SUPREME COURT OF QUEENSLAND

1. McMURDO P:   I agree with the reasons of Williams JA and with the orders proposed.

2. WILLIAMS JA:   On 14 April 1973 the body of a baby, Deidre Kennedy, was found on the roof of a toilet block in a park in Ipswich a few hundred yards from where the Kennedy family was living.  Some of the baby's clothing had been removed and replaced with adult clothing (a slip, a pair of panties and a step-in) which had been taken from a clothes line on a property next door to the Kennedy residence.

3. A post-mortem was carried out by Dr O'Reilly (he had died well before 1985) which established strangulation as the cause of death. During the post-mortem bruises were noted on the lower part of the baby's left thigh; photographs of those bruises were taken.

4. Dr Romaniuk (then Senior Lecturer in oral biology and oral survey with the University of Queensland and a person with considerable experience in forensic odontology) examined the body of the deceased baby and noted "small bruises in the connective tissue"; he described them as bite marks. He obtained prints of the photographs taken at the post-mortem and also obtained models of certain sets of teeth. He signed a statement on 8 October 1973 in which he said:

"Comparisons of these models against that of the bite mark on the dead child were inconclusive because of the meagre detail in the bite mark on the thigh of the child".

Apparently he discussed the matter with a panel of 10 eminent dentists and his statement also contained the following passages:

"In final analysis it was agreed between myself and these 10 men that it would be impossible to establish, with any degree of certainty, as to who would be responsible for this bite mark to the dead child."

"As I have stated previously, the individual bruises caused by the bite are not sufficiently defined to enable one to arrive at any definite conclusion".

5. Extensive police investigation in the months after the finding of the body failed to result in any person being charged.

6. On 11 October 1983 the appellant was first interviewed by the police about the death of the baby. He maintained that in April 1973 he was a member of an RAAF course at the Edinburgh Base in South Australia, that course running from 9 February 1973 to 19 April 1973.  He made no incriminating admissions to the police.  He was arrested for the murder of Deidre Kennedy on 24 February 1984, and his trial commenced in the Supreme Court on 18 February 1985.  The prosecution case against him could be summarised as follows:

1. Opportunity;  he was not, as he claimed, on the RAAF Base at Edinburgh at the time of the murder;

2. He had a propensity for biting small children on the legs;

3. Odontological evidence established that the bite mark on the baby's thigh was caused by his teeth.

The appellant gave evidence at the trial in the course of which in answer to the question "Did you kill Deidre Kennedy?" he swore on oath "I did not".  On 14 March 1985 the jury returned a verdict of guilty of murder.

7. The appellant appealed to the Court of Criminal Appeal and on 27 November 1985 his appeal was unanimously allowed; it was ordered that the conviction be quashed and that a verdict of acquittal be entered. The consequences were the same as if the jury had returned a verdict of not guilty. The reasons for that court so determining are reported at (1985) 19 A Crim R 410. In broad terms the Court of Criminal Appeal dealt with the three major planks of the prosecution case as follows:

1. A reasonable jury could have been satisfied beyond reasonable doubt on the evidence from the Crown witnesses that the appellant was not on the RAAF Base at Edinburgh at the relevant time, but that did not necessarily place him in or near Ipswich.  There was no direct evidence that the appellant was at or near Ipswich at the relevant time.  The appellant's Record of Leave (tendered on the perjury trial as Ex. 25) does not indicate he had any leave in April 1973, but in the light of all the evidence that is probably not conclusive.  There is fairly clear evidence he was on the rifle range on 12 April 1973:  (see Ex. 27, 28 and 30 on the perjury trial).  It should also be noted that prior to joining the Air Force on 5 February 1973 he resided with his mother at East Ipswich;  his mother remained living there for some time after he left;

2. On the then relevant authorities (R v Perry (1982) 150 CLR 580 and R v Sutton (1984) 152 CLR 528) the similar fact (propensity) evidence was not sufficiently probative to outweigh its prejudicial effects and ought to have been excluded;

3. A jury, acting reasonably, must have entertained a reasonable doubt as to guilt based on the odontological evidence.  The court reviewed extensively the evidence from the three experts called by the prosecution:  Dr Romaniuk, Mr Sims, and Mr Brown. 

The verdict of acquittal was entered because the Court concluded that there was no reliable evidence led by the prosecution identifying the appellant as the killer.

8. The state of odontological evidence at the murder trial can be ascertained from a reading of the judgment of the Court of Criminal Appeal and also from evidence given at the perjury trial. The following is a summary of that evidence at the murder trial and relevant observations made thereon in the Court of Criminal Appeal:

1. Dr Romaniuk, Mr Sims and Mr Brown each had considerable experience in forensic odontology;  some detail of their expertise is set out by Kneipp J at 414 and Shepherdson J at 421;

2. Dr Sims in particular, and the others to lesser extent, conceded there was "a body of eminent opinion which holds that valid identification cannot be made by reference only to bruise marks or that they should be referred to only for the purpose of excluding suspects and not for positive identification";

3. the effect of the evidence of all three was that one had to have serious reservations as to the reliability of identification where there are bruise marks only.  As noted by Kneipp J at 415:

"Mr Sims candidly agreed that there was a body of opinion, held by experts whom he would not rate as being inferior in skill to himself, to the effect that identification of teeth by reference to bruise marks is not reliable".

Sims in his evidence referred to a number of problems relevant to that issue:

"Some related to the difficulty of accurate reproduction of the bruising.  A photograph of part of a human body results in reproduction in two dimensions of a three-dimensional object, with possible distortion.  The bruise marks in question were on a curved surface, and this might also cause distortion.  The shapes of bruise marks can alter if the body is placed in a different position from that in which it was when the marks were made.  The marks here were near the knee, and their shapes might be affected by changes in tensions of the skin caused by extension or flexion of the knee joint"; (415)

4. here there was no laceration nor any indentation – there was bruising only;

5. the identification process here was complicated because the accused had had some work done to his teeth in 1976.  It was therefore necessary for the odontologists to reproduce the teeth in the condition they were believed to be in at the time of the death of the baby for comparison purposes;

6. each of the three odontologists was in agreement that the row of bruises in the upper part of the relevant photograph were left by the top row of teeth;

7. the three witnesses spoke of 4 upper teeth and 4 lower teeth as being possible sources of the bruise marks.  The evidence was summarised by Kneipp J at 416 as follows:

"Dr Romaniuk and Mr Brown saw bruising related to all four upper incisors.  Mr Sims related that bruising to only 3 teeth, he did not relate any bruising to the left lateral incisor.  In addition, however, Mr Sims related the left central incisor to a mark to which Dr Romaniuk and Mr Brown related the left lateral incisor.  It would follow, one would think, that Mr Sims would relate bruising to the right central and right lateral incisor which Dr Romaniuk and Mr Brown would relate to the left and right central incisors.  But, finally, all three related a certain mark to the right lateral incisor.  So far as the lower teeth were concerned Dr Romaniuk saw bruising which he would relate to all 4 teeth.  Mr Sims saw bruising which related to 3.  He saw nothing relating to the left lateral incisor.  So far as Mr Brown is concerned, he spoke of a scraping mechanism, causing defuse bruising, and I think it fair to infer he would not relate any particular bruising to any particular teeth".  (Shepherdson J dealt with that evidence at 429-430)

Kneipp J went on to make the comment that the "evidence does not seem to show any satisfactory explanation for these discrepancies";

8. I have already referred to the statement signed by Dr Romaniuk on 8 October 1973.  His oral evidence at trial, of course, was to the effect that a comparison could be made resulting in an identification.  Kneipp J at 416 stated:

". . . he did not account satisfactorily for the change.  He spoke of ongoing research and study in this field, but none of the matters which he mentioned specifically appeared to me to be relevant to his change of opinion".

