Riley v The State of Western Australia - [2005] WASCA 190

JURISDICTION : SUPREME COURT OF WESTERN AUSTRALIA

TITLE OF COURT : THE COURT OF APPEAL (WA)

CITATION: RILEY -v- THE STATE OF WESTERN AUSTRALIA [2005] WASCA 190

CORAM: STEYTLER P
ROBERTS-SMITH JA
MILLER AJA

HEARD: 20 JULY 2005

DELIVERED : 5 OCTOBER 2005

FILE NO/S: CACR 7 of 2005

BETWEEN: PERCY JAMES RILEY
Appellant
AND
THE STATE OF WESTERN AUSTRALIA
Respondent

ON APPEAL FROM:
Jurisdiction              :  DISTRICT COURT OF WESTERN AUSTRALIA
Coram  :H H JACKSON DCJ
File No  :BUN 36 of 2003

Catchwords:

Appeal - Criminal law - Trial by Judge alone - DNA evidence - Conflict between experts - Other evidence not capable of proving guilt - Reasons for decision - Whether adequate - Failure to address conflicting expert evidence or explain why not necessary to do so - Failure to expose process of reasoning

Legislation:

Nil

Result:

Leave to appeal granted
Appeal allowed
Retrial ordered

Category: A

Representation:

Counsel:

Appellant: Mr D S Hunter & Ms K C Parnell
Respondent: Mr M Mischin & Ms S Markham

Solicitors:

Appellant: Legal Aid of Western Australia
Respondent: State Director of Public Prosecutions

Case(s) referred to in judgment(s):

"S" v The Queen, unreported; CCA SCt of WA; Library No 930569; 19 October 1993
Australian Communist Party v The Commonwealth (1950) 83 CLR 1
Beale v Government Insurance Office of NSW (1997) 48 NSWLR 430
Bendixen v Coleman (1943) 68 CLR 401
Carlson v King (1947) 64 WN (NSW) 65
Chapman v Kirke [1948] 2 KB 450
Gardner v Caporn [2005] WASCA 153
Garrett v Nicholson (1999) 21 WAR 226
Lloyd v Faraone [1989] WAR 154
Mount Lawley Pty Ltd v Western Australian Planning Commission (2004) 29 WAR 273
Pettitt v Dunkley [1971] 1 NSWLR 376
Pownall v Conlan Management Pty Ltd (1995) 12 WAR 370
R v Bropho (2004) 36 SR (WA) 328
R v GK (2001) 53 NSWLR 317
R v Karger (2002) 83 SASR 135
R v Riley [2005] WADC 16

Case(s) also cited:

A Child v The State of Western Australia [2005] WASCA 91
Hillstead v The Queen [2005] WASCA 116
Mifsud v Campbell (1991) 21 NSWLR 725
R v Middleton (2000) 114 A Crim R 258
Salmon v The Queen [2001] WASCA 270

STEYTLER P: I have had the advantage of reading the judgments of Roberts‑Smith JA and Miller AJA. Both have referred to the circumstances giving rise to this appeal. There is consequently no need for me to refer to those circumstances except insofar as is necessary for me to explain why, like Roberts‑Smith JA, I consider that the appeal should be allowed.

The Issues at the Trial

The appellant was said by the State to have been one of two men who burgled a home near Bunbury. While the other man looked for items to steal, the appellant was said to have kept watch over the complainant, KM, who had awoken during the burglary, and, in the course of doing so, to have sexually assaulted her. There was no issue as to the fact of the burglary, or as to the fact that KM had been sexually assaulted by one of the two intruders. The only issue was whether the appellant had been one of the burglars and the man who assaulted KM. The appellant denied that he had been so. The evidence established that he had, at the relevant time, lived in Bunbury and worked in Brunswick. Both of those places are within easy reach, by vehicle, of the town where the offences were committed. However, the appellant said that he had not been to that town in recent years. A search of his home had not revealed any of the items which had been stolen in the course of the burglary. Nor did it reveal any items of clothing of the kind said by KM to have been worn by the man who assaulted her.
As appears from the other judgments, KM was able to give only a limited description of the appearance of the person who assaulted her. She said that he had had an oval‑shaped head and black hair with grey through it. She also described what looked like the top of a beard on his cheeks and his dark, bushy eyebrows and dark brown eyes which, she said, were "smallish and oval shaped but not Oriental". She described his height as being similar to that of one of the investigating police officers.
Some weeks after the assault on KM, while she was at work in Bunbury, she saw a man walking past her place of work. She said that, after a while, she realised that the man looked like the person who had assaulted her, "especially his eyes". She reported this to the police.
However, when KM was shown a number of photoboards which included a photograph of the appellant and photographs of some of his relatives, she was unable to make any identification.
The trial Judge found that the description of the offender which KM had given to the police "adequately" matched the appellant's appearance, as he observed it to be, "in many ways" but that, given the inability of KM to identify the appellant from the photoboards, that description (which was wrong in some respects) could not satisfy him beyond reasonable doubt that the appellant had been "the perpetrator".
The trial consequently turned on the DNA evidence, which has been discussed in some detail in the other judgments. Cellular material containing amylase, a component of saliva, had been taken from inside KM's pyjama top in an area which had covered her left breast. This had provided a mixed DNA profile which was consistent with having come from two persons, one male and one female. This profile was analysed by Mr Lawrence Webb, a senior forensic scientist employed by the State of Western Australia at the PathCentre in Nedlands. He concluded that one of the two persons who had contributed to the mixed DNA was KM. He then isolated the DNA which came from the other, male, person and analysed it.
Using standard Australian practice, the male person's DNA was tested or analysed at 10 loci (locations or sites on a strand of DNA). One of these, the amelogenin site, determined gender. The other nine loci were not random, but were chosen because they had differentiating qualities between individuals and were thus better suited to an exclusionary process than other loci. Of the interpreted profile, seven of the 10 loci returned unambiguous alleles (there are two alleles at each locus). The other three returned what Mr Webb described as "masked results". This meant that, in those three instances, only one allele belonging to the male person could be identified and that the other must be shared with one of those belonging to KM.
The profile obtained from the gender locus and the six unambiguous loci was entered into the PathCentre's "identified persons database". At the time, this database contained in excess of 8000 DNA profiles taken from persons of interest to the police and others, including witnesses, laboratory staff, deceased persons and crime scene profiles. It included two profiles taken from samples of the appellant's DNA which had been provided by him some time earlier. Two matches were made in the database, these being with the two profiles which had been taken from DNA samples which had been provided by the appellant.
The appellant having, in this way, become a suspect in the investigation, a fresh sample of his DNA was taken. The profile obtained from this was compared with that of the male DNA taken from KM's pyjama top. The appellant's profile matched at all seven loci and, at the remaining three "masked" loci, he possessed the required alleles identified in the mixture and these could consequently not exclude him as the intruder (if an allele present at any locus on the crime scene DNA is not present at that locus on the suspect's profile, he or she can be excluded as a suspect).
DNA samples were also taken from three of the appellant's male relatives, being two brothers and a cousin. None of these matched the samples taken from KM's pyjama top.

The DNA Evidence

In his evidence at the trial, Mr Webb relied upon what he referred to as the "likelihood ratio", which is used in the case of mixed DNA profiles and which calculates the likelihood that the mixed DNA found is from the complainant and the suspect rather than from the complainant and some other person. His evidence was that, if it was assumed that only two people contributed to the mixture, it was at least 10 billion times more likely that the mixed DNA profile came from KM and the appellant than that it came from KM and an unknown person. In what he considered to be the unlikely event that two people other than KM had contributed to the DNA mixture found, he said that it was at least one billion times more likely that the appellant was one of the two other contributors than that two unknown persons were the contributors.
Mr Webb described these ratios as conservative. He said that they use an arbitrarily selected cut‑off point by reference to the world's population at large and are consequently rounded down from much greater actual likelihood figures. The PathCentre was said by Mr Webb to use a population database comprised of three data sets - Caucasian, Aboriginal and South‑East Asian. These data sets are based upon statistics obtained from the Bureau of Statistics in relation to major racial groups within Western Australia. An international programme is then used to calculate the "likelihood radio" mentioned above. Mr Webb said that, if only the Caucasian population was taken, the actual likelihood ratio in this case (assuming only two contributors) would be one in 566 billion, if only the Aboriginal population was taken, it would be one in 434 billion and, if only the South‑East Asian population was taken, it would be one in 1571 billion.
Mr Webb was cross‑examined at some length. The cross‑examination relevantly focused on five issues, each of which was the subject of evidence from Dr Brian McDonald, a molecular geneticist who gave evidence on behalf of the defence. I will deal with these in turn.

Random Man Non‑Exclusion

The first issue arose out of a debate as regards the use of what is known as a "random man non‑exclusion" calculation. As I understand the evidence, it established that the "random man non‑exclusion" methodology, if applied in the present case, would calculate the probability that some random person, other than the appellant, may have been the male contributor to the DNA profile found on KM's pyjama top. Dr McDonald appeared to prefer this methodology to the likelihood ratio used by Mr Webb. He suggested, in the course of his evidence, that the "random man non‑exclusion" method was used by the Australian Federal Police laboratory and by the Tasmanian laboratory.
Mr Webb, in his evidence, said that the PathCentre did calculations based upon this method and used them as a guide, but did not include these calculations in reports as there was "still no real consensus in the forensic laboratory area" as to how this methodology should be used. If this methodology was applied, the probability of some other man not being excluded was, on Mr Webb's evidence (Dr McDonald took issue with it in respects with which I shall deal under the following heading), one in 453,700, in the case of the Caucasian population, one in 668,900 in the case of the Aboriginal population (although there was no evidence in this respect, it seems to have been accepted that the appellant was part Aboriginal) and one in 3,097,000 in the case of the South‑East Asian population. However, as Roberts‑Smith JA has pointed out, Mr Webb acknowledged, in the course of cross‑examination, that he was aware that in Germany there had been a match of two unrelated people at 10 loci, that in New Zealand there had been "four matches of people with 10 loci" and that in the Northern Territory there had been a brother and a sister with a match at 10 loci.

Number of Loci Used in Calculation

The second issue arose out of the number of loci used in the PathCentre calculation.
Mr Webb's evidence was that the PathCentre currently examines 10 different loci. At each, a DNA profile is determined using a system which, Mr Webb said, is called "Profiler Plus", being "a commercially available kit which has been established worldwide and validated in … [the PathCentre] laboratory, as well as around the world". That level of testing is accepted as appropriate in Australia, according to Mr Webb, although, he said, the PathCentre was currently looking at a system which would enable it to profile 16 sites.
In his evidence, Dr McDonald appeared to challenge the calculations performed by Mr Webb upon the basis that Mr Webb's calculations were based upon 10 loci rather than seven (one of those loci being, as I have said, a gender locus which identified the person in question as being male). Although his evidence on this issue was in some respects unclear (to me at least), it seems that Dr McDonald had two bases for his challenge. The first was founded upon the fact that only the seven unambiguous loci (and not the three "masked results" analysed from the specimen) were entered into the police database for the purposes of a search for possible suspects. It seems that Dr McDonald considered that, having identified a suspect upon this basis (by a means which showed only that the suspect could not be excluded as a donor of the crime scene DNA, rather than that he was the donor), in the absence of other incriminating evidence, it was wrong to make a subsequent comparison using a greater number of loci upon the assumption that the suspect was the perpetrator of the offence. The second basis appeared to be that the three "masked" loci were in any event ambiguous and, consequently, of limited assistance.
Dr McDonald went on to say that, if the "random man non‑exclusion" method was utilised, and if the calculations pursuant to the use of that method were based only upon the six unambiguous differentiating loci used in the initial database search (and the locus which identified gender only), then, if those calculations were limited to the Australian Aboriginal subpopulation, the ratio would come down to about one in 48,000, not taking into account what he thought to be the significantly greater prospect of a match with one of the appellant's relatives. He also made the point that, if more loci were used for comparison, there would be a greater capacity to exclude suspects.
Mr Webb seemingly did not accept either of the first two propositions advanced by Mr McDonald, to the extent that these were put to him in the course of cross‑examination. He based his calculations on 10 loci rather than seven because this produced the most reliable result. He regarded it as irrelevant that the smaller number of loci had been used initially to identify the appellant. He obviously considered that the three "masked" loci, while of less assistance than those which were unambiguous, were nonetheless of assistance in that they indicated the existence of particular alleles which must have been present in the DNA of a contributor. If the appellant had not possessed them, he could have been excluded as a contributor.

