R v Pfennig
[2017] SASCFC 26
•11 April 2017
SUPREME COURT OF SOUTH AUSTRALIA
(Court of Criminal Appeal: Application)
R v PFENNIG
[2017] SASCFC 26
Judgment of The Honourable Justice Blue
11 April 2017
CRIMINAL LAW - EVIDENCE - IDENTIFICATION EVIDENCE - MODES OF IDENTIFICATION - DNA EVIDENCE
CRIMINAL LAW - APPEAL AND NEW TRIAL - VERDICT UNREASONABLE OR INSUPPORTABLE HAVING REGARD TO EVIDENCE
Application for permission to appeal against conviction.
The applicant was convicted after a trial by Judge alone of the murder of Louise Bell in 1983. He seeks permission to appeal against the conviction on the ground that the verdict is unreasonable or incapable of being supported having regard to the evidence. There are two separate limbs to the proposed ground of appeal:
1. The evidence was not capable of proving beyond reasonable doubt that the applicant’s DNA was present on Louise Bell's pyjama top.
2. There was no evidence adduced by the prosecution capable of excluding as a reasonable possibility the innocent transfer of the applicant’s DNA to Louise Bell's pyjama top.
Held:
1. The first limb of the proposed ground of appeal is not reasonably arguable ([at 116]).
2. The second limb raises a question of general importance, justifying the grant of permission to appeal. ([at 131]).
3. Permission to appeal granted ([at 132]).
Criminal Law Consolidation Act s 353, referred to.
Fitzgerald v R (2014) 88 ALJR 779; R v B, FG (2012) 114 SASR 170; R v Doheny and Adams [1997] 1 Crim App R 369; R v Hallcroft (2016) 126 SASR 415, discussed.
R v Milton [2009] SASC 44; R v Parenzee (2007) 101 SASR 456; R v Tran & Tran [2011] SASCFC 153; R v Wymond [2013] SASCFC 12, considered.
R v PFENNIG
[2017] SASCFC 26Application for permission to appeal
BLUE J: This is an application for permission to appeal against conviction.
The applicant, Dieter Pfennig, was convicted after a trial by Judge alone of the murder of Louise Bell in 1983.[1] He seeks permission to appeal against the conviction on the ground that the verdict is unreasonable or incapable of being supported having regard to the evidence.[2]
[1] R v Pfennig (No 2) [2016] SASC 171.
[2] The ground is erroneously expressed as “The verdict is unsatisfactory and unreasonable and against the weight of the evidence”. Although this error is unfortunately not uncommon in the drafting of grounds of appeal, this wording does not accord with section 353 of the Criminal Law Consolidation Act 1935 (SA) which refers to the ground that the verdict is “unreasonable or cannot be supported having regard to the evidence”. Neither of the other two grounds referred to in that section, namely “a wrong decision on any question of law” and “on any ground there was a miscarriage of justice”, is invoked.
There are two separate limbs to the proposed ground of appeal, namely:
1.the evidence was not capable of proving beyond reasonable doubt that the applicant’s DNA was present on the victim’s pyjama top;
2.there was no evidence adduced by the prosecution capable of excluding as a reasonable possibility the innocent transfer of cells of the applicant or pieces of material containing cells of the applicant to the victim’s pyjama top.
The prosecution case
There was no dispute at trial that Louise Bell was abducted from her bedroom during the night of 4/5 January 1983 and murdered soon thereafter. The sole issue was the identity of the murderer.
The prosecution case was entirely circumstantial.[3] It was summarised by the Judge towards the end of his Honour’s reasons for judgment:
[3] The prosecution adduced evidence at trial of two alleged confessions by the applicant but the Judge considered that evidence unreliable. It can be ignored for the purpose of considering permission to appeal.
I therefore find the following items of circumstantial evidence proved:
…
3.The accused lived a short distance from the deceased’s house at the time of her abduction.
4. The accused used to walk the streets alone at night.
5. The deceased went to the same school as the accused’s daughter, Petra Pfennig.
6.On about Christmas Day 1982, the accused and his family went in convoy with other people to Swan Hill and Yarrawonga in Victoria for a canoeing marathon.