9. Thereafter the matter rested until probably 1997 when Swifte, in circumstances hereinafter set out, alleged that the appellant had made a confession of guilt to him in 1984.

10. On 12 February 1999 the appellant was arrested and charged with perjury at his original trial for murder, the alleged perjury being that he swore he did not kill Deidre Kennedy. The indictment presented against the appellant was in these terms:

"That on the 8th day of March, 1985 at Brisbane in the State of Queensland, Raymond John Carroll in a judicial proceeding, namely the trial of Raymond John Carroll for the murder of one Deidre Maree Kennedy knowingly gave false testimony to the effect that he, Raymond John Carroll did not kill the said Deidre Kennedy, and the false testimony touched a matter which was material to a question then depending in the proceeding".

The wording of that indictment followed s 123 of the Criminal Code;  the essential element of the offence is the giving of false testimony touching a matter which is material to a question depending in the subject proceeding.  There can be no doubt that the giving of a false answer on oath to the direct question:  "Did you kill Deidre Kennedy?" would constitute perjury.

11. The appellant raised a number of issues prior to trial on an application brought pursuant to s 592A of the Code.  Those issues were determined by the trial judge, Muir J, as reported at (2000) 115 A Crim R 164.  Section 592A provides that a ruling given on such an application may not be the subject of an interlocutory appeal, but the ruling may be raised as a ground of appeal against conviction.

12. Thereafter the appellant went to trial on the perjury charge and he was convicted on 3 November 2000. From that conviction this appeal is brought. The grounds of appeal raise a number of matters for consideration by this Court, including matters the subject of the s 592A ruling.

13. Before examining the course of the perjury trial I propose to discuss the law applicable to the situation where a person who has been acquitted of an offence is charged with perjury with respect to the gravamen of that offence. On the s 592A application one of the appellant's contentions was that the perjury proceeding was an abuse of process because it infringed the rule against double jeopardy.  In this context one finds expressions such as autrefois acquit, res judicata, issue estoppel and double jeopardy used in the cases.  As Mackenzie J (with the concurrence of McPherson SPJ) said in R v El-Zarw [1994] 2 Qd R 67 at 80:

"This is an area of law in which unless terminology is used precisely it has the capacity to cloud the underlying principles".

In my view it is the underlying principle which is the critical consideration, not the tag used to describe it in a particular case.

14. At the outset of this discussion it should be noted that s 17 of the Code has no application here. It establishes a limited statutory defence which reflects the common law pleas of autrefois acquit and autrefois convict. The appellant could not have been convicted of perjury on the original murder charge, and his acquittal of murder is not an acquittal of an offence of which he might be convicted upon the indictment for perjury. For those reasons s 17 is not applicable: see R v Gordon, ex parte:  Attorney-General [1975] Qd R 301 and R v Viers [1983] 2 Qd R 1 at 4.

15. But the court has a wider power, derived from its jurisdiction to prevent an abuse of process, to stay criminal proceedings if their continuance would involve re‑litigating a matter which had already been disposed of by earlier proceedings.  It was by applying that reasoning that Thomas J stayed the proceedings in Viers, not withstanding the fact that s 17 of the Code had no application.  It is also implicit in the reasoning of the members of the court in El-Zarw that in an appropriate case criminal proceedings could be stayed on that ground.  There are numerous cases in the High Court in which staying criminal proceedings on that ground has been considered and there is no reason for doubting that the Supreme Court of Queensland has that overriding jurisdiction.  (cf. Williamson v Trainor [1992] 2 Qd R 572)

16. In Rogers v The Queen (1994) 181 CLR 251 at 255, Mason CJ observed that the "circumstances in which abuse of process may arise are extremely varied and it would be unwise to limit those circumstances to fixed categories".  Shortly thereafter at 256 the Chief Justice cited with approval a passage from Walton v Gardiner (1993) 177 CLR 378 at 393 to the following effect:

" . . . proceedings before a court should be stayed as an abuse of process if, notwithstanding that the circumstances do not give rise to an estoppel, their continuance would be unjustifiably vexatious and oppressive for the reason that it is sought to litigate anew a case which has already been disposed of by earlier proceedings".

That then led his Honour to say at 256-257:

"Re-litigation in subsequent criminal proceedings of an issue already finally decided in earlier criminal proceedings is not only inconsistent with the principle that a judicial determination is binding, final and conclusive (subject to fraud and fresh evidence), but is also calculated to erode public confidence in the administration of justice by generating conflicting decisions on the same issue".

It is also interesting to note that in the same case at 259, Brennan J observed that "the plea of autrefois acquit has an established territory of operation underpinned by the principle against double jeopardy".  His Honour was there recognising that the principle against double jeopardy is the underlying broader principle which is relevant in situations such as that which exist here.

17. It is not necessary to analyse further the reasoning of the High Court in Rogers.  However, some passages from the judgments in The Queen v Storey (1978) 140 CLR 364 are instructive in the present situation.  The first passage is in the judgment of Barwick CJ at 372 where he said:

"The correct principle relevant to the admissibility in a subsequent trial of evidence given in an earlier trial which has resulted in an acquittal is, in my opinion, no more than this:  that a verdict of acquittal shall not be challenged in a subsequent trial:  the accused in the hearing of a subsequent charge must be given the full benefit of his acquittal on the earlier occasion.  Evidence which was admissible to establish the earlier offence is, in my opinion, not inadmissible merely because it was tendered in the earlier proceedings, but it may not be used for the purpose of challenging, or diminishing the benefit to the accused of, the acquittal.  Such evidence will be admissible, provided it is relevant to the subsequent charge or to a defence to it but must only be allowed to be used to support that charge or negative a defence.  Where evidence which would tend to prove the earlier charge or some element of it is admitted in the subsequent charge, the jury must be duly warned that they must accept the fact of the earlier acquittal and not use the evidence in any way as to reconsider the guilt of the accused of the earlier offence or to question or discount the effect of the acquittal".

That statement was made in the course of a judgment which extensively examined the reasoning in DPP v Humphrys  [1977] AC 1, a case to which reference will be made later.  The other passage from Storey of significance for present purposes is to be found in the judgment of Jacobs J at 409, where he said:

"The only possible exception to this rule is where in the later trial the defendant is charged with an offence consisting of an act or acts which caused or may have caused the jury in the earlier trial to determine an element or elements of the offence in his favour.  A charge of perjury committed in the course of the earlier trial is an obvious example".

El-Zarw is the only reported Queensland case involving a perjury trial with respect to the gravamen of an offence of which the accused had been acquitted.  As Mackenzie J therein noted at 80, there are only a few recorded instances of that occurring in common law jurisdictions.  The explanation is readily understandable.  (See Friedland, Double Jeopardy at 157-160.)

