this considerable body of case law is to

show that in recent years time and again the court has

reiterated that although a warning in suitable terms as to

the danger of a co-defendant having an axe to grind is

desirable, there is no rule of law or practice requiring a

full corroboration direction."

The position is different when an accomplice is a witness for the prosecution. In such a case, the problem of giving a warning that reflects adversely on the accomplice's evidence does not arise, for the accomplice is not in peril of conviction. Indeed, in such cases, the accomplice is usually convicted and punished or the accomplice is pardoned before being called to give evidence. But when an accused, being in peril of conviction, gives self-exculpatory evidence which inculpates a co-accused, two factors will frequently militate against the giving of an accomplice warning. First, as the warning is to be acted on only if the accused witness is an accomplice, the jury must address that question before proceeding to consider the case against the co-accused. That course may be prejudicial to the accused witness. Secondly, as the jury must not be directed to treat the evidence of the accused witness differently from the evidence of other witnesses in considering the case against the accused witness, an accomplice warning to scrutinize the evidence of the accused witness carefully must be clearly qualified as Henning suggests if it is not to carry an implication that runs counter to the principle in Robinson.

19. Without attempting to prescribe a universal rule applicable to cases where an accused witness gives evidence exculpating himself but implicating a co-accused, it is generally preferable not to give an accomplice warning in respect of that evidence unless, in the particular circumstances, the trial judge is of the opinion that the jury might fail to appreciate the risk of acting on that evidence against the co-accused. In such a case, the distinction drawn in Henning must be carefully explained to the jury.

20. This was not such a case. The risk of acting on Hay's evidence so as to convict Webb alone was obvious. Even if it had been right to give some warning about Hay's evidence, the Henning distinction was not clearly drawn. Had I not been prepared to hold that the flower incident makes it necessary to quash both convictions, I would have allowed Hay's appeal on the ground of misdirection.

21. I would allow both appeals, set aside the judgment of the Court of Criminal Appeal and in lieu thereof allow the appeals to that Court, quash the convictions and order a new trial of the appellants.

DEANE J The background facts are set out in the judgment of Toohey J. Except to the extent necessary for the purposes of discussion, I refrain from repeating them. The primary question is whether the conviction of each of the appellants of the murder of Lance Edward Patrick should be quashed on the ground of an appearance of bias on the part of a juror who, towards the conclusion of the appellants' lengthy trial, handed a bunch of daffodils, which she had picked from her garden, to the mother of Mr Patrick's fiancee with the request that she give them to Mr Patrick's mother. After the incident occurred, the appellants, through their counsel, made plain that they objected to the trial continuing with a jury which included the particular juror. The learned trial judge dealt with the matter on the basis that he had power under s.56 of the Juries Act 1927 (S.A.) to excuse the particular juror and continue the trial with a reduced jury. His Honour ruled, however, that the juror should not be excused.

2. In a series of recent cases ((71) See, in particular, Reg. v. Watson; Ex parte Armstrong [1976] HCA 39; (1976) 136 CLR 248 at 258-263; Re Judge Leckie; Ex parte Felman (1977) 52 ALJR 155 at 158; 18 ALR 93 at 97-98; Re Shaw; Ex parte Shaw (1980) 55 ALJR 12 at 14, 16; 32 ALR 47 at 50-51, 54; Livesey v. New South Wales Bar Association [1983] HCA 17; (1983) 151 CLR 288 at 293-294, 300; Re J.R.L.; Ex parte C.J.L. [1986] HCA 39; (1986) 161 CLR 342 at 349-350, 351-352, 359, 368, 371; Vakauta v. Kelly [1989] HCA 44; (1989) 167 CLR 568 at 572, 575; Laws v. Australian Broadcasting Tribunal [1990] HCA 31; (1990) 170 CLR 70 at 81, 87, 96, 99-100.), the Court has formulated the test to be applied in this country in determining whether a judicial officer ("a judge") is disqualified by reason of the appearance of bias, as distinct from proved actual bias. That test, as so formulated, is whether, in all the circumstances, a fair-minded lay observer with knowledge of the material objective facts "might entertain a reasonable apprehension that (the judge) might not bring an impartial and unprejudiced mind to the resolution of the question" in issue. The quoted words in that statement of the test are taken from the judgment of the Court in Livesey v. New South Wales Bar Association ((72) (1983) 151 CLR at 293-294.). In that case, and in a number of the other cases, the test was stated in terms of an apprehension on the part of "the parties or the public" ((73) See, e.g., Reg. v. Watson; Ex parte Armstrong (1976) 136 CLR at 262; Re Judge Leckie; Ex parte Felman (1977) 52 ALJR at 158; 18 ALR at 98; Re Shaw; Ex parte Shaw (1980) 55 ALJR at 14, 16; 32 ALR at 50, 54; Livesey v. New South Wales Bar Association (1983) 151 CLR at 293-294; Re J.R.L.; Ex parte C.J.L. (1986) 161 CLR at 351; Grassby v. The Queen [1989] HCA 45; (1989) 168 CLR 1 at 20.). So stated, the test directly reflects its rationale, namely, that it is of fundamental importance that the parties to litigation and the general public have full confidence in the integrity, including the impartiality, of those entrusted with the administration of justice ((74) See, e.g., Reg. v. Watson; Ex parte Armstrong (1976) 136 CLR at 263; Reg. v. Gough [1993] UKHL 1; (1993) AC 646 at 659.). However, the test is an objective one and the standard to be observed in its application is that of a hypothetical fair-minded and informed lay observer ((75) See, e.g., Stollery v. Greyhound Racing Control Board [1972] HCA 53; (1972) 128 CLR 509 at 517, 519; Livesey v. New South Wales Bar Association (1983) 151 CLR at 299, 300; Builders' Reg. Board of Qld v. Rauber (1983) 57 ALJR 376 at 380, 389-390; 47 ALR 55 at 62, 80; Laws v. Australian Broadcasting Tribunal (1990) 170 CLR at 87- 88, 95-96, 98-100; Vakauta v. Kelly (1989) 167 CLR at 572, 573, 576, 585. And cf. Reg. v. Australian Stevedoring Industry Board; Ex parte Melbourne Stevedoring Co. Pty. Ltd. [1953] HCA 22; (1953) 88 CLR 100 at 116 ("reasonable persons"); Reg. v. Watson; Ex parte Armstrong (1976) 136 CLR at 263 ("fair-minded people").). That being so, it is convenient to frame the test itself in terms of reasonable apprehension on the part of that particular inhabitant of the common law. I have used the word "apprehension" in preference to the word "suspicion" for the reason that the latter word is capable of conveying shades of meaning which are inappropriate in this context. As a practical matter, however, there is little, if any, difference between the content of the two words when prefaced by "reasonable" and I have, in referring to authority in this judgment, on occasion treated them as interchangeable ((76) In particular in fn.(73) (above) and fn.(79) (below).).

3. None of the recent cases in the Court in which the above test was formulated and applied was concerned with an allegation of an appearance of bias on the part of a juror. The "reasonable apprehension" test has, however, been applied by the Court in cases involving a statutory officer other than a judge ((77) See, e.g., Reg. v. Commonwealth Conciliation and Arbitration Commission; Ex parte Angliss Group (1969) 122 CLR 546; Stollery v. Greyhound Racing Control Board [1972] HCA 53; (1972) 128 CLR 509; Builders' Reg. Board of Qld v. Rauber (1983) 57 ALJR 376; 47 ALR 55; Laws v. Australian Broadcasting Tribunal [1990] HCA 31; (1990) 170 CLR 70. See also Metropolitan Properties Co. C'(F.G.C.) Ltd. v. Lannon (1969) 1 QB 577.) and there is no convincing reason of principle why the test applicable to a case involving an allegation of an appearance of bias on the part of a juror entrusted with the discharge of the curial function of deciding questions of fact should be different from that applicable to a judge. In my view, the "reasonable apprehension test" should be applied regardless of whether a question of the appearance of bias arises in relation to a judge, a statutory office holder who is obliged to observe the requirements of procedural fairness ((78) It is unnecessary to consider the position of a domestic tribunal exercising non-statutory powers: cf., Builders' Reg. Board of Qld v. Rauber (1983) 57 ALJR at 390; 47 ALR at 80.) or a juror. I note that the application of that test to cases involving a juror is supported by the weight of recent authority in other courts in this country ((79) See Reg. v. Hodgkinson [1954] VicLawRp 1; (1954) VLR 140; Reg. v. Chaouk [1986] VicRp 70; (1986) VR 707; Reg. v. Emmett (1988) 14 NSWLR 327 at 339; Reg. v. Fielding (1993) 1 Qd R 192.). On the other hand, it conflicts with the conclusion of the House of Lords in the recent case of Reg. v. Gough ((80) [1993] UKHL 1; (1993) AC 646.).