7.The accused returned to Adelaide with other people in their car on the evening of 3 January 1983. The accused’s wife and children went in the family car to Broken Hill.
8.The accused spoke to RP and said he had lost an alibi when told about the police doorknocking. RP treated that comment as a joke.
9. KD was phoned by an unknown person with a European accent.
10. The accused had a German or European accent.
11.The pyjama top of the deceased was deposited on KD’s lawn about two weeks after the above telephone call.
12.The person who deposited the pyjama top on KD’s lawn was the same person who phoned her. That person was in fact the killer of Louise Bell.
13.From the time of the abduction of the deceased, the accused on occasion made comments about the deceased and her abduction.
14.The pyjama top was submerged in the Onkaparinga River and washed in tap water before it was deposited on KD’s lawn.
15.DNA was extracted from the pyjama top and compared with DNA from the accused. Various comparisons differed because of the different methods used and the different information placed into the mix by the laboratories. The results were as follows:
(a) using the random match probability analysis and a Dutch Caucasian database, the chance of finding someone in the Dutch Caucasian population with the profile obtained from the tape lift was 1 in 5,766 billion;
(b) by use of the same random match probability analysis, recalculating (a) but using the Australian Caucasian database, that figure was 1 in 7,537 billion;
(c) using the LRmix Studio probabilistic method, a likelihood ratio for material from the tape lift in support of the proposition that the accused and one other person contributed to the DNA on the tape lift, as opposed to the proposition that the tape lift contained the DNA of two unknown individuals is between 9.6 billion and 36.9 billion;
(d) using STRmix and using a Caucasian database and a Bavarian population database from Germany, it is approximately 17,000 times more likely to have obtained a DNA profile from the fluff (74.2.A) if the accused is a contributor to the DNA profile, rather than some other unknown, unrelated individual;
(e) by using the same method in relation to the material from the tape lift (74.3.3.B), it is approximately 6400 times more likely to have obtained the DNA profile if the accused is a contributor to the DNA profile, rather than some unknown, unrelated individual; and
(f) by use of STRmix and the likelihood ratios method in relation to the tape lift, Colin Bell, Dianne Bell, Rachel Bell and Raymond Geesing were excluded as contributors to the DNA profile.
The DNA evidence formed a major plank in the prosecution’s circumstantial case. The prosecution accepted that, if the DNA evidence were left aside, the remaining circumstantial evidence was incapable by itself of proving beyond reasonable doubt that the applicant was the murderer.
Background
General narrative
As at January 1983 the Bell family had for some years been living in Meadow Way at Hackham West. Louise Bell completed grade 1 at Morphett Vale Primary School at the end of 1978 and moved (together with all other students) to Hackham West Primary School at the beginning of 1979. She completed grade 5 at the end of 1982.[4]
[4] Evidence of Colin Bell T29/26-29.
As at January 1983 the Pfennig family had for some years been living in Holly Rise at Hackham West. This was within a short walking distance from the Bell family house in Meadow Way. The applicant was born in Germany and spoke with a slight German accent.[5] The applicant had difficulty sleeping and was accustomed to go walking or jogging late at night.[6] Petra Pfennig completed reception at Morphett Vale Primary School in 1978 and moved to Hackham West Primary School at the beginning of 1979. Halfway through 1979, she moved from grade 1 into grade 2. She completed grade 5 at the end of 1982.[7]
[5] Evidence of Sandra Pfennig (nee Warren) T476/38 and477/16 and Anthony Van Der Stelt (SAPOL) T1335/28.
[6] Evidence of Sandra Pfennig T487/23-26.
[7] Evidence of Petra Pfennig T513/36-37.
In 1980 Louise Bell and Petra Pfennig were in the same class in grade 3.[8]
[8] Evidence of Petra Pfennig TT513/38 and 514/1.
At least a year before Louise Bell disappeared, Angelina Richards saw the applicant walking from the Hackham West Primary School towards home together with Petra Pfennig and Louise Bell on at least six occasions over a three to six-month period.[9]
[9] Evidence of Angelina Richards T313/23-32, 314/5-7 and 315/19-26.