18. Ultimately it was held in that case that there was no objection in principle to such a perjury charge being laid "if it becomes apparent from substantial evidence that he committed perjury for the purpose of obtaining the acquittal" (84). Later on the same page Mackenzie J said:

"There is support for the proposition that where the Crown brings a charge of perjury against a person who has given evidence in his own defence on another charge and the gravamen of the charge of perjury is that he has sworn that he did not commit the offence, there should be acceptable evidence in addition to that called at the first trial that suggests that at least prima facie that perjury has been committed in the course of obtaining the acquittal.  In the absence of such evidence there is no reason why the acquittal should not be given full effect".

In that case at 76, Ambrose J took the view that there would have to be "substantially different evidence" if the perjury trial was not to be an abuse of process because it infringed the rule against double jeopardy.  Those propositions are supported by the reasoning of the High Court in Storey, of the House of Lords in Humphrys and of the judges in Her Majesty's Advocate v Cairns [1967] JC 37 in Scotland.  The latter two cases, to my mind, provide good examples of when a prosecution for perjury may be brought where the alleged perjury is falsely swearing that he did not commit the crime of which he had been acquitted.  The Full Court in McDermott [1899] 24 VLR 636 upheld the prosecution for perjury where the evidence was identical with that at the trial where the accused was acquitted of the substantive offence.  It was held, correctly, that autrefois acquit was not available, but no consideration at all was given to the principle of double jeopardy or whether the later proceeding was an abuse of process.  The decision should be regarded as overtaken by later High Court authority.  I also regard the reasoning in the Canadian case of Bayn (1932) 59 CCC 89 as irreconcilable with the High Court cases.

19. Cairns had been charged with the murder of a fellow prisoner by stabbing him. On his trial he gave evidence that he neither assaulted nor stabbed the deceased. He was acquitted. On his subsequent release from prison he signed a contract with a newspaper group and pursuant to that contract provided to the newspaper a tape recorded statement which contained a confession that in fact he had murdered his fellow prisoner. The newspaper then published those statements. Not surprisingly Cairns was then charged on an indictment alleging perjury in that he "did depone that you did not, on 26th October 1965, . . . assault and stab . . . the truth as you well knew being that you did . . .". A preliminary point was taken as to the competency of the indictment and that came before the High Court of Justiciary for determination. Much was made on the accused's behalf of the fact that this was the first occasion in Scotland, if not in the United Kingdom, where a person acquitted of a charge was subsequently tried for perjury in denying guilt on oath. The members of the court were not influenced by that circumstance. Each pointed out that it was a matter for the appropriate prosecuting authority to determine whether or not in the circumstances such a charge should be brought.

20. Lord Strachan said at 42 that it "is no doubt true that in order to substantiate the present charge the Crown will have to prove that [the accused] did in fact assault and stab . . . and therefore part of the facts now founded on are the same facts as those which were founded on the earlier trial. That, however, is not sufficient to bar a second trial for a different offence". Lord Wheatley made a statement to similar effect at 46. Importantly the reasoning of Lord Wheatley emphasised (particularly at 47) that what would be investigated on the second trial would be whether the accused committed perjury at his original trial. That approach pervades the reasoning in each of the three judgments in that case. There would be no re-trial of the murder charge, but a perjury trial which incidentally would receive evidence to the effect that the accused committed the offence the subject of the earlier trial. If Cairns could not be charged with perjury that would make a mockery of the law. No reasonable member of society would accept a situation where a person could boast that he actually committed a murder he denied committing on oath and for which he was acquitted and yet the law was powerless to deal with him for perjury. As Lord Wheatley said at 48:

"To give a general immunity to accused persons to commit perjury, however blatant, and perhaps even publicly boast of its success, would only bring the law into disrepute".

21. Humphrys was a somewhat different case.  He had been charged with driving a motor vehicle on July 18, 1972, while disqualified.  The issue on his trial for that offence was whether a police officer correctly identified him as the driver of the motor bicycle on the day in question.  Humphrys gave evidence at that trial in the course of which he denied driving any motor vehicle during 1972.  He was acquitted.  He was subsequently charged with perjury in that he gave false evidence, namely, that he did not drive any motor vehicle during 1972.  It is thus obvious that the alleged perjury was not confined to a denial of the precise charge on which he had been acquitted.  On the perjury trial the prosecution led evidence from a number of witnesses to the effect that Humphrys had driven a motor vehicle on days in 1972 other than July 18.  But they also as part of the case called the police officer in the July 18 matter to give evidence again identifying the accused as the driver of a vehicle on that date.  The question on appeal was whether or not the evidence of the police officer had rightly been admitted.  The House of Lords rejected the proposition that the doctrine of issue estoppel had some place in English criminal law.  It was held that the acquittal at the first trial was no bar to the admission on the trial for perjury of evidence given at the first trial even though, if accepted, that evidence would support the inference that Humphrys was guilty of the original offence.  None of the judgments questioned the operation of the rule against double jeopardy.  The following extracts from judgments in that case demonstrate how the double jeopardy principle is capable of accommodating a subsequent charge of perjury.

22. Viscount Dilhorne said at 26:

". . . evidence given at a trial is not admissible at a second trial to prove the guilt of the person of an offence of which he was acquitted at the first trial or of which he could have been convicted at the first trial but was not, but that determination at the first trial of an issue in the defendant's favour is no bar to the admission at the second trial of evidence given at the first trial directed to establishing perjury at the first trial, even though that evidence, if accepted at the second trial, will lead to the inference that he was guilty of the offence of which he was acquitted at the first trial".

Lord Hailsham at 30 recognised that there would have to be "significant new evidence" supporting the perjury charge before it could be brought.  He pointed out at 33 that the technical plea of autrefois acquit was obviously not available;  the matter had to be resolved by recourse to the broader principle of double jeopardy.  Then at 34 he said:

"No doubt it would be possible for an unscrupulous prosecutor to attempt to use perjury trials as a mere excuse for having a second shot at securing a conviction on the original charge.  I have no doubt in such a case that the court would have an inherent jurisdiction to prevent an abuse of its process and, . . . would at the same time be entitled to quash the indictment as substantially infringing the rule against double jeopardy when this is viewed in its proper light".

His Lordship summarised those matters again at 40-41 of his judgment.  It is sufficient to note that in his view where the evidence was "substantially identical" with that given at the first trial "without any addition" the rule against double jeopardy would be infringed and the perjury prosecution would constitute an abuse of process.

23. Finally, brief reference should be made to the reasoning in that case of Lord Salmon at 46-47.

24. From those authorities the following propositions can be deduced:

1. the focus at the subsequent trial should be on the perjury and not on the crime of which the accused was acquitted at the earlier trial;  the case should not be presented to the jury on the perjury trail as if it were a re-trial of the original offence;

2. evidence given at the earlier trial can be led on the perjury trial, but only to support the charge of perjury.  Generally, there would need to be "substantial" evidence, not led at the earlier trial, which prima facie pointed to perjury to which the evidence led at the earlier trial would be relevant in a subsidiary way;

3. if the evidence at the perjury trial was substantially identical with that at the earlier trial, the rule against double jeopardy would be infringed and the prosecution would amount to an abuse of process.