4. In Reg. v. Gough, the House of Lords held that the appropriate test to be applied by an appellate court (or a trial judge), in determining a question of the appearance of bias on the part of a juror, is whether, in the circumstances of the particular case, it appears to the appellate court (or trial judge) that there was (or is) "a real danger", in the sense of a real possibility, of such bias. The principal speech, with which the other members of the appellate committee expressed their agreement, was delivered by Lord Goff of Chieveley. In summarizing his conclusions, his Lordship said ((81) (1993) AC at 670.):

"I think it possible, and desirable, that the same test

should be applicable in all cases of apparent bias, whether

concerned with justices or members of other inferior

tribunals, or with jurors, or with arbitrators. ...

Furthermore, I think it unnecessary, in formulating the

appropriate test, to require that the court should look at

the matter through the eyes of a reasonable man, because

the court in cases such as these personifies the reasonable

man; and in any event the court has first to ascertain

the relevant circumstances from the available evidence,

knowledge of which would not necessarily be available to an

observer in court at the relevant time. Finally, for the

avoidance of doubt, I prefer to state the test in terms of

real danger rather than real likelihood, to ensure that

the court is thinking in terms of possibility rather than

probability of bias. Accordingly, having ascertained the

relevant circumstances, the court should ask itself whether,

having regard to those circumstances, there was a real

danger of bias on the part of the relevant member of the

tribunal in question, in the sense that he might unfairly

regard (or have unfairly regarded) with favour, or

disfavour, the case of a party to the issue under

consideration by him".

Lord Goff did not expressly refer to the case where the question of an appearance of bias arises in relation to a member of the judiciary other than a local justice or member of an inferior tribunal. However, the tenor of his Lordship's comments in the above passage, particularly the first sentence, and some references in other parts of his speech ((82) Particularly the reference (ibid. at 661) to Dimes v. Proprietors of Grand Junction Canal [1852] EngR 789; (1852) 3 HL Cas 759.) seem to indicate that he considered that the test of a real danger of bias was to be applied generally to cases involving any person "who sits in a judicial capacity" ((83) Metropolitan Properties Co. C'(F.G.C.) Ltd. v. Lannon (1969) 1 QB at 599.). In that regard, it is relevant to note that Lord Woolf, who agreed with Lord Goff's reasons, treated the test as applicable to "a judge" ((84) (1993) AC at 672.).

5. The House of Lords test differs from that accepted in recent cases in this Court as regards both its substance and its reference point. The substance of the House of Lords test is "a real danger of bias". The substance of this Court's test is "a reasonable apprehension of bias". The reference point of the House of Lords test is the appellate court itself or, where the question arises at first instance, the trial judge. The reference point of this Court's test is the fair-minded informed lay observer.

6. Quite apart from the respect which the courts of this country accord any decision of the House of Lords, the decision in Gough is important for present purposes for the reason that this Court's acceptance of the "reasonable apprehension" test in preference to a "real likelihood" or "real danger" test was, to a significant extent, founded upon a perception that English, as well as Australian, authority supported that course. In particular, the judgment of Barwick CJ, Gibbs, Stephen and Mason JJ in Reg. v. Watson; Ex parte Armstrong ((85) (1976) 136 CLR at 258-260.) contains an analysis of English cases which led their Honours to conclude that any conflict between a "real likelihood" test and a "reasonable suspicion" test had been "similarly resolved" (i.e. in favour of the "reasonable suspicion" test) in both Australia and England ((86) ibid. at 260.). Reg. v. Gough demonstrates that, at the time of this Court's decision in Watson, the resolution of any such conflict in England in favour of the reasonable suspicion or apprehension test was at best tentative and that the ultimate resolution was to be in favour of the "real likelihood" or "real danger" test ((87) See also Reg. v. Spencer [1987] UKHL 2; (1987) AC 128 at 144.).

7. Nonetheless, I am of the firm view that the "reasonable apprehension" test should continue to be accepted in this country. That test was adopted only after a careful consideration by the Court of the competing claims of the "real likelihood" or "real danger" test. It cannot be said that the Court would have adopted some different test were it not for its reliance, in Watson, on English authority. In those circumstances the Court would not be justified in overruling the series of recent cases in which it formulated and applied the "reasonable apprehension" test unless it was persuaded that that test is misconceived or inappropriate. I am far from being so persuaded. To the contrary, it appears to me that, in so far as this country is concerned, the "reasonable apprehension" test is the more appropriate one.

8. The adoption of a "real likelihood" or "real danger" test, with the appellate court (or the trial judge) itself as the reference point, would, in my view, go a long way towards substituting, for the doctrine of disqualification by reason of an appearance of bias, a doctrine of disqualification for actual bias modified by the adoption of a new standard of proof (i.e. a real likelihood or possibility rather than probability in the sense of more likely than not). It is true that, as Lord Goff made clear in Reg. v. Gough ((88) See, in particular, (1993) AC at 659 and, to the same effect, per Lord Woolf at 672.), the inquiry which is involved in the application of the real danger test is not directed to an exploration of the actual state of mind of the particular judge or juror. It is directed to the court's assessment of the possibilities in the context of the objective facts disclosed by the material in evidence. Nonetheless, the ultimate question which a court is required to address in an application of that test is whether there was a real danger, in the sense of possibility, of actual bias. The adoption of a test requiring the determination of that ultimate question for the resolution of cases involving no more than an allegation of an appearance of bias would, in my view, be undesirable in this country for the following main reasons.

9. One advantage of the test of reasonable apprehension on the part of a fair-minded and informed observer is that it makes plain that an appellate court is not making an adverse finding on the question whether it is possible or likely that the particular judge or juror was in fact affected by disqualifying bias ((89) See, e.g., Livesey v. New South Wales Bar Association (1983) 151 CLR at 294-295.). In contrast, the real danger test is focused upon that very question. Regardless of an appellate court's care to make plain that its finding is only one of possibility of danger, such a finding is likely to be unfairly damaging to the reputation of the person concerned who will commonly not have been a party to the proceedings before the appellate court and whose subjective thought processes will not have been investigated in the appellate court. In addition, in the ordinary case where there will in fact have been no bias, a finding by an appellate court that there was "a real danger" that a particular judge had been affected by disqualifying bias would be liable to cause unjustified damage to the public confidence in the judiciary which the requirement of the appearance of justice was intended to protect and preserve. Even more importantly, such a finding could, in the case of a judge, quite wrongly undermine "the confidence that his integrity is beyond question" which "supports him not only in his judgment but in all his words and conduct, both that which may be approved and that which may be disapproved" ((90) Reg. v. Watson; Ex parte Armstrong (1976) 136 CLR at 294.).

10. Conversely, an appellate court's awareness of the consequences of a finding of a real danger (or likelihood) of bias on the part of a judge and its confidence in the dedication and integrity of the members of its judicial system could well lead to a situation in which insufficient attention was paid to the rationale of the doctrine of disqualification by reason of an appearance of bias, namely, that justice "should manifestly and undoubtedly be seen to be done" ((91) R. v. Sussex Justices; Ex parte McCarthy (1924) 1 KB 256 at 259.). In that regard, it is well to remember that Lord Hewart CJ was not, in identifying that rationale, referring to the "fundamental importance" that justice should manifestly and undoubtedly be seen by other members of the judiciary to be done. He was referring to the fundamental importance that that should be the perception of both the parties and the general public. That point was well made by Barwick CJ, Gibbs, Stephen and Mason JJ in their joint judgment in Watson ((92) (1976) 136 CLR at 262-263.):

"his (i.e. Lord Hewart's) statement of principle, which was

recently reaffirmed in this Court in Stollery v. Greyhound

Racing Control Board ((93) (1972) 128 CLR at 518-519.) does

go to the heart of the

matter. It is of fundamental importance that the public

should have confidence in the administration of justice. If

fair-minded people reasonably apprehend or suspect that the

tribunal has prejudged the case, they cannot have confidence

in the decision. To repeat the words of Lord Denning M.R.

which have already been cited, 'Justice must be rooted in

confidence: and confidence is destroyed when right-minded

people go away thinking: "The judge was biased."'"