In 1982 Louise Bell and Petra Pfennig were both in grade 5 but were in different classes.[10] They were in the same school basketball team.[11] They had weekly practices and weekly matches between February and November 1982.[12] They both attended a barbecue pool party at the coach’s house in November 1982 to celebrate the end of the basketball season.[13] They both went on a school trip to Mount Gambier towards the end of 1982. Neither had a sleepover at the other’s house.[14]
[10] Evidence of Priscilla Grace T158/37.
[11] Evidence of Petra Pfennig T515/25-26 and Patricia Murrell T186/37-38 and190/22-24.
[12] Evidence of Petra Pfennig T516/12-18 and Ms Murrell T190/18 and191/7–14.
[13] Evidence of Petra Pfennig T516/30-38 and Ms Murrell T193/26-33.
[14] Evidence of Mr Bell T31/7-10 and Petra Pfennig T520/28-32.
Louise Bell was given a pair of pyjamas by her parents as a Christmas present at Christmas in 1982.[15] She wore the top but not the bottom as it was too tight.[16]
[15] Evidence of Mr Bell T52/32-35. A statement by Mrs Bell was tendered suggesting that the pyjamas had been passed on to Louise Bell by her younger sister but the Judge preferred the evidence of Mr Bell and this finding is not challenged on appeal.
[16] Evidence of Mr Bell T52/26-31.
On Christmas morning in 1982, the Pfennig family left Adelaide in company with other families to participate in a canoe marathon on the Murray River in Victoria.[17] On 3 January 1983, the applicant arrived back in Adelaide with the Coppin family, having returned with them in their Kombi van.[18] Mrs Pfennig and their two daughters travelled from Victoria to Broken Hill and did not return to South Australia for at least another two weeks.[19]
[17] Evidence of Sandra Pfennig T491/14-16 and 494/8-12 and Petra Pfennig T521/17-25.
[18] Evidence of Richard Coppin T256/6-11.The Judge made a finding that the accused returned to Adelaide with other people in their car on 3 January 1983 (R v Pfennig (No 2) [2016] SASC 171 at [362]) and this finding is not challenged on appeal.
[19] Evidence of Sandra Pfennig T499/30-38, 500/1-8 and Petra Pfennig T526/3-18.
During the night of 4/5 January 1983, Louise Bell disappeared from her bedroom at the Bell family home. In the morning, Mr Bell found that the flyscreen on the window had been damaged and tent wires and frames that normally obstructed the path had been moved onto the lawn.[20] Louise Bell’s body was never found. Although the flyscreen had already been damaged before that night, the damage found in the morning was worse.[21]
[20] Evidence of Mr Bell T39/19-27 and Brian Riddle T90/31-38 and 91/1-3.
[21] Evidence of Mr Bell T41/10-15, 43/5-9 and 51/1-23.
On 6 January 1983 the police declared the disappearance to be a major crime.[22] They commenced doorknock inquiries in the neighbourhood including in Holly Rise.[23] They have no record of any response to the doorknock from anyone at the Pfennig house.[24] They have no record of any statement ever being given or information received from the applicant about the disappearance.[25]
[22] Evidence of John Woite (SAPOL) T269/27-32.
[23] Evidence of Mr Woite T281/28-38 and 286/10-35.
[24] Evidence of Mr Woite T291/11-20.
[25] Evidence of Mr Woite T292/3-15.
RP lived next door to the Pfennig family in Holly Rise.[26] He watered the lawn at the applicant’s request over the Christmas/New Year period in 1982/83 while the Pfennig family was away interstate.[27] When the police came to knock on the door of the Pfennig house, he told them that the family was away on holiday.[28] One or two days later, he was watering the lawn when he saw the curtain move in the house. After knocking a few times on the door, the applicant answered.[29] They discussed the Bell disappearance. RP told the applicant that, when the police had come knocking, he told them that the applicant was in Broken Hill with his family. The applicant told RP that he had been sprung and had lost an alibi, which RP treated as a joke.[30]
[26] Evidence of RP T359/10-11, 37-38 and 340/1-2.
[27] Evidence of RP T363/30-36.
[28] Evidence of RP T365/12-25.
[29] Evidence of RP T365/33-38 and 366/1-21.
[30] Evidence of RP T367/27-31.