25. I should at this point dispose of one argument advanced by counsel for the appellant. He submitted that as the evidence in question did not affect the result of the 1985 trial it could not be the subject of a perjury charge. In advancing that argument he relied on the passage quoted above from the reasoning of Jacobs J in Storey at 409.  The contention was that it was only evidence which "caused or may have caused the jury" to reach a particular determination which could provide the basis of a perjury charge;  that was not the case here because the jury convicted.  The reason for the subsequent acquittal had nothing to do with the evidence in question.  The submission is based on a misunderstanding of what Jacobs J said;  so long as the evidence may have caused the jury to reach a particular decision the evidence may support a perjury charge.  That is essentially a reference to the wording in the charge:  "touched a matter which was material to a question then depending in the proceeding".  (see also per Viscount Dilhorne in Humphrys at 21‑22.)

26. On the hearing of this appeal I put to senior counsel for the Crown that one got "a feeling . . . from reading the record and the summing-up . . . that it really was a re‑trial for murder rather than a perjury trial".  His reply is instructive:

"I perhaps should stand up and take blame for that because as counsel for the Crown that was the way the case was conducted.  I don't shrink from that.  The Crown took the view that in order to prove there was that lie as particularised in the indictment the Crown in effect had to prove that and that no doubt has flown on to the direction by the trial judge".

The reference at the end of that quote was to the following passage in the summing‑up:

"The only real issue is whether he told a lie when in that trial he said he did not kill Deidre Kennedy.  If he did kill Deidre Kennedy, you may think it plain that that was something well-known to him.  If you conclude that it is not established beyond reasonable doubt that he did kill Deidre Kennedy, it follows that it has not been proved beyond reasonable doubt that the accused gave false testimony and the verdict must be not guilty".

That statement by counsel, and that passage from the summing-up, are at least cause for concern;  in the light of them the trial must be carefully scrutinised with a view to determining whether or not it infringed the law as stated in the authorities to which I have referred above.

27. The prosecution case at the perjury trial could be summarised as follows:

1. The appellant had the opportunity to commit the murder – he was not on the RAAF Base at Edinburgh at the time as he claimed.  The RAAF evidence was substantially the same as on the murder trial;  Exhibits 25, 27, 28 and 30 could arguably have supported the appellant's claim that he was on base at the time;

2. the appellant was in the Ipswich area at the time of the murder – the evidence of Hill which was not led at the murder trial;

3. the appellant had a propensity for biting small children.  Evidence from Ferguson, not called on the murder trial, was led to support this;

4. odontological evidence established that the bite mark on the baby's thigh was caused by the appellant's teeth;

5. the appellant made a confession to a jail inmate that he had committed the murder;  the evidence of Swifte, who was not called at the murder trial.

It will immediately be obvious that the only new evidence (leaving aside for the moment the odontological evidence) is that from the witnesses Hill, Ferguson and Swifte.  The evidence from Hill and Ferguson if totally accepted does not amount to prima facie evidence that the alleged perjury was committed at the earlier trial.  At most total acceptance of the evidence of Hill and Ferguson might strengthen a circumstantial case against the appellant that he was responsible for the death, but that was not the main issue on this trial.

28. One has then to look carefully at the evidence of Swifte in order to determine whether there is "substantial" evidence, not led at the earlier trial, which prima facie points to the alleged perjury. The "alleged confession" was regarded by Muir J on the s 592A application as evidence which was "significantly different and stronger" than that led at the earlier trial (115 A Crim R at 170).  Muir J recognised at 172 that if Swifte's evidence was excluded that would, most probably, result in the prosecution not proceeding.  The grounds on which he was asked at that stage to exclude Swifte's evidence and stay the proceedings were:

1. the general unreliability of a jailhouse confession made some 16 years ago and not reported to police until 1997;

2. the dangers of relying on such a confession identified in Pollitt (1992) 174 CLR 558;

3. the delay prejudiced the accused's opportunity of inquiring into the truth of the confession;

4. Swifte asserted he had told a prison guard of the confession in 1984, but that person was "now dead".

29. Muir J concluded this aspect of the application by saying: "Conducting the balancing exercise required by the above authorities, and having regard to the role appropriate directions may play in safeguarding the accused's interests, it does not appear to me that a stay should be ordered on this ground".

30. It is not necessary to cite passages from Pollitt.  It is sufficient to say that evidence of prison informers is generally regarded as unreliable and a jury should be directed in specific terms that it is dangerous to act on the evidence of a prison informer where there is no substantial confirmation thereof by independent evidence.  One can assume for present purposes that Muir J was justified in ruling as he did on the s 592A application.  This Court now has the advantage of assessing the evidence of Swifte in the light of what emerged at the trial.

31. Immediately before Swifte was called to the stand the Crown prosecutor made the following admission (transcript 452-453):

"The accused was arrested on 24 February 1984, a Friday, on a charge of murder . . . he was held initially at the Ipswich watchhouse.  He appeared in the Ipswich Magistrates Court on 27 February, 1984.  He was received into the Brisbane prison Boggo Road on 27 February, it’s the same day, 1984, a Monday at a time unknown.  He was transported from the Brisbane prison to the Ipswich Magistrates Court on 28 February, 1984 at a time unknown prior to 4pm.  He was released from the Ipswich watchhouse on 28 February, 1984 at a time unknown and he was not returned to custody prior to his trial in 1985".

Indeed, I was the judge who granted bail.  A perusal of my Chamber Book reveals that the application (OS 136/84) was the last matter heard on 27 February 1984;  the material included an affidavit of the applicant-accused sworn that day.  The bail order was signed late that day.

32. The evidence of Swifte is to be found at pp 454-475 of the transcript. He said that in February 1984 he was in the Boggo Road prison. His evidence indicates that he has spent a great deal of time in prison; his record includes offences of dishonesty. The observation should be made that Swifte had difficulty in putting his evidence of relevant conversations into direct speech. According to Swifte the accused "asked me what I knew about bite marks". His evidence would not suggest anything more specific being said about bite marks; in that context, rather surprisingly, Swifte replied: "Mate, as far as I know they're as good as fingerprints . . . you're a shot duck". Then comes a critical piece of evidence; Swifte said: "I spoke to him over the couple of days". He then goes on, it would appear, to give evidence of conversations which occurred at a later time. Again he said: "I can distinctly remember talking to him over a different period of time . . .". It was after making that statement that according to his evidence he asked the appellant "what he was arrested on". He received the reply: "the murder at Ipswich of the baby". He was asked could he be more specific about the conversation and his reply was as follows:

"He had been out snowdropping which means stealing women's clothes, and everything, you know.  Like he'd been out all night and he found you know, like he found a window open.  I can't remember what type of house it was or what – what it was.  Told me that he, you know, been out pinching women's items clothes, looked in and seen the baby, took the baby, took her out, abused her and murdered the child.  It was simple as that.  That's the way he told me and I thought – well you know – I spoke to him at different times why he . . .".

As I understand Swifte's evidence he subsequently saw the appellant in Woodford jail between the conviction on the murder trial and his subsequent acquittal.  According to Swifte he believed from that sighting that the appellant had been convicted.  Whilst it is not entirely clear, Swifte seems to be saying that one of the reasons he did not tell anyone about the alleged confession is that he believed the appellant had been convicted.  But that does not explain the failure to reveal the confession between February 1984 and the trial in February 1985.  He does assert he told a prison officer (now deceased) of it some time during that period and was told to see another person but did not do so as he did not wish to become involved.  This is not a satisfactory explanation.