FST Correction

The next issue related to what was described as an "FST correction". The letters "FST" stand for "Factor Subpopulation and Total". According to Mr Webb, an "FST correction" is a "correction coefficient for common ancestry within the populations" and is calculated from annual data sets. Seemingly, common ancestry within a particular population group has the result that certain alleles may be more likely to occur within that group and this correction co‑efficient takes that into account. Mr Webb said that the PathCentre was currently using an FST value of .03 in its data in relation to "single set profiles" but that this had not been used "in the mixture calculation". He said that the PathCentre's current mixture calculation programme did not allow for FST values to be inserted. The effect of his evidence was that there was no consensus as regards the use of FST values but that, even if an FST value had been factored in, it would have made little difference given that "the actual value for the mixture stat calculation was in fact 434 billion" and the PathCentre had "reduced that to 10 billion just by the very nature of … [its] conservative reporting, and that would more than account for any foibles within the population data sets". He said that the FST value of .03 used in relation to single‑set profiles was considered to be a conservative value. He also said that, from "indications" which he had "heard" in respect of work currently being done in respect of the Australian Aboriginal population, it seemed probable that the FST value in respect of that group was more likely to be closer to 1 per cent than to 3 per cent. He went on to say that there had been "other figures bandied about by various statisticians in regard to the Aboriginal data, that it should be up towards 13 per cent" but that "the studies" had shown that not to be the case. However, he added that "that information" had not yet gone to print and he could not "actually produce anything on that, except to say that as time goes on, further discoveries are made, further information is available, and these FTS's [sic] are certainly not set in stone, and really one statistician may vary significantly from another, and who's to say [who] is right?"
In his evidence, Dr McDonald said that there had been studies on appropriate FST corrections for various subpopulations but not, so far as he was aware, in respect of Australian Aboriginals other than one which had "pooled three different subgroups of Aborigines together and averaged them before they started, so that in effect obliterated the purpose of the exercise". He said that the Australian Aboriginal population was "quite unique compared to any other in the world", given "evidence of tens of thousands of years of isolation" and that to apply "Caucasian type data" for the purpose of calculating "some sort of figure of random occurrence of their genetic make-up" was "just missing the point". He had earlier said that studies in respect of American Indians had reflected a correction factor of 9 to 11 per cent and that in New Zealand "they use 5 per cent" in respect of Maori and Polynesian populations.

NRC Recommendations

The fourth issue related to so‑called "NRC [National Research Council] recommendations". In his evidence, Dr McDonald suggested that these recommendations were to the effect that any likelihood calculation should be adjusted to take into account the size of the database which was used in order to identify the suspect. He said that this should be done by multiplying the probability or likelihood arrived at by the number of people in the database.
This proposition had earlier been put to Mr Webb in the course of his evidence. He rejected it, saying that, because a fresh evidentiary sample was compared to the crime scene mixture sample once the appellant had been identified as a suspect, and because the ratio ultimately arrived at did not rely upon the police database, there was no need to apply the recommendation.

Relatives of the Appellant

The last of the five issues raised by the expert evidence related to evidence given by Dr McDonald to the effect that the likelihood calculations made by Mr Webb were flawed because they did not sufficiently take into account the possibility that the perpetrator of the offences charged might have been a relative of the appellant. Mr Webb had been asked, in the course of cross‑examination, whether his calculations of the likelihood ratio of 10 billion to one had taken into account "calculations of excluding any relatives of … [the appellant] from the calculations". He responded by saying that the calculations which had been made referred to the population at large and were not based upon the appellant's relatives. However, he added, a little later in the course of his evidence, that the calculations which had been made "would certainly take in any account for any subpopulation in that data …" and that the figure arrived at "would certainly account for any relatives in that database".

In his evidence, Dr McDonald said that, from a genetic viewpoint, the persons most likely to match a suspect to the same number of loci as was used to match the suspect to the crime scene DNA are those who are related to that suspect, having inherited similar alleles. He said that there are well‑recognised formulae for adjusting a likelihood ratio in order to take into account the prospect of a match from a relative. He said that the ratio of one in 48,000 which he had arrived at (utilising his preferred random man non‑exclusion methodology and also his preferred use of only the six unambiguous differentiating loci) did not address either the FST correction factor which should be used for the Aboriginal population or what he appeared to regards as the very significant prospect that a relative of the appellant might also provide a match. He said, in this last respect, that "in these sorts of situations in towns like this … they [presumably relatives of a suspect] are the population amongst whom the most likely matches are going to occur".

Grounds of Appeal

As is apparent from the judgments of Roberts‑Smith JA and Miller AJA, the particulars of appeal which were pursued on behalf of the appellant focused primarily on these DNA issues and, in particular, upon what were said to be deficiencies in the trial Judge's reasoning in respect of them. Essentially, five propositions were advanced in that regard. They are that:
(a)the trial Judge did not decide whether the DNA found on KM's pyjama top contained six or nine loci that matched the appellant's DNA;
(b)the trial Judge did not say whether he preferred the evidence of Mr Webb or that of Dr McDonald;
(c)the trial Judge did not say whether or not he accepted the statistical probabilities advanced by Mr Webb or those advanced by Dr McDonald;
(d)the trial Judge made no finding whether or not the probability calculations relied upon by the State were flawed by reason of their failure to take into account the prospect that the perpetrator of the offences was a relative of the appellant; and
(e)the trial Judge made no decision whether or not an FST correction should have been made to the calculations put forward by Mr Webb.
Six other propositions were made in the particulars to the ground of appeal. They are that:
(f)the trial Judge erred in finding that the appellant had denied, in his evidence, having previously given DNA samples to police;
(g)the trial Judge failed to take into account the complainant's inability to identify the appellant either through photoboard identification or through attempts at voice identification;
(h)the trial Judge failed to take into account the fact that none of the stolen items was located at the appellant's house;
(i)the trial Judge failed to take into account the fact that no clothes fitting the description of those worn by the offender were found at the appellant's house;
(j)the trial Judge failed to take into account the fact that the complainant's description and an "identity kit" composite photograph prepared by police with the complainant's assistance were different in height, build and physical features than the appellant; and
(k)the trial Judge failed to take into account that the appellant and the complainant did not know each other when the offence occurred.

Was there error?

There is no doubt that this case turned primarily upon the DNA evidence. It is readily apparent from the trial Judge's reasons that he considered (as was plainly the fact) that, the DNA evidence aside, there was insufficient evidence upon which to convict the appellant. So much is obvious from his comment that the case turned on the DNA evidence (at [24] of his reasons) and that the DNA evidence was "the nub of the trial" (at [33] of his reasons). It is well‑recognised that, in dealing with a case involving DNA evidence, a Court may not substitute reliance on a mathematical expression of probability as a substitute for proof of guilt beyond reasonable doubt: see R v GK (2001) 53 NSWLR 317 at 323, per Mason P, and R v Karger (2002) 83 SASR 135 at 139 [11], per Doyle CJ. What the Court is required to do is to consider the statistical DNA evidence together with all of the other evidence in the case in arriving at a decision whether or not the guilt of the accused has been established beyond reasonable doubt.
In this case the trial Judge was faced with two competing hypotheses. The first was that advanced by the prosecution to the effect that the appellant was the source of the male DNA found on KM's pyjama top. The second was that advanced by the defence to the effect that some other person might have been the source of that DNA. The statistical evidence expressed those competing hypotheses as a ratio (cf Karger, at 140 [15]) and his Honour was obliged to take that ratio into account, together with all of the other evidence, in arriving at his decision. The appellant's contention, so far as the DNA evidence is concerned, is that the trial Judge failed to make findings which were required to be made before he could arrive at a proper appreciation of the value of the statistical evidence as evidence of the appellant's guilt or, if he did make those findings, that he gave no explanation for them.
While it is true that overly elaborate reasons are not required from a trial Judge, especially so in criminal trial proceedings in which a prompt outcome is necessary, where one set of evidence is accepted over a conflicting set of significant evidence a trial Judge is required to set out his or her findings as to how it is that the one has been accepted over the other: Beale v Government Insurance Office of NSW (1997) 48 NSWLR 430 at 443, per Meagher JA, and Mount Lawley Pty Ltd v Western Australian Planning Commission (2004) 29 WAR 273 at 282 ‑ 283. Moreover, it is essential that the reasons adequately disclose the intellectual processes which have resulted in the decision: Garrett v Nicholson (1999) 21 WAR 226 at 248 [73] - [74], per Owen J. It is well accepted that, where there is a right of appeal, the reasons must be sufficient to give effect to that right. If the basis for the decision is not apparent, the losing party cannot know whether there has been a mistake of law or of fact: Carlson v King (1947) 64 WN (NSW) 65 at 66, per Jordan CJ; Pettitt v Dunkley [1971] 1 NSWLR 376 at 387 ‑ 388, per Moffitt JA; Lloyd v Faraone [1989] WAR 154 at 162 ‑ 163, per Malcolm CJ, and Mount Lawley, above, at 282 ‑ 283.
What his Honour did, in this case, was to say, at [58] of his reasons, that he accepted the evidence of the State. What he did not say, in that paragraph, was why he did so, where that evidence conflicted with the evidence advanced on behalf of the appellant. All that he there said was that the test of proof beyond reasonable doubt did "not rely on … statistical probability" (which I take to be a recognition of the proposition accepted in such cases as GK and Karger) and that none of the issues which had been raised by the defence, either singly or together, created a reasonable doubt.
There is no doubt that the trial Judge was aware of the aspects of disagreement between Mr Webb and Dr McDonald which I have earlier mentioned. He referred to them at some length, quoting long passages from their evidence in these respects. However, the only analysis of that evidence, or the only reasoning exposed in respect of it, is that which appears in [55] to [58] of the judgment, those paragraphs having been quoted in full by Roberts‑Smith JA.
The trial Judge said, in [55], that, when cross‑examined, "it became clear that Dr McDonald's main concerns were the use of a one in 10 billion number, without factoring in the issues of population subgroups, and based on an assumption that only two people contributed to the DNA mixture". In [56] he said that the statistical and genetic evidence had to be viewed in the context of the known, and uncontested, facts. He referred, in this respect, to the fact that only two people "were involved", one being KM and the other being "identified by the later as well as earlier DNA samples" and being "a local man of sufficiently matching physical and social characteristics". He also pointed out that there had been no DNA match in the case of the three relatives of the appellant whose DNA has been tested, that no‑one else was "said either to have been involved in the sexual conduct or identified", that the Court had not been provided with evidence of family, racial descent or other evidence (apart from that relating to the testing of the three relatives) able to assist in calculating likelihood ratios and that no alibi evidence was led. In [57] he said that there was "no doubt quite proper debate about the calculation of likelihood ratios and the factors to be studied or included". In [58] he made the findings, to which I have earlier referred, that he accepted the evidence of the State and that none of the issues raised by the defence, singly or together, created a reasonable doubt.
I should say, firstly (although no ground of appeal is advanced in this respect), that I understand the main concerns of Dr McDonald to have gone further than those identified by the trial Judge in [55] of his reasons. As I understand them, they related not only to the factoring in of issues of population subgroups and to the assumption that only two people contributed to the DNA mixture (an issue not relied upon in the appeal) but to the whole methodology adopted by Mr Webb in using the likelihood ratio rather than the random man non‑exclusion method, the number of loci used in making the comparison between the crime scene DNA and that of the appellant, the question of what should be the FST correction factor, the issue of the NRC Recommendations (although this seems to me to have been an issue of no great substance in circumstances in which the search of the police databank was not that ultimately relied upon) and, most importantly of all, in failing to make a sufficient allowance for the prospect that an untested relative or relatives of the appellant who lived in the general area might also have produced an equivalent DNA match.
The conflicts between the evidence of Mr Webb and Dr McDonald were significant in some respects, in my opinion. As will be apparent, this was a case in which the likelihood ratio comparing the prosecution and defence hypotheses was fundamentally important. There was no direct evidence implicating the appellant. KM's description of him, while "adequately" matching his appearance, was flawed in some respects and, as Roberts‑Smith JA points out, could have matched that of many others. I have said that KM had been unable to identify the appellant from a photoboard. While the appellant did live within easy reach, by car, of the town where the offences were committed, he denied having been there and, as I have mentioned, no items of stolen property or of clothing worn by KM's assailant (of which KM gave a reasonably detailed description) were found at a search of his home, conducted some two weeks after the commission of the offences. Consequently, it became critical to evaluate the strength of the statistical evidence advanced by the prosecution in answering the question whether the evidence as a whole was sufficient to establish the guilt of the appellant beyond reasonable doubt.
In my respectful opinion, that evaluation required some assessment of, and the making of findings concerning, Dr McDonald's evidence in respect of the concerns raised by him, more especially as regards what FST correction factor should be applied in the case of the Australian Aboriginal population group (it seemingly having been common cause, as I have said, that the appellant was partly Aboriginal) and what allowance should be made for the prospect that another relative of the appellant than those tested might produce a match with the crime scene DNA. It will be apparent from what I have earlier said that, in Dr McDonald's view at least, those factors made a very significant difference to the calculation of the likelihood ratio or to a calculation using his favoured random man non‑exclusion method (and it may be significant, as regards the issue of allowance for relatives, that the State had considered it necessary to test some of the appellant's relatives). However, no assessment of Dr McDonald's evidence was made by the trial Judge, or of the impact which acceptance of his views might have on the overall weight of the