KD lived around the corner from the Bell house.[31] On 17 January 1983 KD received a telephone call from a man who had a slight European accent (not English or French). He addressed her by her surname. He told her that Louise Bell was with them and did not want to go home but he needed medical information for her. She asked him how she knew that it was genuine call. He told her that if the police went to the corner of South and Beach Roads and looked under a brick they would find Louise’s earrings. He said that the police were stupid and did not notice three things, including that wire frames below the window had been moved to avoid making a noise. He referred to the flyscreen being eased out and, when she asked him whether it was already broken, he said that it was. [32]
[31] Evidence of KD T419/30-32.
[32] Evidence of KD T426/9-20
KD reported the telephone call to the police, who found a pair of earrings under a brick at the corner of South and Beach Roads.[33] They were later identified by Mr Bell as earrings that had been worn by Louise.[34] The building on the corner of South and Beach Roads was the former site of the Morphett Vale Primary School; it then became a library used by the Pfennig family; it then became a community centre at which the applicant played chess.[35]
[33] P11 (Agreed Facts 1).
[34] Evidence of Mr Bell T55/29-35.
[35] Evidence of Sandra Pfennig T481/35-36 and 482/3-4.
On 28 February 1983 KD noticed an item of clothing neatly folded on her front lawn. After a while, thinking that it was rubbish, she picked it up and took it to the back of her house and placed it in a disused flower pot with the intention of disposing of it. After a while, she examined it and realised that it was a pyjama top. She telephoned the police who took possession of it.[36] It was later identified by Mr Bell as the pyjama top that Louise had been wearing on the night of 4/5 January 1983.[37]
[36] Evidence of KD T435/22-33.
[37] Evidence of Mr Bell T53/20-26.
On 28 February 1983 Senior Constable Roemer, a crime scene examiner, examined the pyjama top. He first removed seagrass and algae. He then took five tapelifts from each of the inner front, outer front, inner rear and outer rear surfaces of the top. He finally vacuumed the top and placed the vacuumings in a petri dish. He observed foraminifera and soil containing minerals in the vacuumings, together with a piece of yellow fluff.[38]
[38] Evidence of Theodore Van Dijk (SAPOL) T654/29-35 and 655/2-17. Exhibit P79.
On further analysis by six experts in biology and geology, the vacuumings were found to contain plant life (diatoms), micro-organisms (foraminifera), minerals (glauconite and pyrite) and soil. The seagrass, algae and diatoms found were consistent with the pyjama top having been immersed in the Onkaparinga River or another South Australia estuarine river.[39] The soil contained very high levels of pyrite suggestive of its having been obtained from the Onkaparinga River.[40] The foraminifera and glauconite were strongly suggestive of the pyjama top having been in the Onkaparinga River.[41] The lack of halite and salt were strongly suggestive of the top having been rinsed in tap water after being immersed in an estuarine river.[42] The applicant at trial did not challenge a finding being made that the top had been in the Onkaparinga River and subsequently rinsed and the Judge’s finding to this effect is not challenged on appeal. The applicant had a strong interest in canoeing and in the Onkaparinga River.[43]
[39] Evidence of Professor Waycott T691/19-23 and Professor Fitzpatrick T940/19-23.
[40] Evidence of Professor Fitzpatrick T940/28-38 and 941/1.
[41] Evidence of Professor McCann T838/10-38 and 839/1-38.
[42] Evidence of Professor Fitzpatrick T943/26-38 and 944/1.
[43] Evidence of Reginald Coppin T213/18-24 and Sandra Pfennig T489/30-38 and 390/1-3.
At about the time of Louise Bell’s disappearance, the applicant told a school counsellor at the High School where they both worked that he was not allowed to discuss the Bell case because he was a person of interest, which was not the case.[44] Later, in 1985 or 1986, the applicant told a fellow teacher that because his daughter Petra and Louise Bell were close friends he was able to help the police, which was also not the case.[45]
[44] Evidence of Konstantine Heyer T354/16-32 and 355/1-4 and Mr Woite T291/30-37.
[45] Evidence of Russell McMillon T410/18-26.