33. In March 1997 Swifte was facing a number of criminal charges. His first statement to the police about the alleged confession was made on 19 July 1997. On sentence for the 1997 criminal offences the sentencing judge was asked to take into account in mitigation of sentence the fact that he, Swifte, was co-operating with the police in relation to the murder of Deidre Kennedy.  Swifte maintained under cross‑examination that he did not make the statement primarily for the purpose of obtaining some benefit on his sentence;  reluctantly he conceded he did in fact receive a benefit.  He gave evidence that his reason for going to the police was the following:  "I will tell you why I did it.  Because I woke up and seen a child sitting at the end of my bed one night and I went into horrors for two days, and I said 'no, no', I rang them, then I didn't ring them".  A little later he said:  "I swore that I was speaking to a child . . .", and he claimed it "wasn't a delusion".  He did admit to having had an alcoholic problem, but denied that he was a chronic alcoholic or that  his problems with the drink had something to do with his vision of the child.

34. Finally, some cross-examination as to the period of time over which he spoke to the appellant in Boggo Road jail should be noted. He said he did not think it was the day that the appellant arrived at the jail that he spoke to him. He was "pretty certain" it was the next day. He had a number of conversations, some three or four. These occurred over a "couple of days". He referred on more than one occasion to such conversations taking place over "a couple of days". In one answer he said he could remember speaking to the appellant "on a couple of occasions in the morning and then the afternoon".

35. I turn now to the summing-up with respect to Swifte's evidence. The learned trial judge commenced this segment by saying: "In the case of Swifte, if you accept his evidence, it is direct evidence, also, that the accused bit and killed Deidre Kennedy because it is a confession by the accused of those facts". His Honour then over a number of pages (transcript 1127-1130) summarised the evidence without making much comment on it. Significantly, his Honour did not in that summary refer to the fact that Swifte claimed that the conversations took place over "a couple of days". He did refer to Swifte's criminal history, to the fact that he did receive a benefit on his sentence in 1997 because of his asserted co-operation with the police in this matter, and to the fact that a person such as Swifte "can produce a story by reference to facts which cannot be disproved except by a bare denial". There was no specific reference in that passage to the delay between 1984 when the confession was allegedly made and 1997 when it was revealed to the police; nothing was said about the difficulties inherent in rebutting such evidence after such a lapse of time. Though there was reference to the fact that he told a prison officer about the confession, no point was made that that person was dead by the time of the perjury trial, thus depriving the appellant of the opportunity of checking that story.

36. Finally, the learned trial judge concluded this segment of his summing-up by saying: "For all of those reasons, it would be extremely dangerous to convict in reliance on Swifte's evidence alone and I so direct and I direct you also that his evidence should be treated with extreme caution". What the jury were not told, and in my view there should have been a specific direction on it, was that Swifte's evidence was highly unlikely to be true (perhaps more likely to be untrue) because the appellant had been in Boggo Road for no more than approximately 24 hours. It was conceded that he was received into the Boggo Road prison at some time on 27 February 1984.  It is also a matter of public record that at some time on that day (in theory it could have been before he was taken to Boggo Road) he signed his affidavit in support of the bail application.  At some time before 4pm the following day he was in the Ipswich Magistrates Court, and released from there on bail.  There was clearly no time for Swifte to have had the series of conversations he swore to over a "couple of days".  The underlying premise in Swifte's evidence is that it was not on the first occasion they met in Boggo Road that the alleged confession to murder was made.  That came at a later time.  Leaving aside all the doubts any rational person must have about Swifte's evidence because of his account of the vision of a child and the timing of his statement to the police and its use on his being sentenced in 1997, the confession could not have been made in the circumstances deposed to by Swifte.  That clearly takes the matter far beyond the situation considered by the High Court in Pollitt.  At least the jury should have been very specifically directed that in the light of all that he said on oath it was so highly improbable that the confession was made as alleged that Swifte's evidence should be rejected in its entirety. 

37. Indeed, I would go further. If at the time of the s 592A application all the facts which emerged at trial were known, the evidence should have been excluded. It could not meet the test of being substantial new evidence. If that had been the ruling on the s 592A application the perjury proceedings would have had to be stayed. Not only was it not excluded at trial, but it went to the jury on the basis that they could, subject to a mild warning, accept it as direct evidence of a confession. The fact that Swifte's evidence was left to the jury in that way means, in my view, that the verdict of the jury must be regarded as unsafe and unsatisfactory. That alone is sufficient in my view to require that the conviction be set aside. It is, however, necessary to refer to other aspects of the case before determining whether or not there should be a re-trial.

38. So far as the issue of opportunity is concerned, the evidence at the perjury trial was substantially the same as that at the murder trial, save for the additional evidence from the witness Hill. If one has regard to the evidence from RAAF personnel it could be said that, though there was some evidence either way, a reasonable jury could conclude that the appellant had left the base before the passing out parade. It is fair to say that the significant time lapse between April 1973 and November 2000 meant that most witnesses faced an impossible task in giving evidence from the perspective of a clear recollection. Relevant RAAF documents had been lost and that meant there was little, if any, positive evidence; some possibly relevant exhibits have already been referred to. The problems with this evidence are demonstrated by reference to the evidence of Darryl Stevenson, one of the RAAF personnel. He gave evidence at the murder trial that he could recall the appellant being at the passing out parade; he could recollect introducing the appellant to his relatives at a function after the parade. When giving evidence at the perjury trial he stated that his memory had deteriorated and though he acknowledged what he said at the first trial he did not at the time of the second trial have a specific clear recollection of those events. But at the end of the day, so far as the evidence from the RAAF witnesses are concerned, it is sufficient to say that it was substantially the same as that given at the murder trial.

39. Counsel for the respondent also submitted that the photograph being no. 9 in Ex. 19 proved that the appellant was lying when he said he was on base at the time of the passing out parade. It is correct that on the murder trial he gave evidence of where he was standing at the material time. True that the photograph referred to shows some part of the general area referred to by the appellant. But that does not prove that the appellant was not there; he could have been just outside the area shown in the photograph. The photograph (which is new evidence not available at the murder trial) does not materially advance the prosecution case.

40. In the light of that one must turn to the evidence of Hill; that was put forward as new, substantial evidence to the effect that the appellant was in the Ipswich area at the relevant time. On the s 592A application counsel for the appellant submitted that Hill's evidence should be excluded in the exercise of the Court's discretion because it was too stale to permit a fair trial and the evidence was not "substantial and acceptable" new evidence. One of the points relied on by the appellant was that according to Hill there were two other people present in the house when the appellant allegedly visited her on the day of the killing. Both of those people were dead by the time her evidence came to light and the opportunity of testing her account in that regard was lost.

41. In his reasons on this point at 173, Muir J referred to Brisbane South Regional Health Authority v Taylor (1996) 186 CLR 541 at 551 wherein it was noted that deterioration in the quality of evidence because of the lapse of time is a not uncommon feature of litigation.  His Honour then went on to conclude: 

"I am mindful of considerations such as these, but nevertheless, find against the accused on these contentions.  The evidence of Desley Hill is within short compass.  The evidence is new, but the reasons for its lateness have been appropriately explained, and are not challenged.  The evidence is significant and cogent".