prosecution case. Nor, as I have said, were any reasons given for rejecting Dr McDonald's evidence.

As I have stressed, the trial Judge did say that he accepted the evidence advanced by the State. However, if, by that, he meant that he preferred the evidence of Mr Webb to that of Dr McDonald (and it is not clear to me what his Honour meant by saying that there was "no doubt quite proper debate about the calculation of likelihood ratios and the factors to be studies or included"), then, as I have also said, it was incumbent upon him to say in what respects he preferred that evidence and why, in each respect, that was so. His Honour having failed to do so, and this Court being in no position to make any confident evaluation itself without having the opportunity to see and hear the two witnesses, it seems to me that the appeal must be allowed and that there is no scope for the operation of the proviso.
This conclusion makes it unnecessary for me to deal, separately, with any of the other propositions made on behalf of the appellant and referred to above.

Conclusion

I would consequently give leave to appeal, allow the appeal and order a retrial.
ROBERTS-SMITH JA: This is an application for leave to appeal against conviction. I shall refer to the applicant as the appellant. On 13 to 15 December 2004 he was tried in the District Court at Bunbury before H H Jackson DCJ, sitting without a jury. On 1 February 2005 his Honour convicted the appellant of one count of aggravated burglary, in company, of a habitation, contrary to s 401(2) of the Criminal Code (WA) ("the Code"); one count of sexually penetrating a child between the ages of 13 and 16 years, contrary to s 326 of the Code; and three offences of indecently dealing with a child between 13 and 16 years of age, contrary to s 321(4) of the Code. On 14 April 2005 his Honour sentenced the appellant to 12 months' imprisonment on count 1, 4 years' imprisonment on count 2 and 12 months' imprisonment on each of the three counts of indecent dealing. He ordered that the sentences be served concurrently. There was thus an aggregate sentence of 4 years' imprisonment. His Honour made an order that the appellant be eligible for parole.
The offences were committed at a town near Bunbury in the south‑west of the State in the early hours of the morning one night in

February 2002. The child complainant ("KM") was 14‑years‑old at the time.

DNA was recovered from certain items associated with the offences. When those DNA profiles were run through the police database, they came up with what was described as a "cold hit" on the appellant, a sample of whose DNA was on the database as part of an ongoing operation in the Bunbury area. Comparative tests were subsequently run against the crime scene DNA and samples from the appellant, and when they matched, the police arrested and charged the appellant.
It was common ground on the appeal that there was very little other evidence against the appellant in respect of these offences and indeed that without the DNA evidence there was no case against him.
There is only one ground of appeal, but it is supported by 16 prolix particulars. As they essentially express the appellant's arguments in support of the application, I set them out below:
"1.The verdicts were unreasonable, or could not be supported having regard to the evidence, in that it was not open to the Learned Judge upon the whole of the evidence to be satisfied beyond reasonable doubt that the Appellant was guilty.
Particulars
(a)The critical issue in this trial was whether the offender was the Appellant.
(b)The State relied on DNA evidence to prove the identity of the offender.
(c)The Learned Judge in his written judgment did not determine whether the interpreted DNA found on the inside left of the complainant's pyjama top contained 6 or 9 loci that were the same as the Appellant's DNA. The statistical calculation of the likelihood probability ratio in regard to the DNA on the inside left of the complainant's pyjama top being the Appellant's as opposed to someone else was significantly different, depending on whether 6 or 9 loci of the interpreted DNA found on the inside left of the complainant's pyjama top were determined to be the same as the Appellant's DNA.
(d)The Learned Judge in his written judgment did not indicate whether he accepted or preferred the evidence of the DNA expert called by the State (Mr L G Webb, Forensic Scientist), or the evidence of the expert called by the defence (Dr B L McDonald, Molecular Geneticist) concerning the statistical likelihood probability calculations in regard to the DNA evidence.
(e)Dr B L McDonald gave expert evidence that pursuant to the random man not excluded method of calculating DNA statistical probability likelihoods, and following the National Research Council's guidelines, that the probability of the DNA found on the inside left of the complainant's pyjama top being the Appellant's as opposed to someone else was one in six.
(f)The Learned Judge in his written judgment did not make any determination in regard to what he found to be the statistical probability or likelihood that the DNA found on the inside left of the complainant's pyjama top was the Appellant's as opposed to someone else. The Learned Judge did not indicate in his written judgment whether he accepted the statistical likelihood probability calculations of the State, the defence, or neither in regard to his findings.
(g)The Learned Judge in his written judgment did not determine whether the statistical likelihood probability calculations relied upon by the State concerning the DNA evidence were flawed or needed reducing, due to the State not factoring into the said calculations a statistical correction co‑efficient taking into account the perpetrator of the offences was possibly a relative of the Appellant.
(h)The Learned Judge in his written judgment did not determine whether the statistical likelihood probability calculations relied upon by the State concerning the DNA evidence were flawed or needed reducing, as a result of the said statistical calculations not taking into account a required correctional co‑efficient or FST in regard to sub‑populations (including the Aboriginal sub‑population) having an increased chance of common ancestory [sic] and therefore an increased likelihood of shared DNA within that sub‑population. The expert DNA witness called by the State, Mr L G Webb, Forensic Scientist, when giving evidence conceded a FST calculation incorporating common ancestory [sic] should have been factored into the statistical likelihood probability calculations of the DNA found on the inside left of the complainant's pyjama top being the Appellant's as opposed to someone else. Mr Webb gave evidence the FST calculation was not factored into the State's statistical probability likelihood calculations due to the State not having the relevant technology for this corrected calculation to be done.
(i)The DNA material found on the inside right of the complainant's pyjama top contained DNA material of someone other than the complainant and the Appellant.
(j)The integrity of the DNA material found on the inside left of the complainant's pyjama top was compromised by an unexplained recent 17 cm tear under the right armpit of the said pyjama top. This tear was not mentioned by the complainant in her statement which was read into evidence. The said tare [sic] was not noticed by the police forensic investigator, Peter Lloyd Webster when he collected and took as an exhibit the said pyjama top.
(k)The Learned Judge erred in finding in his written judgment that the Appellant denied in his evidence having previously given DNA samples to police in the light of the evidence of Sgt. Howes. The evidence of Sgt. Howes and the Appellant was in agreement that the Appellant had supplied DNA to police at his house, 28 Picton Road, Bunbury in January 2002, and at Bunbury Regional Prison in August 2002. Sgt. Howes gave no other evidence concerning any other DNA samples being taken from the Appellant. The Appellant in his evidence said prior to 16 January 2002 he had not given any DNA samples to police.
(l)The complainant was unable to identify the Appellant through photo board identification or through voice identification attempts.
(m)Despite the substantial amount of property being taken from the complainant's house on 13 February 2002 none of these stolen items were located at the house of the Appellant when police searched it on 27 February 2002.
(n)No clothes fitting the description of the clothes worn by the offender on 13 February 2002 were found when police searched the house of the Appellant on 27 February 2002.
(o)The complainant's description of the offender and the identity kit composit [sic] photograph prepared by police with the complainant's assistance are different in height, build and physical features than the Appellant.
(p)The Appellant and the complainant did not know each other when the offence occurred. The Appellant lived in Bunbury and the complainant [in a nearby town]. The Appellant when interviewed by police on video in 2002 stated that he had last been to [that town] as a small child. This evidence was similar to the evidence given by the Appellant at his trial."