In about 1988 the applicant was paddling his canoe on the Onkaparinga River in the vicinity of Shelley’s Beach in company with four other canoers. He told one of them that on a previous occasion his paddle had got caught up in some bones which he thought were probably human and referred to the police having found traces of substances on Louise Bell’s pyjamas.[46] He told two of his fellow canoers that Louise Bell’s body was in the area where they were canoeing.[47]
[46] Evidence of Neil Campbell T402/22-38 and 403/4-9.
[47] Evidence of Mr Campbell T403/12-15 and Michael Furga T1224/10-12 and 1225/26-33.
DNA analysis
In June 2011 the petri dish containing the vacuumings (including the piece of fluff) and several of the tapelifts taken by Senior Constable Roemer in February 1983 were sent by the police to Forensic Science South Australia (FSSA).[48] Tapelift 74.3.3.B from the outer rear surface (the Tapelift) and the piece of fluff (item 74.2.A) (the Fluff) ultimately produced DNA profiles subsequently relied on by the prosecution as being the applicant’s DNA.[49]
[48] Evidence of Mr Van der Stelt T1330/1-6 and Ms Mitchell T1402/8-13.
[49] Evidence of Mr Van Dijk (SAPOL) T687/19-38 and Ms Mitchell T1482/25-38 and 1450/1-17.
Between August and December 2011 FSSA extracted, quantified, amplified (using Profiler Plus), detected (by capillary electrophoresis) and analysed (using GeneMapper software) DNA from the Tapelift and Fluff (field samples).[50] This process involved examining alleles at ten loci comprising nine human variable loci plus amelogenin for gender; conducting 28 cycles of amplification; and using an RFU threshold requiring at least 50 RFU.[51]
[50] Evidence of Ms Mitchell T1449/9-38, T1450/1-17 and 1454/5-38.
[51] Evidence of Ms Mitchell T1417/19-38, 1471/8-19 and 1420/30-35 and Professor Linacre T2675/2-12.
Natasha Mitchell of FSSA compared the alleles detected by this process with the alleles of the DNA in a reference sample taken from a swab of the applicant in 2003. She concluded that, where an allele was detected in the Tapelift or the Fluff DNA, it was consistent with the applicant’s DNA as shown in the following table:[52]
[52] Evidence of Ms Mitchell T1452-1453. See exhibits P156, 157 and 158.
Locus Applicant DNA Tapelift DNA Fluff DNA D3S1358 16, 18 16, 18 18 D8S1179 8, 10 8, 10 8, 10 D13S317 8 - 8 amelogenin X, Y X, Y X, Y
In 2011 FSSA used a set of interpretation guidelines requiring certain criteria to be met before proceeding to calculate a statistical weighting in respect of a comparison between a field sample and a reference sample.[53] Those criteria were not met in respect of the Tapelift or Fluff DNA and no attempt was made to calculate a likelihood ratio.[54]
[53] Evidence of Ms Mitchell T1468/36-38 and 1469/1-12.
[54] Evidence of Ms Mitchell T1453/19-25 and 1454/1-4.
The Netherlands Forensic Institute (NFI) used Next Generation Multiplex for amplification involving 34 cycles and detection of 16 loci comprising 15 human variable loci plus amelogenin for gender.[55] The use of 15 rather than nine human variable loci potentially allowed (if sufficient DNA was present at each locus) a much higher level of confidence that a match in loci was the result of a common source of the DNA.[56] The use of 34 cycles rather than 28 cycles of amplification resulted in larger quantities of DNA being detected which potentially resulted in more alleles being detected (whether at nine or 15 sites examined) when the original quantity of the DNA was low.[57]
[55] Evidence of Bart Blankers T1690/28-30,1682/37-38 and T1683/1 and Professor Linacre T2675/16-22. Exhibits P174-176.
[56] Evidence of Ms Mitchell at T2631/20-25, Mr Blankers T2893/21-38 and 2894/1-22 and Professor Linacre T2675/34-38 and 2676/1-18.
[57] Evidence of Professor Linacre T2675/34-38, 2676/1-28 and 2698/1-11.
In 2012 the NFI used random match probability to calculate a statistical weighting and generally used a replicate consensus method when the original quantity of the DNA was low, taking at least three replicates and using the results when common across the replicates or the result of the majority where they differed. This was designed primarily to eliminate artefacts being false positives (“drop in”).[58]
[58] Evidence of Mr Blankers T2488/13-27, Dr Kokshoorn T1782/16-21 and Professor Linacre T2703.