11. There is support in some learned writings ((94) See, e.g., Evans, De Smith's Judicial Review of Administrative Action, 4th ed., (1980) at 264.) and judgments ((95) See, e.g., Reg. v. Altrincham Justices; Ex parte Pennington (1975) QB 549 at 553-554.) for the view that the reasonable apprehension or suspicion test and the real likelihood of danger test should both be retained and applied, either alternatively or cumulatively, depending upon the appropriateness of one or both to the circumstances of the particular case. I do not accept that view. If the test of a reasonable apprehension on the part of a fair-minded observer with knowledge of the material objective facts fell to be applied by reference only to those facts which were apparent at the time, there would be much to be said for the view that the real likelihood or real danger test should be retained to be applied in cases where some of the damaging material facts - whether prior, contemporaneous or subsequent -as ascertained by the appellate court were not known at the time of the proceedings. In my view, however, the material objective facts are not so confined for the purposes of the test. The fair-minded observer is a hypothetical figure. While the question is not settled by any decision of the Court, it appears to me that the knowledge to be attributed to him or her is a broad knowledge of the material objective facts as ascertained by the appellate court ((96) See Laws v. Australian Broadcasting Tribunal (1990) 170 CLR at 87 per Mason CJ and Brennan J: "we must attribute to him or her knowledge of the actual circumstances of the case"; S. and M. Motor Repairs v. Caltex Oil (1988) 12 NSWLR 358 at 368-369, 381; Morris (1991) 93 Cr App R 102 at 106 per Farquharson LJ: "a reasonable and fair minded person sitting in the court and knowing all the relevant facts".), as distinct from a detailed knowledge of the law or knowledge of the character or ability of the members of the relevant court ((97) See Livesey v. New South Wales Bar Association (1983) 151 CLR at 299; Vakauta v. Kelly (1989) 167 CLR at 573, 585; Laws v. Australian Broadcasting Tribunal (1990) 170 CLR at 87-88, 98.). The material objective facts include, of course, any published statement, whether prior, contemporaneous or subsequent, of the person concerned. If, in the particular case, the proper conclusion is that a fair-minded lay observer with a broad knowledge of those facts would not entertain a reasonable apprehension of bias, that is the end of the issue of disqualification by reason of an appearance of bias. Strictly speaking, it is unnecessary for the purposes of the present case, where no allegation of actual bias has been made, to decide whether an appellate court should entertain such an allegation. I would, however, indicate that I consider that the reasonable apprehension test is of such broad and general application that it is unnecessary and inappropriate for an allegation of actual bias to be raised before or determined by an appellate court ((98) See, e.g., R. (De Vesci) v. Justices of Queen's County (1908) 2 IR 285 at 294; Reg. v. Barnsley Licensing Justices (1960) 2 QB 167 at 187; Re J.R.L.; Ex parte C.J.L. (1986) 161 CLR at 356- 357.).

12. The area covered by the doctrine of disqualification by reason of the appearance of bias encompasses at least four distinct, though sometimes overlapping, main categories of case. The first is disqualification by interest, that is to say, cases where some direct or indirect interest in the proceedings, whether pecuniary or otherwise, gives rise to a reasonable apprehension of prejudice, partiality or prejudgment. The second is disqualification by conduct, including published statements. That category consists of cases in which conduct, either in the course of, or outside, the proceedings, gives rise to such an apprehension of bias. The third category is disqualification by association. It will often overlap the first ((99) e.g., a case where a dependent spouse or child has a direct pecuniary interest in the proceedings.) and consists of cases where the apprehension of prejudgment or other bias results from some direct or indirect relationship, experience or contact with a person or persons interested in, or otherwise involved in, the proceedings. The fourth is disqualification by extraneous information. It will commonly overlap the third ((100) e.g., a case where a judge is disqualified by reason of having heard some earlier case: see, e.g., Livesey v. New South Wales Bar Association [1983] HCA 17; (1983) 151 CLR 288; Australian National Industries v. Spedley Securities (1992) 26 NSWLR 411.) and consists of cases where knowledge of some prejudicial but inadmissible fact or circumstance gives rise to the apprehension of bias.

13. Within the first category of case, i.e. disqualification by interest, the general rationale underlying the doctrine is reinforced by the principle expressed in the maxim that nobody may be judge in his own cause ((101) See, e.g., Co.Litt. 141.a; Dickason v. Edwards [1910] HCA 7; (1910) 10 CLR 243 at 259; Australian Workers' Union v. Bowen (No.2) [1948] HCA 35; (1948) 77 CLR 601 at 631.). Indeed, there is one special class of case within that first category in which, subject to the possible operation of the rule of necessity ((102) See Builders' Reg. Board of Qld v. Rauber (1983) 57 ALJR at 385- 386, 392; 47 ALR 72-73, 84; Laws v. Australian Broadcasting Tribunal (1990) 170 CLR at 88-89, 96-98, 102.), the effect of that principle is that disqualification is automatic without there being any "question of investigating, from an objective point of view, whether there was any real likelihood of bias, or any reasonable suspicion of bias, on the facts of the particular case" ((103) Reg. v. Gough (1993) AC at 661 per Lord Goff.). That special class consists of cases in which the judge, juror or statutory officer has a direct pecuniary interest ((104) In the sense of an interest sounding in money or money's worth. See ibid. at 673: "pecuniary or proprietary interest".) in the outcome of the proceedings. In such cases, public confidence in the administration of justice requires that there be disqualification regardless of the particular circumstances ((105) See ibid. at 661.). It is unnecessary, for present purposes, to consider whether that special class of case should be expanded or whether there are other special classes of case in which disqualification by reason of an apprehension of bias is automatic since it is clear that the present case does not fall within any such special class. It would, however, seem appropriate to indicate that I see great force in the view expressed by Lord Goff and Lord Woolf in Reg. v. Gough ((106) ibid. at 664, 673.) to the effect that automatic disqualification should be confined to cases of direct pecuniary interest. That is not, of course, to deny that there will be cases where such a direct pecuniary interest does not exist but where the nature of the relevant interest and/or relationship is such that it is obvious that the person concerned is disqualified by reason of a reasonable apprehension of bias ((107) See, e.g., fn.(99) (above).).

14. The present case falls within the second of the above categories, namely, disqualification by conduct. That being so, it follows from the foregoing that the question for the learned trial judge was whether, in all the circumstances, the conduct of the particular juror would cause a fair-minded lay observer with knowledge of the material objective facts to entertain a reasonable apprehension that the particular juror, and/or other jurors under her influence ((108) cf., Reg. v. Spencer (1987) AC at 146.), might not bring an impartial and unprejudiced mind to the determination of the appropriate verdict in the case of each of the accused persons. His Honour's comments indicate that, instead of addressing that question, he addressed the question whether there was, in his own view, a real possibility of actual bias. That means that he failed to apply the correct test.

15. Nor was the question of the effect of the juror's conduct on a fair-minded informed lay observer addressed in terms by the members of the Court of Criminal Appeal. In dealing with that question, King CJ (with whose judgment Cox and Matheson JJ agreed) said:

"There was no reason to apprehend any prejudice to the

appellants. Any person with human feelings would be likely

to feel sympathy for the mother of a man who had met a

violent death particularly when the mother was present in

court during the trial. The death of the deceased was bound

to excite sympathy for those to whom he was dear. There is

nothing to indicate that that natural sympathy diverted the

jury or any member of it from a dispassionate consideration

of the issue of the guilt or innocence of the appellants,

and there was no basis for a reasonable suspicion of bias on

the part of the jury or any member of it. In my opinion the

learned judge's refusal to discharge the jury and abort the

trial was correct. The learned judge gave the jury a clear

direction designed to ensure that feelings of sympathy would

not divert them from their task" (emphasis added).