42. I now turn to her evidence as it emerged at trial. She was approached by a police officer investigating this matter in March 2000 and she ultimately made a statement about events she claimed to recall happening on 14 April 1973. In the course of her evidence dealing with that first contact with the police officer she said: "As we went through the years, yes, and as he suggest, to jog my memory, yes". It seems clear that some input from the police officer was necessary in order for the witness Hill to be able to place the events in question in point of time. In her evidence in chief she endeavoured to place the relevant date in the context of her pregnancy which resulted in the birth of a child, Natasha, on 25 July 1973. She consistently maintained that Natasha was born 10 weeks premature; that would have placed her conception about early January 1973. She conceded that she did not know she was pregnant until about 6 weeks after that occurred. She informed her then boyfriend, the baby's father, of that fact and he terminated the relationship. On the evidence that must have been around mid-February 1973. According to her evidence, some weeks later she recommenced a relationship with the appellant which lasted for 2‑3 months.  Her evidence immediately runs into problems because there is no doubt that the appellant had joined the Air Force by 9 February 1973 and was on the Edinburgh Base from that date.  The witness Hill maintains that after the relationship recommenced it lasted for some 2-3 months;  all of this occurring before the relevant date, 14 April 1973.  Again, the problems with her evidence are obvious.

43. She says that on a Friday she purchased a cot in preparation for the birth of her child. It was not delivered until the following day, Saturday. She then maintains it was on that Saturday that the appellant visited her at her parent's unit at Woodridge. Essentially she specifies the date by referring to a television news bulletin dealing with the discovery of the body of Deidre Kennedy earlier that day. According to her the appellant became agitated on seeing that news item and left the home quite suddenly shortly thereafter.  According to Hill he said "he had to get home because his mum was sick". The witness was cross‑examined about other events surrounding that date and it is clear that her recollection of other events at about that time is vague.

44. In his summing-up the learned trial judge said with respect to the witness Hill that her "account has some obvious difficulties". He went on to observe that if she fell pregnant in January 1973 "her account of the timing of the break-up with the father of the child and the resuming of the relationship with the accused cannot be accurate". The learned trial judge asked the jury to consider whether she could be mistaken when after 27 years she said she could remember the television program and who was there. He also invited the jury to consider whether she could have a confused recollection, in effect putting incidents that occurred some time apart as happening on the one day.

45. The learned trial judge left Hill's evidence to the jury in that way. That is hardly consistent with evidence described in the ruling on the s 592A application as "significant and cogent". Given the way Hill's evidence emerged at trial it did not significantly advance the prosecution case. As already noted, at the murder trial there was no evidence placing the appellant in the Ipswich district on 14 April 1973. It is clear from the passage dealing with Hill's evidence in the summing‑up, that it did not positively place the appellant in Ipswich on the date in question.  At most for the prosecution it was some other slight evidence which, if accepted by the jury, might have put the appellant in a position where he had an opportunity of committing the crime.

46. But more importantly the evidence was not significant new evidence pointing to the commission of the alleged perjury. Even if it was accepted in its entirety it would amount to no more than additional circumstantial evidence from which a jury might conclude beyond reasonable doubt that the appellant committed the crime in question. But that of course was not the issue here. The evidence certainly was not new significant evidence or substantial acceptable evidence pointing to the fact that the alleged perjury was committed by the appellant.

47. Hill's evidence did not substantially advance the opportunity evidence led at the murder trial.

48. I turn now to the similar fact or propensity evidence. As already noted, relying on Perry and Sutton, the Court of Criminal Appeal held that the evidence of Grinter (named Myer at the first trial) was inadmissible.  That evidence was to the effect that she believed the appellant had bitten their child in 1975 on the thigh leaving a bruise (at the material time Grinter and the appellant were married).  Muir J ruled that consequent upon the more recent decisions (Hoch (1988) 165 CLR 292, Pfennig (1995) 182 CLR 461, and O'Keefe [2000] 1 Qd R 564) such propensity evidence was probative and admissible.  That can be accepted for present purposes.  But, of course, the fact that there had been a change in the relevant law as to admissibility of evidence after the first trial does not mean that the appellant could be re-tried for murder.

49. Muir J in his ruling on the s 592A application at 173-174 noted that Grinter's evidence had been severely criticised at the murder trial, but he was of the view that the additional evidence from Ferguson served to corroborate that of Grinter. However as the evidence emerged at trial there were major inconsistencies between the evidence of Grinter and Ferguson. In particular Grinter referred to one bite mark on the child's thigh, whereas Ferguson spoke of bite marks all over the child's body: torso, thigh, buttocks, and soles of her feet. Grinter not only disputes Ferguson's evidence in that regard, but she does not recall showing Ferguson the child with the bite mark on her. It seems clear that Grinter did not complain about the appellant biting their child until 1984 when she was shown by the police a photograph of Deidre Kennedy and the bruise on her.

50. Those inconsistencies were referred to by the learned trial judge in his summing‑up (transcript 1169) and on the following page he went on to say: 

"Both versions then you might think cannot be accepted in their entirety if you are to accept either of them and you are left with the task of deciding whether to reject all of that evidence or some of it.  In the latter case you must of course decide what you accept and what you reject".

In those circumstances the evidence of Ferguson could not regarded as significant new evidence.  It certainly is not probative of the fact that the appellant committed perjury at his first trial.  Again, it is no more than another circumstance, of dubious weight, which might be considered by the jury on a murder trial as part of a circumstantial case against the accused.  One could not realistically say that the prosecution case, insofar as it relied on propensity evidence, was any different at the perjury trial to what it was at the murder trial.

51. That leaves for consideration the odontological evidence at the perjury trial.

52. I have already noted that at the murder trial the prosecution called three odontologists who purported to identify the appellant as the person responsible for the bite mark on the baby's thigh. In formulating their opinion they had recourse to the report made by Dr O'Reilly of his post-mortem examination, the evidence of Dr Josephson who was present at the post-mortem, the photographs taken at the post‑mortem, and the observations of Dr Romaniuk of the bruise on the body.  They also had the reproduction of the appellant's teeth made by Dr Romaniuk, said to reproduce their condition in 1973.  Each of those witnesses then used a procedure whereby acetate etchings were prepared and compared.  That enabled each, so it was said, to form an opinion that it was the appellant who was responsible for the bite mark.

53. On the perjury trial the prosecution called three odontologists. A S Forrest, who is a lecturer in oral biology at the University of Queensland and employed at the John Tonge Centre as a forensic odontologist; C R Bamford, a forensic dentist from the United Kingdom who holds a Diploma in Forensic Odontology and P J Craig a dentist from Victoria who holds a Graduate Diploma of Forensic Odontology. Their evidence, particularly that of Forrest, was largely based on computer images which Senior Sergeant Garner had prepared from the photographs taken at the post‑mortem and the reproduction of the appellant's teeth made by Dr Romaniuk.  A mathematician, Adams, was called to give evidence as to how the computer imaging could be adjusted or compensated for possible distortion due to the curvature on the relevant surface depicted in the photograph.  The defence also called an odontologist, A K Whittaker, the Professor of Forensic Dentistry and Oral Biology at the University of Wales, Cardiff.

