At the time of his death, Trevor Braun was living in a caravan at a caravan park at Denham Court and the accused lived in another caravan in the same park. Also living in a van in that park was Mr Colin Ludlow. Evidence obtained by police at the time disclosed that, on 26 July 1994, the deceased came home from work early, feeling unwell. He was on medication which made him drowsy. In the early evening the accused and Mr Ludlow went to the deceased's van to check that he was alright. They found him asleep, but his blanket had come into contact with an electric bar heater and was smouldering. Mr Ludlow extinguished the blanket and disconnected the heater, and he and the accused left the caravan. At about 9.00 pm they returned to the van but left again, having found the deceased asleep and the heater still disconnected. Some time after 10.30 pm the accused heard the deceased screaming and saw that his van was on fire. She ran to Mr Ludlow's van to raise the alarm, but when they and others got to the deceased's van, the fire was impenetrable and the deceased could not be rescued.

This was the account given at the time, both by the accused and Mr Ludlow. The death was accepted by the authorities as accidental, and the investigation of the accused's complicity in it was not undertaken until almost a year later in circumstances to which I shall turn in a moment. One of the witnesses who made a statement at the time was Ms Margaret Wardle, who owned the caravan park with her husband. When the investigation was reopened she was interviewed again and, in a statement of 14 June 1995, she told police that after the fire was extinguished she saw the accused crying uncontrollably and heard her say, "It's my fault, it's my fault, I put the heater back on because he was cold." This is the first of the admissions to which objection has been taken.

Observations of the scene after the fire suggested that the heater had been connected to the power point, and may well have caused the fire. Nevertheless, the fire could still have been accidental and, clearly, the accused's statement to Ms Wardle falls short of a confession of homicide. However, there is evidence that the accused admitted having deliberately caused the fire to her general practitioner, Dr Ng, in November 1994; to members of the staff of Waratah House, the psychiatric unit at Campbelltown Hospital, in June 1995; and to police, in an electronically recorded interview in that same month. All this evidence is objected to. To determine its admissibility, it is necessary to examine the circumstances in which the admissions were made and this, in turn, requires an understanding of the accused's personal background and psychiatric history.

The accused is now 22 years old and is the youngest of five children. Her two brothers, one of them the deceased, were from her mother's first marriage, and she has two sisters who are the product of her parents' union. Her father is a Vietnam veteran who suffers from a nervous disorder. Her mother has agoraphobia. Some of her siblings, including the deceased, have had problems with alcohol dependence. One of her sisters has suffered from post natal depression since the birth of her second child. Perhaps not surprisingly in the light of this background, she herself has exhibited behavioural problems since her childhood. Her schooling was marked by disruptive behaviour. She was a poor student and left at the age of 15 at the end of year nine. When she was 13 years old, she underwent a psychological assessment of her intellectual functioning at Rivendell Adolescent Unit and was found to be within the borderline mentally deficient range.

She has a history of admissions to psychiatric units at Campbelltown and Liverpool since 1991, on most occasions after apparent suicide attempts or threats of suicide. Various diagnoses were made at the time of these admissions, but I have had the benefit of the evidence of her general practitioner, Dr Ng, who has now qualified as a Master of Psychological Medicine, and three psychiatrists, Drs Skinner, Phillips and Neilssen, and a psychologist, Mr Guilfoyle, who examined her for the purpose of these proceedings. All these experts agreed that she has suffered from a severe borderline personality disorder which is characterised, among other things, by attention seeking behaviour born of a fear of real or imagined abandonment. It is consistent with this disorder that she has also a history of telling false stories about herself, apparently to elicit people's sympathy or, at least, to attract their attention. Over a period of years she has claimed to have been raped by several men, including her late brother, and to have herself sexually interfered with a small boy. At one stage she claimed to have sat on a needle and to have contracted AIDS, and she continued to tell her sister that she expected to die of the disease after a blood test proved negative. These stories were either inherently improbable, or were shown by her subsequent words and actions to be almost certainly untrue. For the most part, this behaviour - both the suicide attempts and the false claims of some misfortune having befallen her - occurred at the time of some significant family event which, apparently, she saw as diverting attention from her.