It is true that there is probably implicit in those comments, particularly in the words which I have emphasized, a conclusion that a fair-minded lay observer with knowledge of the material objective facts would not have entertained a reasonable apprehension of bias. It seems to me, however, that the circumstances of the present case are such that it is necessary that the question of the effect of the juror's conduct upon the fair-minded lay observer be expressly addressed and answered.

16. While the test of reasonable apprehension on the part of a fair-minded informed lay observer is to be applied in this country in cases involving a judge, a juror or a statutory office holder required to observe procedural fairness, the standard which such an observer would require of each will vary according to the function being discharged and the particular circumstances. This is particularly so in a case of alleged disqualification by conduct. Moreover, in the case of conduct by a juror, the question whether there is a reasonable apprehension of disqualifying bias will, in a case such as the present where the material facts relating to the conduct become apparent at the trial, ordinarily fall to be answered in a context where it can be assumed (by the trial judge) or is known (by an appellate court) that appropriate directions about the need for impartiality will be (or have been) given with the object of removing or minimizing any possibility of either actual or ostensible bias. Thus, in the present case, the question whether there was an appearance of bias by reason of the conduct of the particular juror must be addressed and answered in the context of the clear directions which the learned trial judge gave about the need for impartiality and objectivity on the part of all jurors.

17. It was strongly argued by the Crown that, in the present case, a fair-minded informed lay observer would consider that the conduct of the juror in sending a bunch of flowers to Mr Patrick's mother indicated no more than that the juror felt the sympathy which any normal person would feel for the mother of a son who had been brutally killed, particularly, as King CJ pointed out, when the mother had sat in court listening to evidence of the circumstances of the killing. I was, myself, initially inclined to accept that argument. Further consideration has, however, convinced me that it fails adequately to take account either of the particular background circumstances of this case or of the fact that the material before the Court discloses that the juror's actions were contrary to clear instructions given by the trial judge about the conduct required of members of the jury.

18. As has been said, Mr Patrick had been brutally killed. It was clear that the appellants had both been involved in the circumstances of the killing and that each of them had assaulted him. The issues in the trial arose from the fact that each of the appellants, either directly or by implication, accused the other of inflicting the fatal blow or blows. In these circumstances, it is obvious that members of Mr Patrick's family would inevitably be strongly antagonistic towards the appellants. In a context where his mother had been in regular attendance at the trial and where his fiancee had given evidence for the prosecution, it would also be almost inevitable that the fair-minded observer would be likely to closely identify the mother and other members of Mr Patrick's family and his fiancee and her mother with those involved in the prosecution of the two persons involved in the killing of her son.

19. In this particular case, there was another background circumstance which must be mentioned. It is that Mr Patrick was white skinned while the appellants were both dark skinned, Ms Hay being an Aboriginal and Mr Webb being of mixed Samoan and Indonesian descent. The trial was being held in the South Australian provincial city of Mt. Gambier where the killing had occurred. Before the commencement of the trial, there had been an unsuccessful application on the part of the appellants for a change of venue based, among other things, on an alleged fear of an adverse general attitude of the Mt. Gambier community resulting from an alleged "spate of serious violent crimes ... at Mt. Gambier in which ... the accused (were) aboriginals". The learned trial judge dismissed the application for a change of venue on the ground that the material placed before him did not establish any such general attitude on the part of the Mt. Gambier community. In so doing, his Honour pointed out that it had not been suggested that the residents of Mt. Gambier otherwise had a "prejudicial attitude" to Aboriginals. Nonetheless, it would be to close one's eyes to reality to fail to acknowledge that the racial differences would, in the eyes of a fair-minded lay observer, tend to add emphasis to the division between two "camps", namely, the lawyers and witnesses for the prosecution and those associated with the deceased man on the one hand and the appellants, their lawyers and those associated with them on the other. Clearly, the case was one in which special vigilance was necessary to safeguard the appearance of impartial justice.

20. The law has long recognized the need to ensure an absence of contact between members of an empanelled jury and those interested in, or concerned about, the outcome of the particular trial ((109) See, e.g., Co.Litt at 227.b; R. v. Taylor (1950) NILR 56 at 67-73; Reg. v. Chaouk (1986) VR at 710; Barry, "On the Segregation of Jurors" (1953) 6 Res Judicatae 139.). Notwithstanding the abandonment of past methods designed to ensure that the members of a criminal jury were isolated, the position remains that any extraneous contact between a juror and a person with a special interest in, or concern about, the outcome of the trial is a serious irregularity in the administration of justice. It is common ground that, in the present case, the jurors had, as part of an assembled jury panel in waiting, received clear instructions from the trial judge to the effect that it was incumbent upon them to avoid contact with counsel, witnesses or members of the public in and about the courtroom. The conduct of the juror in asking Mrs Griffiths, the mother of Mr Patrick's fiance, to convey the gift of daffodils to Mrs Patrick was a flagrant and presumably knowing breach of both the letter and the spirit of that direction ((110) Presumably, the juror's conduct was so seen by other members of the jury since, after the incident had been drawn to the trial judge's attention on the following day, the juror informed his Honour that "(t)he jury have just said, why did I not tell them yesterday and they would have told you yesterday."). The fact that the conduct constituted such a breach would be seen by a fair-minded lay observer as demonstrating that the juror's sympathy for, and possible identification with, the mother of the deceased was such as to override the juror's observance of clear instructions about her duty as a juror. It would, in the context of what has been said above, be but a small step for the fair-minded lay observer to conclude that such a degree of sympathy for, and possible identification with, the mother of the deceased would be reflected by a corresponding degree of abhorrence of, and possible antagonism towards, the appellants. In the particular circumstances of this case and notwithstanding the directions of the learned trial judge about the need for objectivity and impartiality, it appears to me that the conclusion is unavoidable that a fair-minded lay observer with knowledge of the material objective facts would be most likely to entertain a reasonable apprehension that the juror concerned would not approach the task of the determination of guilt or innocence either objectively or with an impartial and unprejudiced mind.

21. It follows that the appeals should be allowed and a new trial ordered. Strictly speaking, it is unnecessary that I deal with the separate grounds raised by the appeal of Ms Hay. However, since the other members of the Court have dealt with those separate grounds, it is appropriate that I indicate my views in relation to them.

22. The first of those grounds is that there should have been an order for separate trials. In the particular circumstances of this case, it was strongly arguable that separate trials were appropriate. On the other hand, the trial was destined to be a lengthy one and there were strong considerations, including the desirability of placing the whole picture before the jury, favouring a joint trial in the interests of the administration of justice. The question for this Court is not whether it was, on balance, preferable that an order should have been made for separate trials. It is whether there was, in the event, a miscarriage of justice by reason of the fact that Ms Hay was subjected to a joint trial. In my view, in the context of the directions given by the learned trial judge, it cannot be said that there was such a miscarriage of justice.

23. I would, however, wish to stress that it is important that general comments by appellate judges about the desirability of placing the whole picture before the jury should not be misconstrued as an implicit endorsement of the notion that a consideration favouring a joint trial is that it will enable evidence which is inadmissible against a particular accused to be placed before the jury charged with the determination of the guilt or innocence of that accused. Such comments should be understood as referring only to evidence, such as the sworn evidence of one accused, which is admissible against both accused and which might otherwise be unavailable to be led by the Crown. So far as evidence which is not admissible against both accused, such as a confessional or unsworn statement by one of them, is concerned, the fact that it will be placed before the jury charged with determining the guilt or innocence of the other accused should always be seen as a factor militating in favour of separate trials.