54. Whilst Forrest to a significant extent based his opinion on comparison with the computer images, the other expert odontologists called on the perjury trial essentially formed their opinion based on comparison using acetate etchings, the same procedure used by the three odontologists called at the original trial. Bamford expressly conceded that there had been no changes in principle since 1985. It is clear from a consideration of the evidence of the odontologists called at the perjury trial that there was no new substantive information over and above that available at the murder trial, on which their opinion was based. All of the computer generated images (and the calculations made by Adams) were based on the photographs and reproduction of the appellant's teeth available at the first trial. At the murder trial, for substantive reasons given by the Court of Criminal Appeal, the odontological evidence was insufficient to found positive identification of the person responsible for the bite. The odontological evidence led by the prosecution at the perjury trial amounted to no more than calling another three witnesses to give opinion evidence based on the same material which it was hoped might persuade the jury (and the court) that positive identification had been made. In my view that was a clear breach of the double jeopardy principle. If the conduct of the prosecution case at the perjury trial was endorsed then it would mean that the prosecution could always seek to overcome an acquittal by calling another group of experts who it was hoped would be more convincing in expressing their opinion based on the same basic information available at the first trial. It should also be noted that the witnesses on the perjury trial were not in complete agreement as to which teeth were responsible for particular marks.

55. But in my view there are even greater problems with the prosecution case. Each of the three highly experienced odontologists called by the prosecution at the murder trial were agreed on at least one thing – the row of bruises to the upper point of the photograph (Exhibit 39) were left by the top row of teeth. Each of the four odontologists who gave evidence at the perjury trial was of the view that the top row of bruises were left by the bottom row of teeth. Given that remarkable conflict of opinion it is necessary to consider the evidence given at each trial in some more detail.

56. As already noted, Dr Romaniuk was the only odontologist to actually examine the body of the deceased baby; he was the only odontologist who saw the bruise. I have quoted above from his statement of 8 October 1973 wherein he said that comparisons were "inconclusive" because of the "meagre detail in the bite mark on the thigh of the child". There was also evidence from Sims, noted above, that there was a body of opinion held by experts that identification by reference to bruise marks was "not reliable". But each of those three odontologists at the first trial gave evidence supporting the opinion each held, and that included of necessity evidence supporting the view that the bruises to the upper point of the photograph were left by the top row of teeth.

57. Though the prosecution relied on that evidence in support of a murder conviction on the first trial, counsel for the prosecution made it clear on the s 592A application that the Crown did not intend to call any of the three who gave evidence at the first trial.  Muir J noted at 176 of his reasons on the s 592A application that Dr Brown was "now unavailable to give evidence through ill health".  It was stated during argument on the hearing of this appeal that Dr Romaniuk also could not be called because of ill health.  I am not entirely clear from all the material what the position was with respect to Sims.  It seems from what was said at the s 592A hearing that the Crown did not wish to call him.  Bamford described him as his teacher to whom he spoke probably every week.  Importantly the consequence of all that was that the three expert odontologists called by the prosecution at the first trial were deprived of the opportunity of considering their evidence that the row of bruises to the upper point of the photograph were left by the top row of teeth in the light of the opinions of the four odontologists called at the later trial to opposite effect.  It may well be that, if given the opportunity, they could have advanced good reason for adhering to the view they expressed at the earlier trial.  Or, and this would be just as important from an overall point of view, it may have resulted in a firm view being expressed, in accordance with Dr Romaniuk's initial opinion, that the meagre detail shown in the photograph of the mark was capable of supporting contradictory conclusions.  At the end of the day one is left with the distinct impression that the explanation for the differences of opinion, including differences with respect to particular teeth and marks, were due to the diffuse nature of the bruising observed on the body and depicted in the photographs.

58. That is not fanciful, particularly when regard is had to the evidence of Professor Whittaker, a highly experienced odontologist, given at the perjury trial. In particular he expressed the view that it was impossible to determine actual bite mark impacts from the bruising because the bruises were diffuse without sharp edges. That led him to express the opinion that it was impossible to identify the person responsible for the bite mark with any certainty. If, as was his view, because of lack of focus and definition of the bruising only approximate measurement of teeth marks and gaps could be made, it was difficult, if not impossible, to exclude others as being responsible for the bite mark. Further, though the odontologists who gave evidence at the perjury trial were of the view that odontological evidence could positively identify the biter, it was conceded, particularly by Forrest, that there was a body of opinion amongst forensic odontologists (who were highly respected in the profession) to the effect that it could not be so used – it could only be used to exclude a suspect not as a means of positive identification.

59. Even when experts are involved there is a danger that one sees what one hopes to find. Ordinarily in a criminal investigation an odontologist would start off with no more than a bite mark. The procedure would then be to examine carefully the bite mark, noting and describing all of its features, particularly features which could be important for later comparative purposes. Then, after a cast or impression was available of a suspect's teeth, the odontologist would endeavour to determine whether the features noted in the bite could have been made by the suspect's teeth. In other words the features of the bite mark which could be used for identification purposes would be noted prior to a consideration of the suspect's teeth. That was certainly the approach which Whittaker favoured. As he said (at record 866) it is "very important" that at the stage of examining the bite mark the odontologist has not looked "at the casts of any teeth of the suspect". There is an underlying concern here that the three odontologists called by the prosecution were supplied with a photograph (or image) of the bite mark and a cast of the appellant's teeth (or impression thereof) and asked the question, could those teeth have caused the bite mark. In those circumstances, understandably, there would be concentration on possible points of similarity at the expense of noting all features of the bite mark which might constitute similarities or dissimilarities. The procedure followed by Forrest was of first examining the teeth with a view to saying whether or not there were features about those teeth that would lead one to look for corresponding features in a bite mark (record 575). Having identified such features about the teeth he looked at the bite mark to see whether or not corresponding features were present. The danger, in my view, in adopting that procedure is that where the bite mark in question is in effect a diffuse bruise, it is possible to read more into the bite mark than objectively the expert should. This point in itself would probably not result in the necessary rejection of the evidence of Forrest, but it highlights another critical difference in approach between Whittaker and the odontologists called by the prosecution. It highlights the danger of relying on the odontological evidence to make a positive identification of the person responsible for the bite.

60. It is true that the jury on the perjury trial were well aware at the end of cross‑examination of the evidence given by the three odontologists at the earlier trial.  But the evidence of those odontologists was not before the jury on the perjury trial.  That made it virtually impossible for them to evaluate comparatively the odontological evidence given at each trial.  There would have been a natural tendency for them to prefer the evidence given by the witnesses who appeared before them.

61. At the perjury trial the odontologists called by the prosecution referred to what they said was a second bite mark depicted in the photograph and sought to identify the appellant as the person responsible for it. The witness Garner said that the negatives were of "very high quality" but the prints used in the murder trial were not. He used the negatives to generate the computer images and it was then claimed the second bite mark could be seen. A number of points need to be made about the use of the "second" marks to implicate the appellant:

1. Dr O'Reilly in his post-mortem report identified a number of small bruises but did not record this "second" bite mark;

2. Dr Josephson who examined the body where found and was present at the post-mortem saw one bite mark which was "absolutely clear" – "clearer than this photo would indicate" – but was adamant he saw only one bite mark;

3. None of the three odontologists who gave evidence at the murder trial referred to a second bite mark;

4. The angle of the camera and the curvature of the baby's leg meant there was distortion which had to be corrected before any comparison could be made.  It was here that the evidence of Adams, the mathematician, was relevant;

5. Whittaker did not consider the marks were consistent with a bite.

The matters to which I have referred make the evidence of the odontologists relating to the alleged "second" bite of doubtful quality.  The evidence on this point does not add anything of substance to the evidence with respect to the "first" bruise, and if that evidence is not satisfactory it cannot be saved by adding to it the evidence as to the "second" bite.