At the hearing, counsel for the appellant informed the Court that particular (i) was no longer being pressed and that particular (j) was abandoned.
There was no dispute at the trial that the offences had been committed. The only issue in contest was whether or not the appellant was the offender.
It is not necessary to recount the facts of the offences in any detail beyond that limited extent required to give context to the ground of appeal.
KM and her mother were living in the house with the mother's male partner, who was not home that night.
A few minutes past 3 am, two men entered the home with the intention of stealing. While one man removed items of value, the other took KM into her bedroom, threatened her whilst armed with a weapon, and took the opportunity to sexually violate her. At one point he kissed and sucked her breasts, leaving a mark on her left breast.
Police subsequently seized KM's pyjama top from her bedroom. A sample of DNA material was obtained from the inside chest area of the pyjama top. A mixed DNA profile was obtained. When it was run through the police database that profile was found to be consistent with the DNA of KM and the appellant. Further reference samples were taken from the appellant and they formed the basis for the subsequent court identification of the appellant as the offender.
The police conducted a video record of interview with the appellant. He denied committing the offences. The appellant gave evidence at his trial. He denied committing the offences and said he had been in Bunbury at the time. Apart from the DNA evidence there was no physical evidence connecting him with the offences. There was no other evidence showing he was in the area at the time.
Statements by KM and her mother were read into evidence at the trial. Neither of them testified.
The mother's evidence adds nothing pertinent to this appeal. She remained asleep in the house during the offences and until the following day.
KM's evidence was relevantly that during the night she heard noises and saw a torch light in the house. She thought it was her mother. When she looked into the dining room she saw two people standing there. There was one man holding a torch. He was wearing a black hooded jumper with a black cap underneath. He had the torch light shining in her eyes so all she could see was a shadow. He told her to go back to bed. She went back into her room and they told her to lie down. She did so. She then describes being sexually assaulted by the man without the torch. He was trying to shield his eyes from her but when she moved over on the bed she saw everything above his eyes. He was not wearing anything on his head. There was no lighting in her room at all except for street lighting on the corner outside the house, but there was only a bit of light coming from outside as the blind was about 15 cm up from the bottom of the window.
That man had an oval shaped head and black hair with grey through it. It was not that noticeable but because he had short hair she was able to see it at the front. It was silvery grey. She saw what looked like the top of a beard at the top of his cheeks but could not tell if he actually had a beard or not. She could not see the bottom half of his face. He had bushy dark eyebrows and dark brown eyes. His eyes were smallish and oval shaped, but not oriental. She said she did not know if he was Aboriginal or not because he was neither dark nor white skinned as far as she could tell. He sounded similar to the other person but she was not able to say he was Aboriginal.
The one with the hooded jumper, standing by the door of her bedroom, asked where the money was. She described his voice as "kind of husky and he had a kind of Aboriginal accent. He sounded really uncultured."
Later, when she was speaking to a police officer about what had happened to her, she said that the man who sat on her bed was about early to mid‑twenties, fairly skinny build and about the same height as that police officer. (It was established in evidence that his height was 178cm). She again said he had short black hair with grey through it and dark bushy eyebrows and dark brown oval shaped eyes. He had facial hair which was fairly prickly and also smelt of tobacco and had rough‑feeling hands. She said he "talked feral" and had an olive or tanned complexion. She said he was wearing black pants which did not have any zip so she was pretty sure they were not jeans. She was able to say what he had on underneath because removed his clothing during the incident. He was wearing black and red silky boxers. He also had a black weatherproof type jacket that kept swishing as he walked.
KM made another statement following an incident which occurred in early March 2002 while she was at work in Bunbury. She saw a man walking past by himself and thought to herself that he looked familiar. She had to continue serving customers but when she took a break about 10 or 15 minutes later she thought about it again and realised that he looked like the man who had assaulted her, especially his eyes. She said in a subsequent statement that he did look like his hair was cut a little bit shorter. She subsequently told the police.
In his reasons for decision (R v Riley [2005] WADC 16) the Judge set out the statements made by KM almost in their entirety.
His Honour noted that it was not disputed that a significant amount of property had been stolen from the house, some of which was later recovered in Perth. None was recovered from or in the possession of the appellant or at his home.
His Honour referred to evidence as to the police inquiries which included the preparation of six photo boards from which KM was asked if she could identify either man. The appellant was one of those shown. She was not able to identify either of the offenders. His Honour noted that the police had, with KM's assistance, prepared an identikit likeness of the offender which his Honour said showed "likenesses" to a photo of the accused put in evidence and with the photo of him on the photo board (which was a smaller copy of the first), "… but could not be said to replicate his appearance in all respects".
From his observations of the appellant in court over the three days of trial, the Judge described the appellant as being stocky and brown‑skinned with brown eyes and curly dark brown hair, appearing to be in his early thirties and not particularly Aboriginal in facial features. Nor did he speak in a way peculiarly associated with "Aboriginal English". His Honour then said (at [22]):
"Whilst in the exhibits mentioned, he is shown with a moustache and beard but short hair, at trial he was clean shaven and his hair longer. Nonetheless, the description given by the daughter does, given her age and the frightening circumstances at night in which she observed him, adequately match his appearance in many ways. In saying this, I note that his height of 170 (or 172) cms is less than that of Sgt Anderson, who described his height as about 5' 10" or 178 cms."
His Honour continued (at [23]):
"Of course given her lack of identification either from the photoboards or (which were not attempted) by dock identification, or in an identification parade, such a description could not satisfy me beyond reasonable doubt of his being the perpetrator. Nor was there evidence of admissions by him, of finger prints, of prior association or other connection with the scene, or with the complainants."
His Honour said the matter turned on the DNA evidence and he proceeded to deal with that.
Before dealing with what his Honour said about that, it is necessary to refer to the evidence concerning it. It is also apposite at this point to make some observations about the way in which the appellant's case was presented on the appeal.
It is apparent that the appellant's counsel did a great deal of work in preparing his case. The outline of submissions runs to 36 pages. Attached to that is a 40 page annexure. The annexure is headed "DNA articles presented to the court and expert witnesses but not put into evidence". They include, for example Karen L Ayres, Janet Chaseling and David J Balding, "Implications for DNA identification arising from an analysis of Australian forensic databases" (2002) 129 Forensic Science International 90; Abstract: I Birus, M Marcikic, D Lauc, S Dzijan and G Lauc, "How high should paternity index be for reliable identification of war victims by DNA typing?" (2003) 44(3) Croatian Medical Journal, 322; Ian W Evett, Lindsey A Foreman, Graham Jackson and James A Lambert, "DNA Profiling: A discussion of issues relating to the reporting of a very small match probabilities" (2000) Criminal Law Review, 341; Andrew Ligertwood, "Avoiding Bayes in DNA cases" (2003) 77 Australian Law Journal 317; Bruce S Weir, "Matching and Partially‑Matching DNA Profiles" (2004) 49(5) Journal of Forensic Sciences, 1.
In his outline, counsel also quoted extensively from the judgment of Mazza DCJ in R v Bropho (2004) 36 SR (WA) 328 in support of various of the particulars. That included (at [116] ‑ [120]) part of his Honour's explanation of DNA itself and forensic DNA testing and analysis, as well as observations and findings of his Honour about the DNA evidence in the case before him. Counsel, in effect, sought to incorporate into his submissions as material upon which this Court could rely, much of the content of these articles and the judgment of Mazza DCJ in Bropho.
These were not matters of which judicial notice could be taken. Nor can a Judge's recitation of evidence and findings of fact by that Judge based on that evidence, be relied upon by another court in another case before which different evidence has been adduced.
At common law, in taking judicial notice of particular facts, a court may consult a variety of sources. However these are confined to sources such as reputable dictionaries (Bendixen v Coleman (1943) 68 CLR 401; Chapman v Kirke [1948] 2 KB 450, 454) and other recognised reference works (Australian Communist Party v The Commonwealth (1950) 83 CLR 1, 196). Section 72 of the Evidence Act 1906 (WA), allows a court, in matters of public history, literature, science, or art, to refer, for the purposes of evidence, to such published books as they consider to be of authority on the subjects to which they relate. The section has not been taken to involve any significant extension of the common law.
In "Cross on Evidence" (7th Aust Ed) the position in respect of contentious material is explained at [3035]:
"The judge, when undertaking such a reference to works of authority, either pursuant to the statutory provisions or at common law, is not entitled to obtain information on any contentious matter without giving the parties the opportunity to controvert or comment upon the work of which reference has been made (In the Marriage of Dean (1988) 94 FLR 32 at 37 (Fam C of A FC); Malaysian Airline System v Wood [1988] WAR 294 at 303 (FC); Gordon M Jenkins & Associates Pty Ltd v Coleman (1989) 23 FCR 38; 87 ALR 477 (Fed C of A); Australian and Overseas Telecommunications Corp Ltd v McAuslan (1993) 47 FCR 492 (Fed C of A FC)).
'It is one thing to use the section [s 64 of the Evidence Act 1929] for the purpose of discovering or verifying objective facts or figures about which there can be no real dispute, such as historical or geographical data of an uncontroversial nature or mathematical tables of life expectancy or interest calculations …. [E]ven (Arnold v Norris [1936] SASR 287) there I think … the parties should be given notice of the Court's intention and an opportunity to be heard on the result of its researches. It is quite another thing for the tribunal of its own motion to seek to inform itself out of court on a question of fact or opinion vital to the issue and by no means free from controversy. This is, of course, doubly objectionable if done without the parties having any opportunity to state their views on the specific authorities consulted. But in my view, even if they are offered this opportunity, the court should not embark on such investigations except by consent. It would be preposterous to suppose, for example, that in a claim for damages for workmen's compensation where divergent medical opinions have been expressed by expert witnesses on each side, the court should be at liberty without consent to pursue independent inquiries of its own on the point through medical journals or text books not referred to by the witnesses (Cavanett v Chambers [1968] SASR 97 at 101. See too: Horman v Bingham [1972] VR 29 (position of seeds on cannabis plant); Fairbank v Jones (1975) 10 SASR 367 at 370‑1. See [3130]‑[3140] below.)'"
In the present case, the issues raised in respect of the DNA evidence were central to the contest between the appellant and the State. The evidence of the DNA experts was opposed on critical aspects. The content of the articles sought to be relied upon on the appeal patently was not material of which the court could take judicial notice, nor to which it could have regard, unless and to the extent it was referred to by the witnesses and agreed to or adopted by them. The particulars to the ground of appeal must therefore be approached having regard to the evidence actually given before the Judge.
The State relied upon the expert evidence of Mr Laurie Webb. Mr Webb is a senior forensic scientist in forensic biology at the PathCentre in Nedlands, Western Australia. He is currently in charge of the major crime area.
Some of the contents of the articles were put to Mr Webb in cross‑examination but he was not prepared to agree with them, nor that they were accurate, and in fact expressed reservations about, and even disagreed with them. Specifically, he declined to adopt what was put to him from them. Those materials are therefore not evidence on this appeal.
In the course of Mr Webb's evidence‑in‑chief, the State prosecutor tendered a report by him dated 22 July 2002 (Ex 25A) together with two addenda, dated 26 September 2002 and 23 July 2003 (Ex 25B and 25C respectively). The relevant findings concerning the pyjama top were expressed in the narrative to the report as:
"PW 17 Pyjama top: The blue, Now brand, size 8 long sleeved button down shirt with a stars and moon pattern was in a worn and soiled condition.
One button was missing and another loose. A 17cm recent tear was noted under the right armpit.
The inside of the left and right chest areas were sampled for cellular material and submitted for DNA analysis. A mixed DNA profile consistent with having come from two people was recovered from the cellular material from the inside right of the pyjama top.
Amylase, a component of saliva, was not detected in this sample.
The mixture could be separated into major and minor components. The major component DNA profile matches [KM's] DNA profile.
The minor component is too weak to provide a meaningful conclusion.
Amylase, a component of saliva, was detected on the sample from the inside left chest area.
A mixed DNA profile consistent with having come from two people was recovered from the cellular material from the inside left of the pyjama top. The results appear below."
A summary of DNA typing results were set out in tabular form. The results insofar as they relate to the appellant, KM, and the pyjama top sample, were:

Name/Description	D3S1358	vWA	FGA	Amel	D8S1179	D21S11	D18S51	D5S818	D13S317	D7S820
[KM]	17/18	16/16	24/24	X/X	12/13	30/30	14/15	12/13	8/11	10/13
RILEY P J	14/19	17/18	21/22	X/Y	12/16	30/30	17/20	11/13	8/11	8/13
- PW 17 Pyjama top-inside left (mixture)	14/17/ 18/19	16/17 /18	21/22 /24	X/Y	12/13 /16	30/30	14/15/ 17/20	11/12 /13	8/11	8/10 /13