24. Nor do I subscribe to the view that the reasons which favour the joint trial of persons who are charged with committing an offence jointly are particularly strong in cases where such persons seek to cast the blame on one another. Particularly where the accused has made a confessional statement and in jurisdictions where an accused is permitted to make an unsworn statement, the dangers of unfair prejudice to one accused from material which is inadmissible against him or her being placed before the jury seem to me to be exacerbated in such cases. Far from the desirability of avoiding "inconsistent verdicts" assuming particular importance, there is a particular danger in such cases that popular notions of the need for consistent verdicts may tend to subvert the requirement of proof beyond reasonable doubt. If, for example, each of two defendants seeks to exculpate himself or herself from guilt of a crime, which both or one of them undoubtedly committed, by casting the entire blame on the other, it is difficult to see any particular relevance of the need for consistent verdicts apart from the superficial and mistaken notion that there would be something "inconsistent" about an acquittal of both. Indeed, where there is a joint trial in such a case, it is desirable that the trial judge stress to the jury that, while the jury may think it apparent that the crime was committed by at least one of the accused, there would be nothing inconsistent in their finding that the guilt of neither had been proved beyond reasonable doubt.

25. I agree with the reasons of King CJ in the Court of Criminal Appeal for concluding that, in the particular circumstances of this case, it cannot be said that a miscarriage of justice resulted from the inadvertent disclosure to the jury that Ms Hay had been in prison. That disclosure, made in the course of examination-in-chief of a prosecution witness, was a serious irregularity. In some cases, such a disclosure would clearly give rise to a situation in which an application for the discharge of a jury would necessarily succeed. In the overall context of the present trial, however, it obviously played no significant part at all in the ultimate verdict and no miscarriage of justice resulted from it.

26. The final separate ground concerns the "accomplice warning" given in relation to Ms Hay's evidence to the extent to which it was "against Mr Webb". On this aspect of the case, I am in general agreement with the judgment of Toohey J. As the New South Wales Court of Criminal Appeal (Gleeson CJ, Campbell and Mathews JJ) observed in Reg. v. Henning ((111) Unreported, 11 May 1990.) in a passage quoted by King CJ in the present case:

"But different principles apply when the supposed accomplice

who gives evidence against a co-accused is himself an

accused giving evidence in his own case. It would be

difficult indeed to seek to apply inflexible rules to such

situations. For the interests of justice will almost

certainly require different responses in different

circumstances. Considerable latitude must be allowed in

order to enable trial judges to address the situation in a

manner which will adapt to the competing interests in the

particular case."

If, in such a case, a trial judge considers it necessary or appropriate to give an "accomplice warning" to protect a co-accused, the critical thing is that it be made clear that the relevant comments relate only to the use of the evidence as against the co-accused. In the present case, his Honour adequately satisfied that requirement when he expressly confined the "accomplice warning" to the use which might be made of Ms Hay's evidence to "convict Mr Webb". It is true that, at an earlier stage of his summing up, the trial judge had given a direction which offended against what was said by this Court in Robinson v. The Queen (No.2) ((112) [1991] HCA 38; (1991) 65 ALJR 644; 102 ALR 493.). The error involved in that direction was, however, adequately corrected by the redirection which his Honour subsequently gave.

27. In each case, the judgment of the Court of Criminal Appeal should be set aside and, in lieu thereof, it should be ordered that the appeal to that court be allowed, the conviction quashed and a new trial ordered.

TOOHEY J The appellants were tried jointly on a charge of murder by judge and jury in the Supreme Court of South Australia. They were both found guilty of the charge and their appeals to the Court of Criminal Appeal were dismissed ((113) Rulings of the trial judge, Debelle J, are reported in Webb and Hay (1992) 64 A Crim R 38. The decision of the Court of Criminal Appeal is reported in Reg. v. Webb and Hay (1992) 59 SASR 563.).

2. In this Court special leave to appeal was granted to the appellant, Webb, "limited to the ground that the learned trial judge ought to have discharged the jury". The notice of appeal filed pursuant to the grant of special leave complains that the Court of Criminal Appeal erred in upholding the trial judge's refusal to discharge the jury and that "given the combined circumstances it ought to have ruled that there was a miscarriage of justice on account of pre-trial publicity". As the argument for Webb developed, in the light of what had been said by the Court on the application for special leave to appeal, the only ground on which it was urged that the trial judge should have discharged the jury was that relating to the "flower incident" discussed later in these reasons.

3. Special leave to appeal was granted to the appellant, Hay, "limited to: one, the ground that the learned trial judge ought to have discharged the jury; two, the correctness of directions given by the learned trial judge as to the assessment of the evidence of the applicant, including the accomplice warning". The notice of appeal filed pursuant to that grant likewise complains of error on the part of the Court of Criminal Appeal in upholding the trial judge's refusal to discharge the jury. It further asserts that the Court should have found that the trial judge misdirected the jury as to the effect of Hay's evidence, that he erred in categorising her evidence as that of an accomplice "and the effects that followed from that direction" and that he further erred "in failing to direct the jury that it was their function to determine whether the applicant was in fact an accomplice". The argument that the trial judge should have discharged the jury centred on the "flower incident" but, as will appear, ranged more widely.

4. It is necessary to say something of the circumstances giving rise to the charge against the appellants and also something of events surrounding their trial.

5. The deceased, Lance Edward Patrick, met the appellants in a hotel in Mount Gambier one evening. The three left the hotel at about 10.25 p.m., taking with them a cask of moselle, and began drinking in a bus shelter. They were still there at 11.50 p.m. Just after midnight Webb was seen walking across a car park adjacent to the bus shelter, wearing the deceased's boots. Hay was in a telephone box across the road. Between 11.50 p.m. and just after midnight a person living nearby heard a loud, strong scream by a male voice, followed by two or three other screams. King CJ, with whose judgment Cox and Matheson JJ agreed, said ((114) ibid. at 566.) that the "overwhelming inference is that the incident which caused the death of the deceased occurred between 11.50 pm and 12.05 am". The appellants spent the rest of the night in a motel room which they entered without permission. At some point Webb deposited the deceased's jacket and belt in a used car yard but he retained the deceased's boots. Webb gave the buckle from the belt to another occupant of the house in which he and Hay were living. He burned the deceased's boots and his own jacket.

6. The deceased's body was found that morning. It had extensive injuries to the face, throat and upper body, consistent with kicks, and injuries to the face, consistent with blows from the buckle of the deceased's belt. The cause of death was a fracture of the cricoid cartilage in the throat, a fracture which could have been caused by blows inflicted by a boot. Footmarks discernible on the throat matched the pattern of the sole of Webb's shoes and, in one case, the pattern of the sole of Hay's shoes.

7. When interrogated by the police, Webb gave several versions of events but he did not give evidence at trial. There was no statement by Hay adduced in evidence but she gave evidence at trial.

8. At this point it is convenient to say something of events surrounding the trial. On 10 June 1992 the appellants entered pleas of not guilty and a jury was empanelled. Following the opening address by the Crown, Webb changed his plea to guilty. The trial judge made an order prohibiting publication of anything referring to Webb's plea of guilty. In addition his Honour discharged the jury. Webb applied for leave to appeal against the conviction entered by reason of his guilty plea but, instead, the trial judge granted an application for leave to change his plea to not guilty. On 23 June 1992 a fresh jury was empanelled from a different jury pool and each juror was asked whether he or she knew anything of the facts involved in the trial. The trial judge refused an application for a change of venue from Mount Gambier and he also refused to order separate trials.

9. On the following day the trial judge gave what counsel described as "a general address to a new jury panel". Such an address, this Court was told, is an invariable practice in South Australia whereby, at the commencement of a sittings, the trial judge addresses the entire jury panel and tells them about the procedure followed in the criminal court, including such matters as the onus of proof, presumption of innocence and the right of silence. The address generally includes a warning to the potential jurors to avoid contact with others and not to discuss the case, except among themselves. Counsel for the appellants relied upon this practice as evidencing the likelihood that all jurors would have been aware of the need to avoid contact with anyone connected with the trial, other than the members of the jury.

10. The Crown case concluded on 24 July. As already mentioned, Webb did not give evidence. Hay gave evidence and her case concluded on 28 July. The Crown addressed on 28 and 29 July. Counsel for Webb addressed on 29 and 30 July. Hay's counsel addressed on 30 and 31 July. The trial judge's summing up began on the afternoon of 31 July.