In September 1994, about two months after her brother's death, the accused was admitted to the Macquarie Clinic, the psychiatric unit at Liverpool Hospital, after she had been found attempting to walk in front of cars on the Hume Highway. She told staff that she was experiencing flashbacks of the scene of her brother's death, and that she wanted to die to be with him. She was admitted as a voluntary patient and the following day was taken home by her parents. She had been treated by Dr Ng on and off since 1989, and he saw her on 16 November 1994. During this consultation she told the doctor that she dreamt about her brother's death, and said that she felt guilty because she could have done something to help him. On the morning of 23 November, she told him that she could have taken the heater away from the deceased, adding that she had wanted to but he did not want it moved. However, on the afternoon of that same day, she saw the doctor again and told him that she had killed her brother by putting blankets over the heater. Given her history of telling false stories, Dr Ng did not believe this and, after consulting a colleague and taking legal advice, he did not report the matter.

In the evening of 29 May 1995, the accused was admitted to Waratah House at Campbelltown after having tried to hang herself. Her history of this attempt was corroborated by injuries observed on physical examination. She told staff on admission that her desire to end her life arose from her ineradicable memories of the death of her brother, together with other factors in her life at the time, which it is unnecessary to recount. Over the ensuing fortnight she told a number of the hospital staff that she deliberately caused the fire in which her brother died by going back to his caravan alone, reconnecting the heater, and placing a blanket over it. It is unnecessary to set out in detail what she said on each of these occasions. It is sufficient to say that her various accounts of what she did were somewhat laconic, and yet not entirely consistent. However, on other occasions she retracted this confession and gave an account consistent with the fire having been an accident (whilst still expressing a sense of guilt that he might not have died if she had been more attentive or resourceful). At times she appeared to be confused about what she had said to whom, and unaware of clear inconsistency between her various statements. Some days after her admission to the hospital, police were notified of what she had been saying and staff were requested to make detailed notes of any statements she might then make touching on her brother's death.

Whilst in the hospital, the accused told her parents on several occasions that she had caused the fire, although again she retracted this on one occasion. On 5 June, a nurse understood the accused to have made a similar admission in a conversation which she overheard between the accused and a Legal Aid solicitor who was to represent her in a hearing under the Mental Health Act.

On 30 May and 1 June, she was certified by medical staff as both mentally ill and mentally disordered, within the meaning of the Mental Health Act 1990. The risk of her harming herself was considered to be such that she was classified as a Category A patient, which required a nurse to be within arm's reach of her at all times. Her demeanour in the first days at the hospital was consistent with a very disturbed state of mind. Later her behaviour settled, but on two further occasions, she behaved in a violent and irrational way and needed to be restrained, on one of those occasions during an attempt to abscond from the hospital.

On 5 June, a magistrate ordered that she be detained as an involuntary patient for one month. However, on 15 June she was discharged into the custody of the police. There was still considered to be a significant risk of self harm, such that the police kept her handcuffed for her own safety until her appearance at Campbelltown Local Court that afternoon. The precaution was well advised, because she attempted to strangle herself with a sock in the court cells after her appearance.

On that morning, Detectives McDonald and Grant escorted the accused from the hospital to Campbelltown Police Station, where she participated in an electronically recorded interview for about an hour, commencing a little after 11.20 am. In the course of that interview she made admissions of having caused the fire, to much the same effect as those she had made to hospital staff, although she added to the account of her movements a matter of some significance which she had never mentioned before. (This appears in the answers to questions 152 and 153, but for the present purposes no more need be said about it.) She was later charged with murder and was taken before a magistrate after 2.00 pm that day.

In her conversations with the hospital staff, the accused said on one occasion that she had had an argument with her brother on the afternoon before he died (although she did not elaborate upon this) and, on another occasion, said that she had caused the fire, not because her brother had raped her, but because of the anxiety which his lifestyle had been causing to her parents. However, to the police, she said nothing about either of these things and maintained that her relationship with her brother had been good. She said that her earlier claim that he had raped her was false, and she was unable to describe in any way her state of mind at the relevant time.

For obvious reasons, expert evidence has loomed large in this voir dire examination. I could not say that I was uniformly impressed by the opinions of the medical and nursing staff from Waratah House who gave evidence, and I derived considerably more assistance from the evidence of Dr Ng and of the psychiatrists and psychologists engaged for forensic purposes. I should say that I was particularly impressed by the practical and insightful evidence of Dr Ng who, of course, has the advantage of longstanding association with the accused.