Mr Webb's description of his conclusions about the pyjama top was as follows:
"PW 17 Pyjama top
A mixed DNA profile consistent with having come from two people was recovered from the cellular material on the inside left chest area of the pyjama top.
If it is assumed that there are two people contributing to the mixture, then it is at least 10 billion times more likely to find this mixed DNA profile if it comes from [KM] and Percy James RILEY than if it comes from [KM] and an unknown."
According to Mr Webb's evidence, DNA is basically the genetic building blocks of one's own body. It is contained within every cell except for circulating red cells which do not have a nucleus. Otherwise, the DNA is contained within the nuclei of the cells of the body and "wrapped up" tightly in the chromosomes. DNA is the genetic material which determines the physical characteristics of an individual. It is inherited from a person's biological parents. For an individual, the DNA throughout the body is the same; but between individuals the DNA is quite unique. When speaking of the uniqueness of DNA, Mr Webb said he was referring to the full genome, which spans out to about a metre in length, so that if one was looking at it microscopically there would be a vast number of particular sites or loci. Different individuals may have matching DNA at the same loci, but so far as science is aware, no two individuals other than identical twins can have the same full DNA profile - that is, matches across all loci in the genome. Forensic DNA examination involves looking at particular sites (loci) on the molecule which have been chosen because they vary greatly between individuals. The PathCentre currently looks at 10 different sites. At each of those sites they then determine the DNA profile using a system called "Profiler Plus". That is a commercially available kit which has been established world‑wide and validated in the PathCentre itself, as well as around the world. There are other systems available which allow examination and comparison of more sites and in fact the PathCentre is presently looking at a system which would allow them to look at 16 loci simultaneously. Within that field of forensic expertise, there is general acceptance that a match at 10 loci is determinative. One of those, however, is the amelogenin locus which actually determines gender (with either two X chromosomes or an X and a Y chromosome) and which therefore cannot otherwise be used for the purposes of comparison. The other nine loci are differentiating loci.
An individual inherits one part of the genetic information ("allele") from each of the parents. Thus, each of the loci will have at least one, but at most two, of the particular alleles inherited from the parents. If there is a common allele from both mother and father, that will show as a single allele, but in fact there will be actually two which are exactly the same.
The particular alleles are given numerical values which refer to the actual repeating units within the DNA. As an example, Mr Webb explained that at one site a person might have 11 from the mother and 11 from the father, which would be seen as one peak on a computer generated chart and that would be an 11.
DNA profiling is essentially exclusionary - that is to say, if at a particular locus there is an allele which is the same as that of a person, then it is said that person is "not excluded" from being the source; however if there is an allele at any locus which is not present in the person's profile at that locus, then the person is excluded from being the possible source of the sample.
Mr Webb described the process which was followed once DNA was obtained from cellular material found at a crime scene.
Once the DNA is extracted from the cells and is in solution, it is possible to quantitate how much DNA is present. A system referred to as "polymerase chain reaction" (PCR) is then used to amplify or replicate the DNA at the selected site. That is then passed through a genetic analyser, which is a machine which incorporates a capillary for the actual determination of the DNA profile. The analyser is based on fluorescent technology and the DNA is effectively able to be seen on a graphical representation, or electropherrogram, which gives series of peaks in graphical form. There are then interpreted by two independent scientists, as well as the genotype program in the machine itself. Once verified, the result is loaded into the DNA database system.

There are strict rules about whether or not an allele can be identified from an electropherrogram result. There must be at least 50 reference points before a particular result can be identified as an allele. There are similar rules about a homozygote peak, which is basically when the two genetic information pieces that have come from both parents are the same, to produce one large peak at the particular locus. He explained other rules governing what the laboratory can or cannot call as being part of a DNA profile.
If there is a mismatch between the crime scene sample and the subject sample, then the person who is the source of the subject sample is excluded. If there is a match, a weight is then attached to that in statistical form using a Western Australian population database.
The population database used by the PathCentre comprises three data sets - Caucasian, Aboriginal and South-East Asian. They are based on Bureau of Statistics figures in relation to the major racial groups within the State. An internationally recognised computer program is then applied to produce a probability or likelihood ratio.
Where the sample is a single DNA profile, the result will be expressed in terms of the probability ratio, namely the probability of the particular DNA profile occurring and it being from the suspect. If the sample is a mixture of DNA profiles, the likelihood ratio will be used. That relies on the prosecutor's hypothesis and the defence hypothesis. The former is an assumption that the suspect is the source of the crime sample; the latter includes everyone in the population. The ratio between them is expressed as the "likelihood ratio".
Mr Webb described the forensic investigative process which resulted in the comparison described in his report. The crime scene sample produced a mixture of DNA profiles. The sample came from at least two people and for reasons which he explained, Mr Webb assumed it was not likely to be more than two. One of them matched KM's DNA profile. Given that the sample came from her pyjama top and the circumstances in which it was found, he was prepared to assume it was her DNA. He then interpreted the second DNA profile from the mixture and interrogated that against the PathCentre database. At that time it contained about 8000 DNA profiles of reference samples from persons of interest, suspects, witnesses, laboratory staff, deceased persons and crime scene profiles.
Two matches came up on the system and they were both reference samples from the appellant. There was no other match.
Once that match was made, further buccal samples were obtained from the appellant pursuant to the Criminal Investigation (Identifying People) Act 2002 (WA) and his full DNA profile was then compared to the crime scene mixture in total. The result was the appellant could not be excluded as contributing to that mixture. Application of the statistical program then produced the likelihood ratio result mentioned above.
In this case, Mr Webb explained that the prosecutor's hypothesis was one alleging that the appellant was part of the mixture together with KM. The defence hypothesis was that there were two unknown persons in the mixture.
It should be mentioned that DNA samples from three of the appellant's male relatives were also tested against the crime scene sample and each was found to be excluded.
The profile checked against the database in the first instance was based on alleles identified at seven loci. According to Mr Webb a minimum of six loci are required for an interrogative search. A match at seven loci is certainly acceptable within the National Crime database protocol. As he explained however, the actual comparison of the crime scene sample to the accused's DNA upon which his evidence was based used a comparison at 10 loci.
Mr Webb testified that he had actually recalculated his statistical findings on the likelihood ratio to an assumption that the mixture came from three persons rather than two. The result then was that it was at least one billion times more likely to find that mixed DNA profile if it came from KM, the appellant and an unknown person, than if it came from KM and two unknown persons. On that assumption, the likelihood ratio reduced from 10 to one billion times - which he said he considered to be a "very conservative view".
In cross‑examination Mr Webb said he was aware that in Germany, two people had been found to have DNA that matched 10 loci; in New Zealand four people matched 10 loci and in the Northern Territory there is a brother and sister who have a match at 10 loci. However, he said that although the PathCentre has had (different) people with matches at six loci, and the Centre has the largest searchable DNA database in Australia, they have not experienced any greater matches than that.
Counsel for the appellant put before Mr Webb a copy of a letter written by him to counsel dated 16 February 2004 (later tendered as Ex 26). He drew Mr Webb's attention to a paragraph in that in which Mr Webb wrote that genotypes at three loci (D8S1179, D5S818 and D7S820) could not be conclusively assigned to the interpreted DNA profile, as he could only be certain about one of the alleles at each locus. There were accordingly only seven loci (including AMEL, the sex determinant locus) which could be searched against the database. Counsel put to Mr Webb that this was different to the findings in respect of those loci in his report Ex 25A and in his evidence.
Mr Webb explained that they were two different things. What he was reporting in Ex 26 was the interpreted results from the crime scene sample. By "interpreted" results he meant the DNA profile in the mixed sample which he had "interpreted" by removing the DNA he knew to be that of KM. What was left he took to be that of the offender. Alleles were identified at only seven loci. The database search on those produced a match with the appellant. The results reported upon in exhibit 25A were of a comparison between the full profile (not just the interpreted) crime scene mixed sample and the buccal swab reference samples taken from the appellant, across 10 loci - all of which matched. He disputed that the different results were inconsistent, pointing out that the statistical conclusion was based on the full DNA profile of the mixture and the appellant's reference sample. Statistical reports are not done on database searches used as an investigatory tool.
Asked later about his statistical calculations, Mr Webb explained (AB 164) that the calculation of one in 10 billion was "the most conservative":
"In relation to the mixed DNA profile recovered from the inside left of the pyjama top, PW17 - in relation to the mixture, the likelihood ratio for the Caucasian population was 5.6 by 10 to the 11, so that is 566 billion. The likelihood ratio for the Aboriginal population was 434 billion, and the likelihood ratio for the South‑East Asians was 1571 billion; in other words, 1.5 trillion.
So how did you come by the 10 billion number?---That is a figure that we use in our laboratory as a conservative figure; the 10 billion in relation to the world's population as a relationship. We currently don't report the actual figures because they are quite extreme at the best. We have a cut‑off value of 1 in 10, or a probability cut‑off of 10 billion, and in the same way in our relationship with the mixture calculations. So you can see we've referred to three data sets and I've given in fact a very conservative figure in relation to that because the most conservative figure in this case was in fact from the Aboriginal population, which was 434 billion.
Are you aware that in other databases in the world they have a lower cut‑off rate than ‑ ‑ ‑?‑‑‑I'm aware of that. They may do. That's their choice.
Are you aware that in London for instance on a 10 locus match, they have a general figure of 1 in 1 billion for a 10 locus match?‑‑‑Yes.
And the PathCentre figures are simply different, are they, based on different criteria?‑‑‑Most states in Australia actually report the actual values. We report 10 billion."
Mr Webb agreed that there is a factor which can be applied to account for common ancestry within populations. This is called the "FST value". He described it as a correctional co‑efficient for common ancestry within subpopulation groups in the annual population data sets. He said the figure currently used by the PathCentre in relation to single set profiles is 0.03. He said the Centre does not routinely perform that calculation, but the conservative figure taken in this case would certainly account for any relatives in the database. The PathCentre does not use the FST value in mixture calculations because the current mixture profile calculation program technically does not allow for FST values to be inserted. He explained further that he was not prepared to make a calculation using an FST value of 0.03 or give a ball park figure based on that, because it was not just a matter of applying that figure across the board; one has to look at each locus independently and it requires a fairly complex program to do that, which he did not have available to him. Asked whether he thought it would give a significant reduction in the likelihood ratio, he pointed out that the actual value for the mixture statistical calculation was in fact 434 billion. He had reduced that to 10 billion just by the very nature of the conservative reporting applied by the PathCentre. They do not report figures higher than 10 billion, but that would more than account for any foibles within the population data sets given that if the FST correction were made, it would be applied to the actual value, which was here 434 billion.
Asked whether the reason the FST correction factor is there is because of a wide acceptance that many sub‑populations, due to long years of interrelationship, cause a likelihood of shared alleles among the sub‑population, Mr Webb said that was correct but there were different schools of thought and no agreement between statisticians as to what the correct FST value is. Some consider the 0.03 is very conservative; in fact 1 per cent is probably more to the point. Work is currently being done around Australia on this and the indications are that the FST value is more likely to be closer to 1 per cent than to 3 per cent.
Asked whether there was internal PathCentre documentation which indicates the PathCentre will adopt a value as high as 0.13 if it is relevant, Mr Webb said there is nothing to that effect currently in their method manual. Reference had been made to it but it was subsequently shown that the FST value of 0.13 given by the authors of a particular paper was far from the correct value. Those authors were dealing with endogenous communities whose peoples were actually practising inbreeding as a part of their religion and so one would expect to find that most of them were actually related. He reiterated that there is a huge area of dispute as to what the actual FST value should be in regard to Aboriginal sub‑populations, and indeed that it is unknown at this time.
Taken again to reports of different people with matching DNA profiles, Mr Webb variously observed that in one case when the matches were investigated, a number of them were found to be in fact the same person but from different jurisdictions, others were identical twins; an extract of an article reporting 39 people had seven full loci matches was problematical because the methodologies were not explained and they came from an examination of skeletal remains in Croatia which were likely to involve significant contamination.
Mr Webb was then asked about the Random Man Not Excluded ("RMNE") methodology. He explained that looks at the probability that anyone at random may have contributed to the crime scene DNA profile. It asks what is the probability that any person could contribute to the profile. That was not the same methodology or calculation which he had used. He said the PathCentre used it as a guide but do not include it in their reports. There is still no real consensus in the forensic laboratory field as to how the RMNE approach is to be used. Each allele is interpreted and looked at in the general population, comparing it to the three data sets in the statistics. He had done the three calculations for the mixed sample in this case. The RMNE score probabilities were calculated as being Caucasian - one in 453,700; Aboriginal - one in 668,900; and in South‑East Asian - one in 3,097,000. The FST correction was not taken into account in those calculations.
Dr Brian McDonald is a molecular geneticist with a PhD in Pathology. He works as a private consultant. He was called by the appellant.
The major point Dr McDonald made was that the identification of the appellant had not come from evidence other than the database search. Normally, what would happen, he said, is that evidence regarding a crime would be gathered and would identify an individual. By virtue of that identification a DNA sample would be taken and compared with the crime sample. In this particular instance however, the appellant was identified not through other evidence but directly from the DNA hit on the database. Dr McDonald said there were a couple of possible explanations for that. What he described as the two fundamental ones were that it was not the appellant but someone else. In that situation the most likely person that would be would be one of his relatives, because relatives shared DNA much more frequently than someone picked at random from the unrelated population. He said the one in 10 billion figure only addresses the theoretical population, by picking out numbers from a frequency graph and multiplying them together. He said no attempt had been made to indicate the likelihood of relatives of the appellant having the same combination, which required different types of calculation. The second was that it was someone other than the appellant and not being one of his relatives. That would involve calculating a probability ratio based on RMNE methodology.
In relation to the calculation if the offender were not the appellant but one of his relatives, Dr McDonald said he had not done the calculation but it would increase the likelihood that there would be a match. He said there is a standard formula which enables calculations to be done without actually knowing what the DNA profile of the relatives is. He distinguished between that and the FST correction factor which he explained allowed for a certain level of relationship within an otherwise unrelated database, but which would not take account of relatives.
He was asked whether there was a difference in the FST correction for Aboriginal people and Caucasians, to which he responded (AB 205):
"Well, there's been no studies on the FST for Aborigines, and the only study that I know of, which is in one of the papers that was quoted, the Ayres paper, actually pooled three different subgroups of Aborigines together and averaged them before they started, so that in effect obliterated the purpose of the exercise. So there have been no studies on Aborigines. The FST figure of 1 per cent for Caucasians which, following analysis of the database where they saw all these matches - it was up, to be conservative, to 3 per cent. As I say, it's based largely on Caucasian data. The very few studies on indigenous people - and the ones quoted I think were in the American Indians - you're looking at 9 to 11 per cent. In New Zealand, they use 5 per cent for Maoris and Polynesians. I've asked to see that data but I've never seen it, so presumably they have data that shows that they get matching a lot more frequently than you would expect by chance, and so they've had to increase their figure."
As his Honour noted in an exchange with the witness, the FST factor is not based on any racial characteristic but is connected to family structure and familial relations. Dr McDonald said it is to do rather with mating patterns within a sub‑population. He reiterated that the data does not exist in respect of Aboriginal populations notwithstanding that (AB 206‑207):
"… the Australian Aboriginal population is really quite unique compared to any other in the world, and the sort of use, particularly of Caucasian type data, to apply to Aboriginal groups which clearly have evidence of tens of thousands of years of isolation - you know, to apply this sort of view that there's this randomly mixing homogeneous group from whom you can calculate some sort of figure of random occurrence of their genetic makeup is just missing the point."
The fundamental point of difference between Mr Webb and Dr McDonald was whether the calculations should be based on six loci or nine. Dr McDonald's view was that only the original six should have been used because that was the basis upon which the appellant was identified in this case. He explained it this way (at AB 210):
"… the real critical issue in this whole thing is the identification of Mr Riley. He was not identified by any other evidence that appears in this case. It is a consequence of his profile happening to be on this database that the West Australian police have, so now this evidence - part of this evidence was responsible for identifying him. Then we sort of seem to take on this new persona that his DNA must be in there. He's not excluded but being not excluded does not mean that his DNA is in there. So then we move on, and the term was used purely as an intelligence source, but it's the critical intelligence source. That was searched with six loci. Then we go on and then the next step is that whilst we only had six loci to identify him and we found someone who is not excluded - another situation could be where - this could be a complex mixture where 30 or 40 or 50 per cent of the population would not be excluded, but if we said because it identified one person who was not excluded, we will assume that the person is now a contributor, as opposed to being not excluded, so the issue is moving away from the evidence …"
Notwithstanding that, he said he was not making any judgment on whether six loci were enough or not; he was not advancing the proposition that was too small a set. He said the one in 10 billion figure only addressed the chance of matching a particular theoretical population; it had not actually asked the question "if it's not Mr Riley, who else could it be?" There has been no attempt at all to address the most likely people that it would be if it is not Mr Riley.
Dr McDonald said that for this mixture on the six loci - and not allowing any factor for Aborigines or relatives - the figure would be about one in 48,000, which is probably a ten‑fold reduction on Mr Webb's calculation of approximately 660,000. Furthermore, in this sort of situation in towns like this, relatives are the population amongst whom the most likely matches are going to occur. He added that if the appellant was identified as a consequence of evidence in the case, such as being picked out of a line‑up, and his DNA then taken and compared to the crime scene sample, the whole population figure could be used. However, that was not how he was identified. For some reason he was already on the database and was matched there, so the appellant immediately became the focus.
He said that working on the six loci, and allowing for relatives, applying the RMNE methodology to a database of 8000 people, multiplied by the RMNE figure of 48,000, the probability ratio becomes one in six. In other words, there would be almost a 100 per cent chance that someone other than the appellant in the 48,000 would not be excluded as a contributor to the crime scene sample mixture.
In cross‑examination Dr McDonald said he was not challenging any of the methodology that had been used by Mr Webb nor the results obtained; what he was challenging was the appropriate statistical figure and whether it ought to have been based on six rather than nine loci. His contention was that as the evidentiary sample produced identifiable alleles at only six loci, they were the loci which ought to have been used to do the statistical calculations, based on RMNE methodology. That produces a probability ratio, not a likelihood ratio.