In March 2012 FSSA sent to the NFI in the Netherlands the DNA extracts from the Tapelift and Fluff DNA and reference samples from the applicant and others.[59]
[59] Evidence of Ms Mitchell T1461/5-31 and Dr Kokshoorn Exhibit P171.
In April 2012 Bart Blankers at the NFI divided the extracts from each of the Tapelift and Fluff DNA into three replicates. He performed Next Generation Multiplex amplification at 34 cycles; electrophoresis detection and analysis using GeneMapper software. These were independently interpreted by two technicians. Mr Blankers generated a consensus profile. The whole process was independently reviewed by Dr Bas Kokshoorn who concurred with Mr Blankers’ conclusions.
Mr Blankers concluded that, where an allele was detected in the Tapelift or the Fluff DNA, it was consistent with the allele or alleles from the applicant’s DNA as shown in the following table:[60]
[60] Evidence of Mr Blankers T1723-1730, 1742/1-9, 1743/23-38 and 1744/1-13 and Dr Kokshoorn T1798/27-33 Exhibits P174, 175 and 176.
Locus Applicant DNA Tapelift DNA Fluff DNA D10S1248 13, 16 13, 16 13, 15, 16 vWA 17, 18 17, 18 16, 18 D16S539 10,11 11F[61] - D2S1338 17, 25 25F - D8S1179 8, 10 8, 10 8, 10, 13, 15 D21S11 28, 29 28, 29 29F D18S51 12, 18 12F 12F D22S1045 16 16F 16F D19S433 13 13F 13F TH01 8, 9.3 8F 8, 9.3 FGA 23, 25 25F 21, 23, 25 D2S441 10, 11 10, 11 10, 11 D3S1358 16,18 16, 18 16, 18 D1S1656 12, 15.3 12, 15.3 12, 14, 15.3 D12S391 17, 18 18F 18, 19 amelogenin X, Y X, Y X, Y [61] F indicates possibility of either heterozygote or homozygote donor. It denotes possibility of an allele from a heterozygote donor having dropped-out (T1728, exhibit P175).
In respect of the Tapelift, there was a clear major contributor. There was an additional allele present at a very low level which had been contributed by a second minor individual (or may have been an artefact).[62] Mr Blankers concluded that the applicant could not be excluded as the source of the DNA in the Tapelift and proceeded to calculate a statistical weighting.[63] He calculated the random match probability using the Dutch Caucasian male database and the Australian Caucasian male database. He concluded that the chance of finding someone in the Dutch Caucasian population whose DNA matched the profile obtained from the Tapelift was 1 in 5.766 trillion and of finding someone in the Australian Caucasian population whose DNA matched the profile obtained from the Tapelift was 1 in 7.537 trillion.[64]
[62] Evidence of Mr Blankers T1726/6-15 and 1730/12-19 and Dr Kokshoorn T1797/35-38 and 1798/1-2.
[63] Evidence of Mr Blankers T1745/8-15 and Dr Kokshoorn T2471/2.
[64] Evidence of Mr Blankers T1745/14-15 and 1752/3-5, Dr Kokshoorn T1804/15-17 and 1809/17.
In respect of the Fluff DNA, there was a mixed profile representing at least two and quite possibly three contributors without a clear major contributor. The NFI used a set of interpretation guidelines requiring certain criteria to be met before proceeding to calculate a statistical weighting in respect of a comparison between a field sample and a reference sample.[65] Those criteria were not met in respect of the Fluff DNA because of the complex profile with no clear major contributor and no attempt was made to calculate a random match probability.[66]
[65] Evidence of Mr Blankers T2896/3-10 and 2897/10-29.
[66] Evidence of Mr Blankers T2897/19-22 and Dr Kokshoorn T1796/23-25, 2896/3-2897/38 and
2516/19-30.