Mr Guilfoyle, psychologist, tested her intellectual functioning and arrived independently at the same conclusion as the testing some years before at Rivendell Adolescent Unit, that is, that she is operating at the borderline mentally deficient level. His personality testing confirmed the diagnosis of borderline personality disorder. The forensic psychiatrists who examined her did not accept without reservation that she is as intellectually impaired as the test results would suggest but, nevertheless, were agreed that she is of low intelligence. All the experts were of the opinion that her confession could be the product of her personality disorder and be, in fact, false, although Dr Skinner was more guarded about this than the others. They explained that the accused's disorder is such as to cause a blurring between reality and fantasy and, for this reason, Dr Phillips referred to her history of false stories as one of telling "untruths", rather than "lies". In evidence he said that he considered her confessional statements to be of "doubtful reliability or validity". Dr Neilssen, in his report, concluded that he "would advise caution in seeking to convict Ms Braun on the strength of her confession alone." All were of the view that her confession, if it were false, might have been influenced by her sense of guilt that she did not do more to prevent her brother's death.

Dr Ng shared these views and expanded upon the interaction of her intellectual deficits with her personality disorder. He agreed that her confession was different from other false stories she had told (with the possible exception of her claim that she had sexually assaulted a boy) in that they were claims of misfortune befalling her, whereas the confession was an admission of wrongdoing on her part. However, he explained that she may have implicated herself in her brother's death to gain the attention which she undoubtedly did receive but, because of her low intelligence, may not have appreciated the serious implications for herself of what she was saying. As it happens, she spent about a month in custody following her arrest and, when she saw him after her release, she said that her confession was false. (She also repudiated her confession to the three forensic psychiatrists and to Mr Guilfoyle although, even then, her account to each of them of what did occur was not entirely consistent.) Dr Ng considered that her experience of custody had awoken her to the consequences of what she had said and had accelerated a process of maturation which would, in any event, have occurred as she got older.

The foregoing is a brief but, I hope, adequate summary of the considerable volume of evidence led on the voir dire. I have been ably assisted by the submissions, both oral and in writing, of counsel for the Crown and the accused. I am particularly indebted to the Crown prosecutor for the thoroughness, realism and fairness with which he has presented his arguments.

Objection to the evidence of Ms Wardle, that the accused told her on the night of the fire that she had turned the heater back on, was but faintly pressed. Clearly, the accused's statement is relevant although, standing alone, it would be insufficient to establish her complicity in her brother's death. Nor has it been suggested that there is anything in the circumstances in which it was made militating against its admissibility. However, this is the only part of the challenged evidence which I would admit.

The Crown prosecutor did not press the admission of this evidence. Quite apart from the circumstances at the time these admissions were made (to which I shall turn in a moment), he foreshadowed that both parents would exercise their right under s 18 of the Evidence Act, 1995 to object to giving evidence against their daughter, and acknowledged that he would have great difficulty in persuading me that they should be compelled to do so.

Objection to this evidence was based upon s 90 of the Evidence Act, conferring a discretion upon me to reject it if, having regard to the circumstances in which the admission was made, it would be unfair to the accused to admit it. That discretion is relevantly similar to the common law discretion to exclude confessional evidence in the interests of a fair trial expounded in cases such as The King v Lee [1950] HCA 25; (1950) 82 CLR 133. The reliability of the admission is a matter properly to be considered and, particularly in the case of mentally disturbed people, the circumstances in which the admission was made might call for the exercise of the discretion, even though there was no impropriety on the part of the person to whom it was made: cf R v Parker (1989) 19 NSWLR 177, per Gleeson CJ at 183.

I am satisfied that it would be unfair to the accused to allow the evidence of this admission. I share Dr Ng's scepticism about its reliability. It is no more than a bare assertion that she had started the fire by putting blankets over the heater and, given that she had undoubtedly gone to the caravan earlier with Mr Ludlow and seen a blanket smouldering after contact with the heater, it is a statement which she could easily have fabricated. It was made at a time when she was expressing feelings of guilt that she might have been able to prevent her brother's death, and two months after the experience of that death had engendered in her the bizarre behaviour which led to her admission to the Macquarie Clinic at Liverpool. The effect of those circumstances upon the reliability of what she said, against the background of her personality disorder and her history of false stories, is such that the evidence should be rejected.