11. On the morning of 31 July an unusual incident occurred. Before leaving for court, a juror picked two bunches of daffodils from her garden and brought them with her to court. Apparently it was the juror's intention to give them to another juror whose wife was in hospital. While in the courtroom building she spontaneously decided to give one bunch to Mrs Patrick, the mother of the deceased. Mrs Patrick had been in court during the trial and her identity was known. It does not appear that she was a witness. The juror handed the flowers to a woman whom she did not know but who had been in court and who, as it happened, was the mother of the deceased's fiancee. She did so with a request that the woman give the flowers to the mother of the deceased, saying: "Could you give these to Mrs. Patrick? I cannot talk to you."

12. The matter was brought to the attention of the trial judge by the Crown prosecutor. Counsel for each of the appellants asked for a "mistrial". The juror was identified and said:

"I plead guilty. I had beautiful daffodils in my garden.

There was a lady I did not know at the door, I said, 'I

cannot talk to you, those are for Mrs. Patrick', that is

all I said.

...

I humbly apologise to you (his Honour) and the court."

The jury then retired while submissions were made. After they had returned, his Honour asked the juror whether the other jurors had known of her actions. She answered:

"No one knew ... The jury have just said, why did I not

tell them yesterday and they could have told you

yesterday."

13. The trial judge heard submissions from counsel. He said that the question he had to determine was "whether there was a real danger that the position of the accused had been or might have been prejudiced by what had occurred" ((115) (1992) 64 A Crim R at 70.). He ruled that the trial should proceed. He told the jury his reasons for taking this course. Essentially they were that the trial had proceeded for six weeks so that the interests of the people present including Mrs Patrick were apparent, and that the jury had heard a great deal of evidence as to the manner in which the deceased met his death and they had seen photographs of the deceased, so that it might be expected that all jurors would in any event feel sympathy for his family and fiancee. What the juror had done was on the spur of the moment. His Honour stressed again the need for the jury to have regard only to the evidence and to consider it in a dispassionate manner, putting all feelings of sympathy or emotion to one side and weighing the evidence in a dispassionate manner.

14. King CJ endorsed the approach taken by the trial judge in this regard, saying ((116) (1992) 59 SASR at 568.):

"The death of the deceased was bound to excite sympathy for

those to whom he was dear. There is nothing to indicate

that that natural sympathy diverted the jury or any member

of it from a dispassionate consideration of the issue of

the guilt or innocence of the appellants, and there was no

basis for a reasonable suspicion of bias on the part of the

jury or any member of it ... The learned judge gave the

jury a clear direction designed to ensure that feelings of

sympathy would not divert them from their task."

15. Before ruling that the trial should proceed, the trial judge said to counsel:

"Is not the question I have to consider this: The question

is, whether the act of the juror in giving the flowers

shows such a degree of prejudice as would suggest some

expression of bias either towards the prosecution or the

defence and would infringe upon her proper consideration of

the evidence."

Counsel for the appellant submitted that "may" was a more accurate expression than "would" where that word appeared in his Honour's formulation and his Honour said: "That might be so." Counsel for the Crown, although making some response to the trial judge, seems to have accepted the trial judge's formulation as being in conformity with what was said by the Court of Appeal in Bliss ((117) (1986) 84 Cr App R 1.). The situation in England is now governed by Reg. v. Gough ((118) [1993] UKHL 1; (1993) AC 646.), judgment in which was delivered after the decision of the Court of Criminal Appeal in the present case. Lord Goff of Chieveley, with whom the other members of the House of Lords agreed, accepted that in relation to jurors the test was as formulated by the Court of Appeal, namely, whether "there is a real danger of bias affecting the mind of the relevant juror or jurors" ((119) ibid. at 669.). Lord Goff said that there was no practical distinction between this test and a formulation in terms of "real possibility of bias" or "a real likelihood, in the sense of a real possibility, of bias" ((120) ibid. at 668.). His Lordship added ((121) ibid. at 669.):

"Even if the judge decides that it is unnecessary to do more

than issue a warning to the jury or to a particular juror,

and thereby isolate and neutralise any bias that might

otherwise occur, the effect of his warning is not merely to

ensure that the jurors do not allow any possible bias to

affect their minds, but also to prevent any lack of public

confidence in the integrity of the jury."

16. Gough was concerned with an allegation of apparent rather than actual bias on the part of a juror. It is not clear that the challenge in the present case is so restricted and it must be dealt with accordingly. It can be said immediately that there is no evidence of actual bias on the part of the juror. Certainly she went out of her way to express her sympathy for the mother of the deceased. But sympathy was an emotion all members of the jury might have been expected to feel, as the trial judge acknowledged. It did not manifest a view of the guilt of the appellants or either of them or a bias against them. Sympathy is an emotion that jurors will often manifest in the course of a murder trial in the face of testimony, whether oral, written or photographic, bearing on the circumstances in which the deceased met his or her death. Of itself it cannot be treated as an indication of bias against an accused.

17. As to the submission based on apparent bias, there is, as Lord Goff pointed out in Gough, a public aspect involved, a matter of public confidence in the integrity of the jury. But, again, it is necessary to stress that in the present case what was being manifested was sympathy for the mother of the deceased, not hostility to the appellants. This is very different from a case such as Reg. v. Giles ((122) [1959] VicRp 76; (1959) VR 583.), in which a juror, after the Crown outlined the case against the accused who was charged with sexual offences against children, said audibly: "You dirty bastard." The trial judge, after cautioning the jury, allowed the trial to go on. The Full Court declined to interfere with the exercise of the judge's discretion. It is unnecessary to express a view as to the correctness of that decision; but clearly that situation is a far cry from the present one.

18. This Court has dealt with the question of apparent bias on a number of occasions ((123) See, in particular, Reg. v. Watson; Ex parte Armstrong [1976] HCA 39; (1976) 136 CLR 248; Livesey v. New South Wales Bar Association [1983] HCA 17; (1983) 151 CLR 288; Re J.R.L.; Ex parte C.J.L. [1986] HCA 39; (1986) 161 CLR 342; Vakauta v. Kelly [1989] HCA 44; (1989) 167 CLR 568; Grassby v. The Queen [1989] HCA 45; (1989) 168 CLR 1; Laws v. Australian Broadcasting Tribunal [1990] HCA 31; (1990) 170 CLR 70.), though not in relation to the actions of a juror. As Gough makes clear, the underlying principle is the same, whether judges, jurors or members of a tribunal are concerned, though naturally its application will differ in those cases. However, the language in which the test of apparent bias has been expressed in this Court does not accord with that of some English decisions. Thus in Livesey v. New South Wales Bar Association this Court said ((124) (1983) 151 CLR at 293-294.):

" It was common ground between the parties to the present

appeal that the principle to be applied in a case such as

the present is that laid down in the majority judgment in

Reg. v. Watson; Ex parte Armstrong ((125) (1976) 136 CLR at

258-263.). That principle is

that a judge should not sit to hear a case if in all the

circumstances the parties or the public might entertain

a reasonable apprehension that he might not bring an

impartial and unprejudiced mind to the resolution of the

question involved in it."

19. In the majority judgment in Reg. v. Watson; Ex parte Armstrong, Barwick CJ, Gibbs, Stephen and Mason JJ examined a number of English authorities and also earlier decisions of this Court, in particular Reg. v. Australian Stevedoring Industry Board; Ex parte Melbourne Stevedoring Co. Pty. Ltd. ((126) [1953] HCA 22; (1953) 88 CLR 100.) and Reg. v. Commonwealth Conciliation and Arbitration Commission; Ex parte Angliss Group ((127) (1969) 122 CLR 546.). In Australian Stevedoring Industry Board Dixon CJ, Williams, Webb and Fullagar JJ spoke ((128) (1953) 88 CLR at 116.) in terms which the majority in Watson ((129) (1976) 136 CLR at 261.) regarded as requiring "a real likelihood of bias". But, the majority continued, "if doubts were left by that decision as to the correct approach to this question, they were removed by (Angliss Group)". In Angliss Group the Court used language suggesting a test of reasonable apprehension of bias ((130) (1969) 122 CLR at 553-554.). The test of "reasonable apprehension" was applied in Vakauta v. Kelly ((131) (1989) 167 CLR at 573-574, 575, 584-585.) and in Grassby v. The Queen ((132) (1989) 168 CLR at 20.). That test must be regarded as the prevailing test in this country.