It is submitted that the error made by his Honour was apparently in attributing Mr Webb's evidence of this to police witnesses.
Mr Webb's evidence otherwise was to the effect that the database contained DNA profiles from a wide range of sources; that included samples from convicted offenders, but also a large number of other people

and obtained in a variety of circumstances which did not necessarily have any connection with offending.

I accept that as a matter of fact the issue whether the appellant had given samples to police officers in 2000 and 2001 was of little importance as such. In that respect, all that was important was that his DNA was in fact on the database. However, the content of the questions put in cross‑examination was not evidence, and the conflict erroneously perceived by his Honour probably did have importance to him as going to the appellant's credibility and at least was a (if not the) determining factor in his Honour rejecting the appellant's sworn evidence that he was not in the town at the time of the offences. His Honour expressly linked the two. This was therefore a factual error which in all likelihood, at least in part, led to rejection of the appellant's evidence on a critical issue. That amounted to a miscarriage of justice and I would also uphold the ground on this basis.

Conclusion

This is not a case for the application of the proviso. The respondent has not shown that there was not a substantial miscarriage of justice. I would grant leave to appeal and allow the appeal.
As it cannot be said that it was not open to the Judge on the evidence to be satisfied of the appellant's guilt beyond reasonable doubt, there should not be a judgment of acquittal. I would order a retrial.
MILLER AJA: I have had the opportunity of reading in draft the reasons for judgment of Roberts‑Smith JA. His Honour has set out the facts of the case, relevant excerpts of evidence, and the grounds of appeal. None of this needs to be repeated.
The ground of appeal raised by the appellant contends that the verdicts reached by the learned trial Judge were unreasonable or could not be supported having regard to the evidence because it was not open to the learned trial Judge upon the whole of the evidence to be satisfied beyond reasonable doubt of the guilt of the appellant. This ground became an effective contention that the learned trial Judge failed to give adequate reasons for the decision which he reached.
The appellant faced an indictment which contained five counts. There was one count of aggravated burglary in company, one count of sexual penetration of a child between the ages of 13 and 16 years without her consent and three counts of indecent dealing with a child between the

ages of 13 and 16 years, each of which alleged a different means of indecent dealing.