The same must be said of this evidence. For the most part these admissions were made while the accused was clearly in a disturbed state of mind after what may well have been a serious suicide attempt. Quite apart from her admissions of causing the fire, the accused continued to voice a sense of moral responsibility for her brother's death (even if it had been an accident). As I earlier observed, these admissions are laconic, although more detailed than anything she said to Dr Ng, and they also could easily have been fabricated. As the Crown prosecutor fairly observed, their "scope and detail" varied, and on more than one occasion she recanted them.

The Crown prosecutor distinguished admissions made to the hospital staff before the involvement of police from those made later, after the police had requested staff to maintain detailed notes of what she said. As to the latter, he acknowledged that the staff might be seen to have induced the accused to speak of her involvement in her brother's death at a time when she was psychiatrically ill, and had not been warned that whatever she said might be used in evidence against her. That concession was properly made, but I remain of the view that fairness to the accused dictates the rejection of the whole of this evidence.

The Crown prosecutor did not press the evidence of the conversation overheard between the accused and her solicitor prior to the magistrate's hearing. The nurse's recollection of that conversation does not accord with that of the solicitor, who gave evidence before me. In any event, it was clearly intended to be a confidential communication between solicitor and client, and the nurse overheard it only because the accused's status as a Category A patient did not permit her to be out of earshot. The unfairness of admitting this evidence is obvious.

Of course, the very same circumstances affect these admissions made at a police station immediately after her discharge from Waratah House, while she was handcuffed for her own safety and still considered at risk of harming herself. It is not to the point that the video tape of the interview reveals that the police officers treated her courteously and that she appeared to understand what was being asked of her and to respond appropriately. I have not found it necessary to resort to s 90 in determining the admissibility of this evidence. These admissions were made in the course of "official questioning", within the meaning of s 85 of the Evidence Act, and by sub-s (2) of that section evidence of them is inadmissible unless I am satisfied that the circumstances in which they were made were such as to make it unlikely that the truth of them was adversely affected. In the light of the whole of the evidence, and having regard to the matters which I am required by sub-s (3) of that section to take into account, I am not so satisfied.

Again, with commendable fairness, the Crown prosecutor did not press the admission of this evidence. Indeed, in written submissions, he acknowledged that, given that the accused's personality disorder manifested itself in attention seeking behaviour, "the very nature of the official questioning process might be likely to exacerbate her tendency to confabulate".

Counsel for the accused also submitted that the police evidence should be rejected on the grounds of public policy embraced in s 138 of the Evidence Act, in the light of behaviour on their part which was either unlawful or improper and, to an extent, the Crown prosecutor acknowledged the force of that submission. As I have determined that the evidence is inadmissible in any event, it is unnecessary to express any concluded view about this argument. However, one thing does need to be said. It is clear that, when the police escorted the accused from the hospital on 15 June 1995, she was under arrest for the murder of her brother, although she was not told that in so many words. Det McDonald's evidence to that effect was disarmingly frank, and Det Grant's evidence to the contrary was, to say the least, disingenuous. That being so, her detention at the police station for the purpose of questioning was clearly unlawful: Williams v The Queen (1986) 161 CLR 287. It is a matter of grave concern that nine years after that High Court decision, police in this state should still feel free to flaunt the clear statements of principle of the majority of the Court in the interests of what they perceive to be sound investigative practice. Given that s 138 now casts the burden upon the Crown to show that confessional evidence should be admitted once it has been established that it was obtained illegally, it may be that the police evidence here should have been rejected on that basis alone.

I have arrived at this conclusion about the evidence, well aware that the truth of a confession is normally a matter for the jury to determine. A judge does not lightly reject evidence which, on its face, is relevant and probative of guilt, particularly in a case as serious as this: cf Collins v R [1980] FCA 72; (1980) 31 ALR 257, per Brennan J (as his Honour then was), at 314. On the other hand, it is the responsibility of the trial judge to do so where the interests of justice require it, in accordance with well established principles developed at common law and now enshrined (and to some extent, modified) in the Evidence Act. This is such a case.