20. The test of "real danger" accepted in Gough might suggest a somewhat more rigorous test than that of "reasonable apprehension". If that be so, the judgments in this Court to which reference has been made point to the latter as the yardstick by which to measure apparent bias. But when proper emphasis is placed on the reasonableness of any apprehension likely to arise, that is, apprehension on the part of a fair-minded observer ((133) Livesey (1985) 151 CLR at 300.), there may be in truth little difference in the application of the two tests. In formulating the test as one of "a real likelihood, in the sense of a real possibility, of bias", Lord Goff, as already mentioned, saw no practical distinction between this test and the one adopted, namely, "a real danger of bias". Be that as it may, the test of "reasonable apprehension" is to be applied in the present case.

21. In applying the "real danger" test the trial judge was in error. But when the correct test is applied, the charge of apparent bias in the case of this juror cannot be sustained. There can be no reasonable apprehension of her actions being construed as bias against the appellants or either of them as distinct from sympathy for the mother of the deceased. While that sympathy was manifested in an unusual way, it remained to the fair-minded observer a case of sympathy. Any dangers associated with it were readily capable of being avoided by the express directions the trial judge gave to the jury. And the juror's own reaction, her apology and expression of concern, points to the likelihood that she, as well as the other jurors, would observe those directions.

23. The appellant, Hay, relied upon the "flower incident" as justification for the argument that the trial judge should have discharged the jury. To that extent her appeal must fail. But she relied also upon some other matters which, together with that incident, were said to warrant discharge.

24. The first of these other matters is the failure of the trial judge to order separate trials for the appellants. The justification, indeed the alleged necessity, for separate trials lies in the fact that in three records of interview with Webb, which could be expected to be and were adduced in evidence by the prosecution, Webb made assertions that Hay had engaged in a violent and sadistic attack on the deceased. And, it was said, although the trial judge warned the jury that this evidence was not admissible against Hay, such a direction could not cure the overwhelming prejudice inevitably caused to Hay.

25. King CJ dealt with this ground by pointing out that there are "strong reasons of principle and policy why persons charged with committing an offence jointly ought to be tried together. That is particularly so where each seeks to cast the blame on the other" ((134) (1992) 59 SASR at 585.). What King CJ referred to as "strong reasons of principle and policy" were discussed by his Honour in Reg. v. Collie ((135) [1991] SASC 2996; (1991) 56 SASR 302 at 307-311.). I respectfully agree with that discussion which emphasises that when accused are charged with committing a crime jointly, prima facie there should be a joint trial. There are administrative factors pointing in that direction but, more importantly, consideration by the same jury at the same trial is likely to avoid inconsistent verdicts, particularly when each accused tries to cast the blame on the other or others ((136) Reg. v. Demirok [1976] VicRp 19; (1976) VR 244 at 254.). There are of course dangers for an accused in a joint trial by reason of the admission of evidence which would not be admitted at the trial of one accused. That risk must be obviated by express and careful directions to the jury as to the use they may make of the evidence so far as it concerns each accused ((137) Reg. v. Harbach (1973) 6 SASR 427 at 433.).

26. In the end the critical question before an appellate court in these circumstances is whether, by reason of the joint trial, there has been a substantial miscarriage of justice or, put another way, whether improper prejudice has been created against an accused.

27. In the present case adequate directions were given by the trial judge. It is true that Webb did not give evidence and was therefore not subject to cross-examination, though the jury heard his statements to the police implicating Hay. But, as King CJ observed ((138) (1992) 59 SASR at 585.):

"That is a common feature of a joint trial and does not of

itself render separate trials necessary."

Properly instructed by the trial judge, as they were, the jury were capable of appreciating the use they could make of evidence as against each of the appellants. It has not been shown that a substantial miscarriage of justice is likely to have occurred.

28. A further ground upon which discharge of the jury was sought was the disclosure to the jury that Hay had been in prison. The disclosure was inadvertent and was made in the course of the examination-in-chief of a prosecution witness. The witness was asked when he or she first met Hay; the following ensued:

"A. I wouldn't know exactly what date but it was when she

was doing some time in prison, I used to go and visit.

Q. Had you known her for about a year or longer.

A. Well, at that time I had only just met her through

Cindy when she was in gaol."

The trial judge refused an application by Hay's counsel to discharge the jury. His Honour gave a direction to the jury which King CJ described as "an entirely clear and appropriate direction designed to remove any prejudicial effect of the disclosure" ((139) ibid. at 578.). The matter, King CJ said, was one for the discretion of the trial judge and he correctly exercised that discretion. I agree with his Honour's conclusion.

29. Where evidence prejudicial to an accused is elicited inadvertently, it is a matter for the trial judge to decide whether the jury should be discharged. If the jury are not discharged, on appeal the question for determination is as mentioned earlier in these reasons. That is, the appeal "is not against the failure to discharge the jury but against the conviction" ((140) Maric v. The Queen (1978) 52 ALJR 631 at 634; 20 ALR 513 at 520 per Gibbs J). The question then is whether a substantial miscarriage of justice has occurred.

30. In Reg. v. Glennon Mason CJ and I said ((141) [1992] HCA 16; (1992) 173 CLR 592 at 604.):

"Reception of inadmissible evidence of a prior conviction

has been said to offend against one of the most deeply

rooted and jealously guarded principles of our criminal

law."

Glennon was concerned with a case of pre-trial publicity as a result of which a permanent stay of proceedings was sought. That is not to say that the passage is inapposite in the present case. But, again, the question on appeal is whether, in the present case, Hay suffered a substantial miscarriage of justice by reason of the inadvertent disclosure that she had been in prison.

31. Nothing was said at trial of the offence which resulted in Hay's imprisonment. Again I agree with King CJ when he concluded ((142) (1992) 59 SASR at 578.):

"It would have come as no surprise to the jury, having

regard to the evidence of her drinking and conduct on the

present occasion, that she had had a brush with the law

resulting in imprisonment. Her admitted behaviour on the

present occasion would have reduced any prejudice arising

from the disclosure that she had been in gaol previously,

into insignificance."

32. The next complaint by Hay related to the correctness of directions given by the trial judge as to the evaluation by the jury of her evidence. In dealing with this complaint, it must be remembered that, while Hay gave evidence, Webb did not. The crux of his Honour's direction on this aspect may be found in the following passage:

"You will bear in mind that any person in the position of

the accused, and I am speaking quite generally now, any

person in the position of the accused will obviously be

under a strong temptation to consider his or her own

interest exclusively and, if need be, to play down his or

her own part in the matter, if need be, at the expense of

the co-accused. So you must bear in mind the possibility

of that kind of distortion in a trial of this kind even to

the point of deliberately false evidence. However, it is

necessary (sic) for me to say more on the subject than that

general observation about evidence that any co-accused

might give in a trial in which more than one person is

jointly charged before a jury."

33. This direction produced a request for a redirection which his Honour gave the following day in these terms:

"One other thing I wish to say just to backtrack a little

on what I said last night. Towards the close last night

almost at the very close, I suggested to you that a

co-accused who gives evidence might seek to play down his

involvement in a matter to the point of giving false

evidence. What I said then was a general warning only. I

was not intending, in any way, to suggest that Ms Hay was

a suspect witness. I did not, in any way, seek to suggest

that you subject her evidence to any different kind of

scrutiny from that which you would apply to the evidence of

any other witness. My purpose was to lead into the topic

I'm now about to discuss.

The topic on which the law does require me to give a

warning, that is to say, give a warning about the evidence

of a witness who is an accomplice. I emphasize indeed what

I have just been saying and what I said at an early stage

in the course of my summing up, that you should examine and

test the evidence of Ms Hay in the same way as you test the

evidence of other witnesses. You should deal with her

evidence just as you would deal with any evidence of any

other witness."