The appellant was tried by Judge alone and the trial ran over three days. The learned trial Judge took time to consider judgment and on 1 February 2005 he convicted the appellant of all counts on the indictment. His Honour's reasons were extensive. They were approximately 30 pages in length and recounted much of the evidence in considerable detail. This was particularly so in relation to DNA evidence, which was the central contest at the trial.
The trial was really one of identification. The question was whether the appellant was the offender who had committed the offences and, in particular, those of sexual assault. The learned trial Judge found that there was no doubt that the offences alleged in the indictment had taken place on the date and at the place alleged and that the child was 14 years of age at the relevant time. As he put it, "the only issue has been whether the accused committed the offences or any of them".
The evidence of the complainant was received by the admission, by consent, of her written depositions. They revealed (relevantly) that on 13 February 2002 she was asleep in her room when she was awoken by torchlight flashing through a space under the door. She awoke at a time which she estimated to be 3 am. Because the light was annoying her, she got up on the assumption that her mother was in the dining room. When she looked in the dining room, she saw two people standing there. One, holding the torch, was wearing a black hooded jumper with a black cap underneath it. The complainant could only see a shadow because he was shining the torch at her. He told her to get back into bed. The second person told her she was dreaming. She went to her room and she was told to lie down. She was unsure who told her to do that. The man without the torch got onto the bed where she was lying. He sat on the bed, and when she looked at him he tried to shield his eyes from her. She saw everything above his eyes and could see that he was not wearing anything on his head.
There was no light in the complainant's room, but there was street lighting outside. This let in some light and it was sufficient for the complainant to identify that the man on the bed had an oval‑shaped head and black hair with grey through it. He had what looked like the top of a beard on his cheeks, but she could not tell whether it was a beard. He had dark, bushy eyebrows and dark brown eyes. His eyes were "smallish and oval shaped but not oriental". She did not know whether he was Aboriginal.
The man on the bed asked the complainant her age. When she said she was 14 years of age he said he did not believe her. He asked her whether she was a virgin, to which she answered in the affirmative. He then put his hand in her pants and "groped her", placing his fingers inside her vagina. The complainant was wearing blue, silky pyjamas at the time and she was not wearing underwear. The placing by the offender of his fingers inside the complainant's vagina constituted the offence the subject of count 2 on the indictment.
The offender then committed indecent assaults on the complainant. He first put his hand through a gap in her pyjama top and started to feel her breasts. He then opened up the top, leaned over and kissed her breasts. He sucked her and left what she described as "a hickey on my left breast which has left a mark". The next indecent assault was when the offender put his right hand down the complainant's pyjama pants again and "groped her vagina", but without putting his fingers inside her.
The final indecent assault occurred when the offender pulled down his pants and boxer shorts, which the complainant identified as being red and black in colour and silky, and he grabbed the complainant's left wrist with his right hand and forced it onto his penis, which became erect as she did so. He then pulled up his pants and sat on the bed, before leaving with the other offender.
In a second statement which was received in evidence and made about a month after the commission of the offences, the complainant said that on 6 March 2002, whilst working in Bunbury, she saw a man walk past the shop. He was alone. The complainant thought he looked familiar. She later realised that he looked like the man who had assaulted her, "especially his eyes".
The police investigation of the incident was extensive. Police prepared six photoboards from which the complainant was asked if she could identify either or both of two suspects. The first photoboard included a photograph of the appellant. The second contained no photograph of the appellant, but of different men, two of whom were named Riley. The third and fourth photoboards contained no photographs of any persons named Riley. The fifth photoboard contained a photograph of another Riley and the sixth photoboard contained a photograph of yet another Riley. The complainant was unable to identify any offender from the photoboards she was shown.
The complainant prepared an identikit likeness of the offender, but the learned trial Judge, having considered this identikit likeness, reached the conclusion that the likeness did not "replicate the appearance of the accused".
The learned trial Judge saw the appellant in court. He described him as being stocky, brown‑skinned with brown eyes and curly, brown hair, in his early 30s, and not particularly Aboriginal in facial features. Nor did he speak in a way peculiarly associated with what the learned trial Judge termed "Aboriginal English".
The learned trial Judge referred to the description of the offender and the identikit likeness as "a person of approximately 20‑25 years of age, Caucasian, olive/tanned, dark, short hair, slim build and approximately 175 cm tall". The learned trial Judge noted that the height of the appellant was 170 centimetres. This was eight centimetres shorter than a police officer whose height the complainant had thought matched that of the offender.
However, the learned trial Judge considered that notwithstanding that the description given by the complainant was somewhat inconsistent with the appearance of the appellant at trial, nevertheless "given her age and the frightening circumstances at night in which she observed him, [her description] adequately match[ed] his appearance in many ways".
In his reasons, the learned trial Judge concentrated on the DNA evidence which had been led at trial. On the inside of the complainant's pyjama top, material had been located from which DNA analysis could be conducted. The prosecution case was that on analysis, that DNA material proved to be that of the complainant and of the appellant. The prosecution case was that DNA from the pyjama top matched DNA analysis of material previously taken voluntarily from the appellant in connection with Bunbury police inquiries concerning local burglaries, and also samples taken from him after the commission of the offences against the complainant. DNA samples taken from Charles Douglas Riley and Paul Anthony Riley did not match the complainant's pyjama top.
The appellant was from Bunbury and the offences occurred at a nearby town. At his trial he said that in February 2002, he was working in Brunswick. The learned trial Judge found both Bunbury and Brunswick to be within easy reach, by vehicle, of the town where the offences occurred.
The learned trial Judge stressed that the question at trial was whether the DNA material taken from the complainant's pyjama top established beyond reasonable doubt that the DNA of the appellant had been left on that clothing. The evidence in relation to DNA has been extensively documented by Roberts‑Smith JA. Two witnesses were called in relation to the issue. They were Mr L G Webb, a forensic biologist, employed at the Pathology Centre at Nedlands and Dr B L McDonald, a molecular geneticist in practice as a private consultant.
The evidence of Mr Webb was comprehensively reviewed by the learned trial Judge. It established that the material taken from the complainant's pyjama top was sufficient for analysis and the appellant was not excluded by analysis as a person whose DNA it could have been. Nor did the DNA match the known DNA profile of any other person.
Using the standard Australian practice, the material from the pyjama top was tested at 10 loci, one of which determined gender and the other nine of which were differentiating loci.
The end result of Mr Webb's analysis was that statistically and genetically, the possibility could not be totally discounted that two people, one of them the complainant, could have contributed to the mixture of material found on the pyjama top. This was because two loci produce masked results. That, however, left seven identifying loci which Mr Webb said were enough to identify the appellant. The learned trial Judge found that the seven identifying loci also enabled Mr Webb to calculate "exclusionary odds of very large dimensions". The exact conclusion reached by Mr Webb was expressed in his report, which he identified in evidence, in the following terms:
"PW 17 Pyjama top
A mixed DNA profile consistent with having come from two people was recovered from the cellular material on the inside left chest area of the pyjama top.
If it is assumed that there are two people contributing to the mixture, then it is at least 10 billion times more likely to find this mixed DNA profile if it comes from [the complainant] and Percy James Riley than if it comes from [the complainant] and an unknown."
The case for the appellant at trial was that the number of matched loci on the database search of Mr Webb was insufficient. Mr Webb conceded in cross‑examination that it would be possible to test at a greater number of loci and that some laboratories did this. The Pathology Centre in Perth is presently considering whether or not it would do so.
Mr Webb testified that the figure of "1 in 10 billion" that he had taken was a conservative figure, based on statistical and genetic calculations. Relevant excerpts from his testimony are as follows:
"And the calculation of 1 in 10 billion was which?---Is the most conservative. …
In relation to the mixed DNA profile recovered from the inside left of the pyjama top, PW17 - in relation to that mixture, the likelihood ratio for the Caucasian population was 5.6 by 10 to the 11, so that is 566 billion. The likelihood ratio for the Aboriginal population was 434 billion, and the likelihood ratio for the South‑East Asians was 1571 billion; in other words, 1.5 trillion.
So how did you come by the 10 billion number?‑‑‑That is a figure that we use in our laboratory as a conservative figure; the 10 billion in relation to the world's population as a relationship. We currently don't report the actual figures because they are quite extreme at the best. We have a cut-off value of 1 in 10, or a probability cut-of of 10 billion, and in the same way in our relationship with the mixture calculations. So you can see we've referred to three data sets and I've given in fact a very conservative figure in relation to that because the most conservative figure in this case was in fact from the Aboriginal population, which was 434 billion."
Mr Webb conceded that there are other databases in the world where there was a lower cut‑off rate. As he pointed out, that was the choice of the particular database. However, most Australian States report actual values. The Pathology Centre in Western Australia reports to 10 billion.
When asked about the Aboriginal subpopulation, Mr Webb said:
"It would certainly take in any account for any subpopulation in that data, yes, or any ‑ ‑ ‑
Isn't it necessary to do a further calculation in regard to excluding relatives?‑‑‑We don't perform that, no, not routinely, but that figure would certainly account for any relatives in that database.
What about in regard to the Aboriginal subpopulation ‑ ‑ ?‑‑‑Sorry, are you referring to an FST value or are you referring to actual relatives?
Actual relatives at the moment? Now you've mentioned it, what exactly is an FST value? Could you please explain that?‑‑‑It's correction coefficient for common ancestry within the populations; annual data sets.
And that FST calculation is because there's some alleles in some subpopulations that are in disequilibrium? Is that why it's used?‑‑‑Not necessarily. It's really to account for, as I said, that common ancestry. We use the subpopulation model, which is based on the work from Buckleton and Walsh and Weir, and we look at those, accounting for those subpopulations; currently using an FST value of .03 in our data in relation to single set profiles.
Did you use that calculation in this calculation?‑‑‑Not in the mixture calculation, no.
Why is that?‑‑‑Because our current mixture calculation program does not allow for FST values to be inserted."
Mr Webb agreed that a range of FST values was used in places such as New Zealand and the United States of America in respect of subpopulation groups. He was also aware that by reason of long‑term isolation there was some disequilibrium evident in the DNA of Aboriginal people at two loci. The frequency at these two loci would not be expected in a randomly mating population.
Mr Webb spoke, in his evidence, of the "random man not excluded" methodology. He said:
"There's still no real consensus in the forensic laboratory area as to how we will use the random man not excluded. Some laboratories, like the Australian Federal Police laboratory, do use it. I know that the Tasmanian laboratory do use it. But there's certainly not - we're not in unanimous agreement as to how we report it and how we use it. But certainly in our laboratory, and it's practice - that we actually use it in our files. We certainly look at the figure that we obtain and relate that to the actual mixture at hand.
…
Yes?‑‑‑The chance of the random man – the RMNE score or the random man not excluded – the probabilities are Caucasian, one in 453,700; Aboriginal, one in 668,900; and in South-East Asian, one in 3,097,000.
That's not taking into account the FST correction?‑‑‑That's not taken into that account, no."
The upshot of Mr Webb's testimony was that by the use of DNA profiling techniques in practice at the Pathology Centre in Western Australia, it was at least 10 billion times more likely to find the mixed DNA profile if it came from the complainant and the appellant than if it came from the complainant and an unknown person. Whilst there may be different likelihood ratios for different ethnic groups within the population, the calculation "1 in 10 billion" was still very conservative. For the Aboriginal population, the figure would actually be one in 434 billion. The calculations did not take into account the exclusion of relatives, nor did the calculation allow for FST values to be inserted. Using the random man not excluded methodology, the probabilities became one in 686,900 for Aboriginals.
The appellant's witness, Dr B L McDonald, was critical of the evidence of Mr Webb. He identified the following:
(1)No attempt had been made by Mr Webb to indicate the likelihood of relatives of the appellant having the same DNA combination.
(2)The FST correction for Aboriginals was not incorporated in the analysis - although there have been no studies on Aboriginals.
(3)Working on the basis of the appellant being identified on only six loci, a probability of one in 48,0000 on a random man not excluded basis was the appropriate calculation.
As the learned trial Judge pointed out, on cross‑examination of Dr McDonald it became clear that his primary concern was the use of a one in 10 billion number without factoring in the issues of population subgroups and because the analysis was based on an assumption that only two people had contributed to the DNA mixture.
The learned trial Judge was left in a position in which the evidence led by the prosecution conservatively established that there was only a one in 10 billion chance that somebody other than the appellant had left his DNA in the mixture which was taken from the complainant's pyjama top. The calculation was made in accordance with the methodology used at the Pathology Centre in Western Australia.
Evidence led on behalf of the appellant at trial suggested that there were other bases upon which DNA analysis might be performed. Reference was made to a number of variations in the result depending upon what factors were used in the DNA analysis.
He learned trial Judge was, of course, required to be satisfied beyond reasonable doubt that the DNA analysis conducted by Mr Webb and which (on his testimony) conservatively estimated that there was only a one in 10 billion chance that it was somebody other than the appellant who had contributed to the mixture found on the pyjama top constituted proof beyond reasonable doubt that it was the appellant who was the offender.
Having contrasted the evidence given by Mr Webb with that given by Dr McDonald, the learned trial Judge clearly accepted that of Mr Webb. His Honour said, "There is no doubt quite proper debate about the calculation of likelihood ratios and the factors to be studied or included", but appears nevertheless to have been satisfied beyond reasonable doubt that the evidence of Mr Webb was compelling evidence. In my opinion, it was. It was tested and questioned by reference to practices in other jurisdictions, but there was no evidence to suggest that the practices adopted at the Pathology Centre of Western Australia are inadequate or based on any false assumption. At its highest, an expert witness was called by the appellant who put forward what he contended were preferable bases upon which DNA analysis should be made. A number of the tests or methodologies to which he referred are in use in other jurisdictions. That is not, however, to say that they are necessarily preferable or more accurate. It is all a question of what methodology is used in what jurisdiction.