62. Given all that I have said about the various odontological opinions in these reasons, I have come to the conclusion that a verdict that the appellant was responsible for the death of the child based on the odontological evidence given at the perjury trial was unsafe and unsatisfactory. But, as I have already said, the matter should never have got to the jury. The prosecution case on the perjury trial was essentially a re‑trial for murder based on odontological evidence from a fresh set of witnesses, asking the jury to disregard contrary opinion from the odontological experts called at the first trial where the basic data on which each opinion was based was the same.

63. My reasoning is focused on the specific, rather unusual, circumstances surrounding these trials for murder and perjury. As cases such as Rogers, Storey, El-Zwar, Humphrys and Cairns demonstrate, there is a fine line dividing what is permissible and what is not permissible when a person is charged with perjury in swearing that the substantive offence was not committed.  Advances in technology and science may well mean that, in a particular case, a charge of perjury could properly be brought in such circumstances where the new substantive evidence resulted from developments in technology and science.  But for the reasons which I have given, this is not such a case.

64. When one considers the conduct of the perjury trial, and the evidence as it there emerged, it is clear that the principle of double jeopardy was substantially breached. In essence, the prosecution set out to prove that the appellant murdered the child by calling different witnesses but without there being any new substantive acceptable evidence to that led at the original trial.

65. As the learned trial judge noted in delivering his reasons on the s 592A application the critical evidence was that of Swifte. As he said, without Swifte's evidence the prosecution would, in all probability, not proceed. But, for the reasons given above, it became clear by the end of the trial that no weight at all could be attached to his evidence. Indeed, many may well think it ironic that such evidence was led on a perjury trial although his alcoholism may suggest he was unreliable rather than dishonest. Without Swifte the prosecution had no case; their only hope was to re‑run the murder trial under the guise of a perjury trial.  That in substance is what happened.

66. There were some other matters raised during argument which can be dealt with relatively briefly.

67. As at 14 April 1973, C H (Nugget) Carroll (no relation to the appellant) resided with the witness Borchert and others next door to the Kennedy family. He was then a man aged 73. He occupied a bedroom which was surrounded on two sides by a verandah. On the western side there was a "glass push-up window" which opened onto the verandah where two rope clothes lines were strung. He gave a statement to the police on 16 April 1973 in which he stated that on the evening of 13 April 1973 that clothes line contained numerous articles of female and children's clothing.  About 10pm that evening he was smoking a cigarette whilst lying on the bed.  The window referred to was open approximately 6 inches.  He saw a male person on the verandah and initially thought it was another occupant of the residence.  He noticed that person removing something from the clothes line.  He kept the person under observation for about 5 minutes.  Eventually he realised that the person was not an occupant of the house and in consequence he walked to the kitchen and spoke to other occupants, including Borchert.  They then walked onto the verandah but no person was located.  Nugget Carroll described the prowler on the verandah as a male, aged about 18-20, about 5' 9" tall, of thin build and about 10 stone.  He also said the person's hair was brown to fairish and about collar length.  Those observations by Nugget Carroll were extremely significant because there was a formal admission made by the prosecution that the panties and step-in found on the deceased child had been hanging on the clothes line referred to by Nugget Carroll on the evening of 13 April.  The inference is irresistible that the person seen by Nugget Carroll taking clothes from that line was the person responsible for the death of the child.  The evidence was that Nugget Carroll died early in 1974.  His statement given to police on 16 April 1973 was tendered in evidence at both the murder trial and the perjury trial.

68. The description given by Nugget Carroll of the person he saw taking clothing from the verandah is significant because the appellant at the material time had jet black hair which was short in length, and was six foot one inch in height. The importance of Nugget Carroll's evidence should have been obvious; the issues were discussed at some length by Shepherdson J at 431-432 of his judgment on the appeal from the murder conviction. Counsel for the appellant at the perjury trial specifically asked for a direction in the summing-up with respect to it. There was a complaint on the hearing of the appeal that the learned trial judge failed to direct the jury as to the importance of the statement of Nugget Carroll. Specifically it was submitted the learned trial judge failed to remind the jury that it was clothing taken from the verandah in question which was found on the body of the child and the description did not match that of the appellant. As the evidence identifying the clothing was in the form of a Crown admission that is something which could have been overlooked by the jury given the length of the trial and the detail of the odontological evidence.

69. In my view there is some force in the submission made by counsel for the appellant. A perusal of the summing-up reveals that the learned trial judge merely recounted some of the contents of the statement of Nugget Carroll but did not at any stage of the summing-up specifically draw the jury's attention to the fact that at the material time the appellant had short dark hair, was taller than the man described and the fact that clothing taken from the clothes line on the verandah was identified as being the clothing on the deceased child. However, because I have reached a definite conclusion on other aspects of the case it is not necessary to consider whether those deficiencies in the summing‑up would alone have caused the trial to miscarry.  It is sufficient to say that such considerations add additional support to the conclusion that the verdict of the jury was unsafe and unsatisfactory.

70. Counsel for the appellant also submitted that because of the time lapse of 27 years between the death of the child and the perjury trial the appellant did not have, and could not have had, a fair trial.  A number of potential witnesses, some of considerable importance, had died in the intervening period.  RAAF records which might have thrown additional light on the appellant's whereabouts at the material time had been lost.  Recollections had faded.  In some instances recall of relevant events had been prompted by police questioning.  Generally the lapse of time impacted heavily on the appellant's opportunity of testing the veracity of much of the evidence led against him.  Courts have recognised that the fairness of a trial may often be impaired by a long delay between the occurrence of events and the trial;  reference need only be made to Longman v The Queen (1989) 168 CLR 79 at 91 and R v Johannsen and Chambers (1996) 87 A Crim R 126.  Against that background the learned trial judge gave directions to the jury with respect to the relevance of delay (Record 1120-1121);  he concluded those passages by saying:

"I direct you that for those reasons it would be dangerous to convict on the oral evidence as to events of 27 odd years ago unless after having taken heed of this warning and having scrutinised the evidence with great care, you are satisfied as to the truth and accuracy of the evidence which you accept."

It was submitted that in the circumstances that warning was not adequate;  in particular it was submitted that his Honour should have detailed with greater specificity areas where it was contended the delay had particularly impacted on the capacity of the appellant to adequately test the case against him.

71. Again, because I have reached a positive conclusion based on other grounds, it is not necessary to determine whether there was a deficiency in the summing-up in this regard which alone would have resulted in a miscarriage. It is again sufficient to say that the submissions with respect to the aspect of delay tend to support the conclusion that the verdict of the jury in all the circumstances was unsafe and unsatisfactory.

72. In my view the trial should have been stayed as an abuse of process. But in any event when the evidence is carefully considered I am of the view that the verdict returned by the jury was unsafe and unsatisfactory. On either ground the verdict should be set aside and an acquittal entered.

73. There was also an appeal by the Attorney-General against sentence which should formally be dismissed.

74. The orders of the Court should therefore be:

1. Appeal allowed.

2. Conviction set aside and verdict of acquittal entered.

3. Appeal by Attorney-General against sentence dismissed.

75. HOLMES J:  I agree with the reasons for judgment of Williams JA and with the orders he proposes.