34. King CJ considered that there had been no infringement by the trial judge of the principles enunciated by this Court in Robinson v. The Queen (No.2) ((143) [1991] HCA 38; (1991) 65 ALJR 644; 102 ALR 493.) regarding a direction to a jury as to how they should treat the evidence of an accused. There can be little doubt that the initial direction offended against what was said in Robinson. The question is whether that direction, coupled with the redirection, is susceptible to challenge.

35. Counsel for Hay stressed that her evidence was lengthy and that, even with the redirection, the jury must have been led to conclude that her evidence should be scrutinised more carefully than that of any other witness. All this may be true but the situation was one in which, unlike in Robinson, each appellant sought to blame the other and thereby exculpate himself or herself. The trial judge's earlier remarks must be understood in this light. While capable themselves of leading to a misunderstanding by the jury, the remarks were sufficiently balanced by his Honour's later direction. He had to deal with, and therefore say something about, a situation in which each appellant was trying to place the responsibility for the deceased's death on the other.

36. Hay's role in the killing of the deceased leads on to another challenge to the decision of the Court of Criminal Appeal. Hay's evidence implicated Webb in the killing and, although admitting her own participation in violence against the deceased, she claimed that Webb had already struck the fatal blow. In those circumstances the trial judge gave an "accomplice direction". Hay's complaint is that such a direction should not have been given as it was prejudicial to her defence, alternatively that the trial judge should have left to the jury the question whether she was an accomplice.

37. His Honour spent some time speaking of the way in which the law regards the evidence of accomplices, including the care with which their evidence should be scrutinized, in particular if it is uncorroborated. He said that on one view of the evidence Webb or Hay could be regarded as having been an accomplice to murder but he then went further and said:

"Without expressing any view about the guilt or innocence of

the two accused, I'm not concerned with that question now,

I direct you that each of them should be regarded, in law,

as accomplice for the purpose of this special warning.

Ms Hay has given evidence which, to some degree, is against

Mr Webb. You should not convict Mr Webb on the evidence of

Ms Hay unless you find the evidence is corroborated, or

unless, after you have given it very careful consideration

in the light of the warning that I am now giving, you are

convinced it is reliable. Just to put that same matter in

other terms. If the evidence that Ms Hay gives against

Mr Webb is not corroborated, you should not take it into

account against the other accused unless you are convinced

of its reliability."

38. His Honour told the jury: "Whether the evidence of Ms Hay against Mr Webb is corroborated is a matter for you to decide." He then identified to the jury parts of Hay's evidence relating to Webb that were corroborated, namely, that Webb was at the scene and that he kicked the deceased on a number of occasions. He also said that in so far as Hay's denial that she used the belt buckle on the deceased implied that Webb had done so, that was not corroborated.

39. As King CJ pointed out ((144) (1992) 59 SASR at 581. See Davies v. Director of Public Prosecutions (1954) AC 378 at 399.):

"In the classic formulation of the rule requiring a warning

that it is dangerous to convict on the uncorroborated

evidence of an accomplice, the requirement is confined to

the evidence of a witness for the prosecution."

40. There are several English decisions to the effect that a full corroboration warning is not mandatory where the witness is a co-accused ((145) R. v. Barnes (1940) 2 All ER 229; Reg. v. Prater (1960) 2 QB 464; Reg. v. Stannard (1965) 2 QB 1; Reg. v. Bagley (1980) Crim LR 572; Loveridge (1982) 76 Cr App R 125; Knowlden (1981) 77 Cr App R 94; Reg. v. Cheema (1994) 1 WLR 147.). There is no common approach to be discerned in the decisions of Australian courts. The full corroboration warning in the case of a co-accused is required in Victoria ((146) Reg. v. Teitler [1959] VicRp 54; (1959) VR 321.) and in Queensland ((147) Reg. v. Allen and Edwards (1973) Qd R 395.). The warning is regarded as discretionary in New South Wales ((148) Reg. v. Henning, unreported, Court of Criminal Appeal, 11 May 1990.). In South Australia itself views have varied. In Reg. v. Rigney Bray CJ, expressing a preference for Reg. v. Teitler ((149) [1959] VicRp 54; (1959) VR 321.), said ((150) (1975) 12 SASR 30 at 40.):

" There is no doubt that if there is evidence on which

a reasonable jury could find that the witness was a

'participant' ... the issue of accomplice or not should be

left to the jury."

Hogarth J favoured a discretion in the trial judge to exercise a discretion whether to give such a warning; his Honour disagreed with Teitler ((151) ibid. at 53-54.). In Reg. v. Wilson ((152) (1987) 47 SASR 287.), where there were four co-accused, White J took the view that he had such a discretion and directed the jury, not by reference to accomplices, but by alerting the jury to the possibility, indeed likelihood, that each accused had a strong reason for blaming the others and exculpating himself.

41. In the present case King CJ preferred the approach taken by the Court of Criminal Appeal in Henning which is encapsulated in this passage from its judgment ((153) Henning, unreported, 11 May 1990 at 47.):

"But different principles apply when the supposed accomplice

who gives evidence against a co-accused is himself an

accused giving evidence in his own case. It would be

difficult indeed to seek to apply inflexible rules to

such situations. For the interests of justice will

almost certainly require different responses in different

circumstances. Considerable latitude must be allowed in

order to enable trial judges to address the situation in a

manner which will adapt to the competing interests in the

particular case."

42. There is already a strong opinion that the law of corroboration has become unduly and unnecessarily complex and technical ((154) Reg. v. Cheema (1994) 1 WLR at 158.). In Vetrovec v. The Queen ((155) (1982) 136 DLR (3d) 89.) the Supreme Court of Canada held that it is no longer a rule of law that there is a special category for accomplices requiring a special warning that it is dangerous to act on their uncorroborated evidence. Dickson J, delivering the judgment of the Court, said ((156) ibid. at 99.):

" None of these arguments can justify a fixed and

invariable rule regarding all accomplices. All that can

be established is that the testimony of some accomplices

may be untrustworthy. But this can be said of many other

categories of witness. There is nothing inherent in the

evidence of an accomplice which automatically renders him

untrustworthy. To construct a universal rule singling out

accomplices, then, is to fasten upon this branch of the

law of evidence a blind and empty formalism. Rather than

attempting to pigeon-hole a witness into a category and

then recite a ritualistic incantation, the trial judge

might better direct his mind to the facts of the case, and

thoroughly examine all the factors which might impair the

worth of a particular witness. If, in his judgment, the

credit of the witness is such that the jury should be

cautioned, then he may instruct accordingly. If, on the

other hand, he believes the witness to be trustworthy,

then, regardless of whether the witness is technically an

'accomplice' no warning is necessary."

43. The question here is a somewhat narrower one but the reasoning in Vetrovec reinforces the need for flexibility in the direction which a trial judge gives when co-accused blame each other. The approach taken in Henning and by King CJ in the present case gives effect to this need. It follows that in the present case the trial judge was not obliged to give the accomplice corroboration warning but that he was not in error in doing so. The problem is whether, in so directing the jury, his Honour placed the evidence of Hay in a disadvantaged position. He had to maintain a balance between the interests of Webb on the one hand and Hay on the other. This ground of appeal can only succeed if it is shown that his Honour failed to maintain that balance and that, as a result, Hay suffered a substantial miscarriage of justice.

44. I am not persuaded that she did suffer a substantial miscarriage of justice. His Honour had to say something about the use of Hay's evidence against Webb. This he did. The earlier direction, to which reference has been made, went further and tended to focus unduly on Hay's playing down of her own role in the killing of the deceased. But that was corrected by redirection. Overall, the jury were sufficiently alerted to how they should regard Hay's evidence both as it bore on her own defence and as it implicated Webb.

45. Clearly Hay was, on her own evidence, an accomplice to the killing of the deceased. It is hard to see how she could have been unduly prejudiced by his Honour's invitation to the jury to treat her as such. In all the circumstances this was probably a preferable course to instructing the jury that, in relation to Hay's evidence implicating Webb, they should first determine whether she was an accomplice and then, in relation to the case against her, decide on the criminal onus of proof whether she was guilty of murder. However, it would have been better still to avoid any reference to accomplice and deal with the strengths and weaknesses of the evidence generally.