The learned trial Judge was acutely aware of the fact that Mr Webb's testimony had been the subject of considerable scrutiny. He was also well‑aware of Dr McDonald's opinions as they were expressed in evidence. Nevertheless, he was in a position where he was called upon to reach a view on the expert testimony before him as to whether the evidence led by the prosecution satisfied him beyond reasonable doubt that the DNA analysis of Mr Webb was such that it could confidently be said beyond reasonable doubt that the odds against the offender having been anybody but the appellant were so great that there was only one result open, and that was a finding that the appellant was, indeed, the offender.
The learned trial Judge made some statements in the concluding paragraphs of his written reasons which have been the subject of criticism. He said:
"It seems clear to me that all this evidence, taken from a statistical or genetic viewpoint, must in a court situation be placed in the context of the known, and uncontested, facts. This is clearly a situation in which two people were involved - one, the daughter, is undoubtedly identified by the DNA. The other identified by the later as well as earlier DNA samples is a local man of sufficiently matching physical and social characteristics. It is relevant to point out that DNA analysis of the material did not match that taken by police from other persons each described as relatives of the accused, Bert Junior Riley, apparently a cousin, and Charles Douglas Riley and Paul Anthony Riley, apparently brothers of the accused. No-one else is said either to have been involved in the sexual conduct or identified nor was the court provided with evidence of family, racial descent or other evidence able to assist in calculating likelihood ratios save that the only three family members tested were excluded by the DNA tests and a number not identified on photoboard. No alibi evidence was led."
His Honour then concluded:
"The court, however, has to adopt a test of proof beyond reasonable doubt. As has been said many times those words mean what they say without further explanation. They do not rely on such statistical probability. Here I have been satisfied to that legal standard. The onus is on the State whose evidence I accept. None of the issues raised by the defence, singly or together, create such a doubt."
I am unable to see any shortcomings in the conclusions reached by the learned trial Judge. In the first instance, it was true that the DNA evidence was to be placed in the context of the known and uncontested facts. As the learned trial Judge put it, the complainant was undoubtedly identified as a person who had contributed to the DNA sample. The other contributor to the sample was not identified in any absolute sense. However, if Mr Webb's evidence was accepted, it was 10 billion times more likely that the DNA profile found in the material on the pyjama top which was not that of the complainant came from the appellant rather than from an unknown person. They were enormous odds.
The learned trial Judge correctly stated that the appellant's physical and social characteristics were "sufficiently matching" with those described by the complainant. True, there were some variations, but the learned trial Judge did not suggest that there was an identical match between the appellant and the offender described by the complainant. Nor did he rely on these "matching physical and social characteristics" otherwise than by way of pointing out that it was not the case that the appellant was an entirely different person in physical and social characteristics to the offender identified by the complainant.
The learned trial Judge also made reference to the fact that the DNA analysis did not match DNA taken by police from a cousin and two brothers of the appellant. This was undoubtedly correct. The evidence established it. Further, a number of other Rileys whose photographs were placed on photoboards were not identified by the complainant as the offender.
The learned trial Judge was correct to say that the standard of proof applicable to criminal proceedings does not rely on statistical probability. The question was whether the evidence put forward by the prosecution was sufficient to satisfy the learned trial Judge beyond reasonable doubt of guilt. He found that it was. In my opinion, the evidence was capable of satisfying the standard of proof required and the learned trial Judge was therefore entitled to accept it and accept it to the standard required. As he put it, nothing raised on behalf of the appellant created a reasonable doubt. This included the testimony of the appellant who gave evidence at trial denying that he had committed the offences and saying that he was in Bunbury at the time. It also included the evidence of Dr McDonald, who expressed professional and expert criticisms of the evidence of Mr Webb, but which failed to satisfy the learned trial Judge that they were sufficient to cast a reasonable doubt upon the testimony which the prosecution had led.
The reasons of the learned trial Judge were, in my view, quite adequate. They met the test of sufficiently disclosing the learned trial Judge's reasoning process. There may not have been a detailed analysis by the learned trial Judge as to why he preferred the evidence of Mr Webb on the one hand from that of Dr McDonald on the other. However, it seems to me that it is sufficiently clear from the reasons that, having detailed the evidence of Mr Webb and contrasted that of Dr McDonald, his Honour was concluding that whatever may be the case in other jurisdictions, the methodology of DNA analysis at the Pathology Centre of Western Australia is sufficiently reliable to enable the results produced by Mr Webb in this case to be acted upon and accepted as proof beyond reasonable doubt that the appellant was the offender.
In my view, his Honour's reasons fell within the description of what constitute appropriate reasons for judgment set out by Owen J in Garrett v Nicholson (1999) 21 WAR 226 at [73] ‑ [74] as follows:
"At common law there was (and is) a duty on a decision maker required to act judicially or quasi-judicially to give reasons for decision. The reasons must disclose adequately the intellectual process which has resulted in a particular determination. I will repeat what I said on the general question of a duty to give reasons (with which White J agreed) in Frichot v Zalmstra, unreported; FCt SCt of WA; Library No 981291; 13 May 1998 at 14:
'The authority most often cited in this regard is the dicta of Moffitt P, in Pettit v Dunkley [1971] 1 NSWLR 376 at 387‑388. That passage is so well‑known that I will not repeat it other than to say that it has been adopted in this Court in cases such as Lloyd v Faraone [1989] WAR 154 at 163.
In Wright v Australian Broadcasting Commission [1977] 1 NSWLR 697 at 702 the court indicated that in a position such as we have here the appealable error lies not in the result of the trial but in the failure to give relevant reasons in circumstances which deprive a party of an effective right of appeal conferred by statute.
To that I would add that in our judicial system, where proceedings are to be conducted in public, the duty to give reasons is an adjunct to or in some respects a part of the overall obligation to afford to litigants procedural fairness. Put in that way, in addition to securing the statutory right of appeal, the obligation to give adequate reasons is part of the fairness to a litigant who comes to the court to know why it is that he or she succeeded or has been unsuccessful. ... [T]hat does not mean that every piece of evidence, every exhibit, every word that has fallen from counsel during submissions must be alluded to expressly or even by implication in the course of giving reasons. It would neither be necessary nor feasible for an obligation of that nature to be imposed on trial courts.
It is sufficient if the reasoning process which led to a result is disclosed with sufficient certainty to enable the litigant to know why it is that the result ensued and to ensure that the statutory right of appeal has been secured.'
The tribunal of fact in that case was the District Court. However, I believe the principles are equally applicable to the Court of Petty Sessions. To the authorities mentioned in that extract I would add Soulemezis v Dudley (Holdings) Pty Ltd (1987) 10 NSWLR 247 at 259, 277‑281; Danagher v Racing Penalties Appeal Tribunal (1995) 13 WAR 531 at 539‑40, 548 and Kiama Constructions Pty Ltd v Davey (1996) 40 NSWLR 639 at 640, 647. I would also add that it is necessary to look at the reasons as a whole to see if they give the sense of what was intended in a way that achieves the results to which I have referred. It is sometimes the case that by dissecting the reasons line by line, word by word the reader loses the sense of what is conveyed by a particular passage. It is often the case that the true sense can be gleaned only by looking at the passage in the context of the whole. In this way something which appears at first glance to be a material omission may be adequately explained."
Further, the evidence of Mr Webb met the requirements for admissibility of expert opinion evidence as they were set out by Anderson J in Pownall v Conlan Management Pty Ltd (1995) 12 WAR 370 at 387 and 390:
"The general rule is that an expert opinion is of no value unless the facts upon which it is based are proved by admissible evidence: see Ramsay v Watson (1961) 108 CLR 642; Paric v John Holland (Constructions) Pty Ltd (1985) 59 ALJR 844; Steffen v Ruban (1966) 84 WN (Pt 1) (NSW) 264. …
Expert opinion is to be judged like any other evidence. It must be comprehensible and reach conclusions that are rationally based. The process of inference that leads to the conclusions must be stated or revealed in a way that enables the conclusions to be tested and a judgment made about the reliability of them."
In Beale v Government Insurance Office of NSW (1997) 48 NSWLR 430, Meagher JA set out (at 442 ‑ 444) what is considered to be the content of an adequate statement of reasons for judgment. His Honour made reference to the need for a trial Judge to refer to all the evidence led in the proceedings to indicate which of it is accepted or rejected. He pointed out however, that reasons need not necessarily be lengthy or elaborate. Accordingly, the content of the obligation to deliver adequate reasons is not the same for every judicial decision. His Honour then said at 443:
"No mechanical formula can be given in determining what reasons are required. However, there are three fundamental elements of a statement of reasons, which it is useful to consider. First, a judge should refer to relevant evidence. There is no need to refer to the relevant evidence in detail, especially in circumstances where it is clear that the evidence has been considered. However, where certain evidence is important or critical to the proper determination of the matter and it is not referred to by the trial judge, an appellate court may infer that the trial judge overlooked the evidence or failed to give consideration to it: North Sydney Council v Ligon 302 Pty Ltd (1995) 87 LGERA 435. Where conflicting evidence of a significant nature is given, the existence of both sets of evidence should be referred to.
Secondly, a judge should set out any material findings of fact and any conclusions or ultimate findings of fact reached. The obvious extension of the principle in North Sydney Council is that, where findings of fact are not referred to, an appellate court may infer that the trial judge considered that finding to be immaterial. Where one set of evidence is accepted over a conflicting set of significant evidence, the trial judge should set out his findings as to how he comes to accept the one over the other. But that is not to say that a judge must make explicit findings on each disputed piece of evidence, especially if the inference as to what is found is appropriately clear: Selvanayagam v University of the West Indies [1983] 1 WLR 585; [1983] 1 All ER 824. Further, it may not be necessary to make findings on every argument or destroy every submission, particularly where the arguments advanced are numerous and of varying significance: Rajski v Bainton (Court of Appeal, 6 September 1991, unreported).
Thirdly, a judge should provide reasons for making the relevant findings of fact (and conclusions) and reasons in applying the law to the facts found. Those reasons or the process of reasoning should be understandable and preferrably [sic] logical as well."
Finally, his Honour entered this caveat at 444:
"Whilst it is desirable to address these elements in giving reasons for decision, it is the purpose which the reasons serve which assumes primary importance in determining the content of the reasons. That purpose must be weighed against other considerations. It has been noted by this Court that the content required of a statement of reasons is to be measured against the burden that the provision of reasons imposes on the judicial system: Sinak v Tess (Court of Appeal, 15 March 1995, unreported). The reason for this is that the giving of overly elaborate reasons can serve to undermine public confidence in the judiciary and in the judicial system in the same way that insufficient reasons can. On the one hand, the provision of inadequate reasons can lead to a sense of injustice and a reduced appreciation or understanding of legal rights and obligations. On the other hand, an overly onerous duty to provide reasons increases costs and delay in the judicial system which has the effect of undermining public confidence in the judicial system."
In the present case, the learned trial Judge met the requirements of an adequate statement of reasons for judgment. His Honour referred in detail to the DNA evidence which was the critical evidence in the trial. His Honour made it clear that he accepted the evidence of Mr Webb, that there was only a one in 10 billion chance that somebody other than the complainant and the appellant had left DNA material on the pyjama top. This led his Honour to be satisfied beyond reasonable doubt that the prosecution had proven its case. The process of reasoning was thus understandable and logical.
It is true that greater detail might have been provided in relation to the analysis of the respective positions of Mr Webb and Dr McDonald, but as Meagher JA pointed out in Beale v Government Insurance Office of NSW (supra), overly elaborate reasons are not required and may be counter‑productive. It must be remembered that this was a criminal trial by Judge alone. It was an exception to the general rule that a trial on indictment is before Judge and jury. The trial Judge was required to give reasons because he judged the matter alone. In the end, the question was whether the evidence adduced by the prosecution was sufficient to satisfy the trial Judge beyond reasonable doubt that the appellant was the offender. This decision hinged upon the acceptance of the evidence of Mr Webb. His Honour accepted that evidence and explained why he did. He considered the evidence which had been led for the appellant, including that of the expert called by the appellant, but was nevertheless satisfied beyond reasonable doubt that the evidence of Mr Webb was compelling, and that evidence, and that alone, was sufficient to constitute proof of guilt to the required standard. I am unable to conclude that the learned trial Judge was in error in any way in reaching that view. I would therefore dismiss the appeal.