Subsequently, the NFI created software entitled “LRmix Studio” to calculate a different form of statistical weighting known as a likelihood ratio. This software was created under the supervision of Professor Peter Gill. Professor Gill, Dr John Buckleton and others were pioneers in the use of probabilistic software to create a likelihood ratio.[67]
[67] Evidence of Mr Blankers T2887/35-38 and 2888/1-31, Professor Gill T1948/3-34 and 2942/21-24 and Dr Buckleton T2825/21-31.
In November 2015 Dr Kokshoorn used LRmix Studio to calculate a likelihood ratio using the GeneMapper results for the Tapelift DNA and the applicant’s reference sample. He used the results from all three replicates as opposed to the earlier use of only the majority of the replicates where there was a relevant difference. He calculated that the probability that the Tapelift sample contained DNA of the applicant and one unknown person compared to the probability that it contained DNA from two unknown persons was between 9.6 and 36.9 billion to 1. He expressed the opinion that this was not inconsistent with, and supported, the conclusion earlier reached using random match probability.[68]
[68] Evidence of Dr Kokshoorn T1827/4-9.
In the meantime, Dr Duncan Taylor and Dr Buckleton in Australia and New Zealand had created software entitled “STRmix” to calculate a likelihood ratio. This software predated, but performed a similar function to, LRMix Studio. LRMix Studio is semi-continuous probabilistic software whereas STRmix is continuous probabilistic software.[69] When FSSA introduced and validated STRmix, it reviewed the RFU threshold which until then had been set at 50 RFU. It changed the threshold to 30 RFU.[70]
[69] Evidence of Dr Taylor T1543/30-32 and Dr Buckleton T1841/35-38 and 1842/1-9.
[70] Evidence of Ms Mitchell T2249/14-24.
In October 2012 Ms Mitchell used a lower threshold of 30 RFU to detect DNA in the Tapelift and Fluff DNA. She concluded that, where an allele was detected in the Tapelift or the Fluff DNA, it was consistent with the applicant’s DNA as shown in the following table:[71]
[71] Evidence of Ms Mitchell T1486/31-38 and 1487/1-5. Exhibits P158, 160 and 161.
Locus Applicant DNA Tapelift DNA Fluff DNA D3S1358 16, 18 15, 16, 18 16, 18 vWA 17, 18 17 - FGA 23, 25 - 23 D8S1179 8, 10 8, 10 8, 10 D21S11 28, 29 - - D18S51 12, 18 - - D5S818 10, 11 10 - D1S317 8 8 8 D7S820 8, 9 - - amelogenin X, Y X, Y X, Y
Ms Mitchell used STRmix to calculate a likelihood ratio in respect of the Tapelift DNA, the Fluff DNA and the applicant’s reference sample. She calculated that the probability that the Tapelift sample contained DNA of the applicant and one unknown person compared to the probability that it contained DNA from two unknown persons was approximately 6,400 to 1. She calculated that the probability that the Fluff sample contained DNA of the applicant and one unknown person compared to the probability that it contained DNA from two unknown persons was approximately 17,000 to 1.[72]
[72] Evidence of Ms Mitchell T1492/8-13 and 1498/3-13.
Criteria for permission to appeal
The criterion for permission to appeal is that the ground of appeal is reasonably arguable.[73] Although this is usually identified as the sole criterion, if the appeal would raise a question of general importance, I consider that this may be taken into account as more readily leading to the grant of permission to appeal (although the converse does not apply that absence of any question of general importance is a reason to refuse permission if the ground of appeal is reasonably arguable). By way of analogy, whether an appeal would raise a question of general importance or matters of public interest is relevant in the context of applications by the Director for permissions to appeal.[74] For example, in R v Hallcroft,[75] this Court held that the importance of appellate guidance on the proper approach to sentencing for offences of murder gave rise to public interest reasons for granting permission to appeal.[76]
[73] R v Parenzee [2007] SASC 316, (2007) 101 SASR 456 at [22] per Doyle CJ (with whom Anderson and Kelly JJ agreed); R v Milton [2009] SASC 44 at [7] per Gray, Sulan and Kourakis JJ.
[74] R v Tran & Tran [2011] SASCFC 153 at [7] per Kourakis CJ, Vanstone and Peek JJ; R v Wymond [2013] SASCFC 12 at [1] per Kourakis CJ, Vanstone and Blue JJ; R v Hallcroft [2016] SASCFC 137, (2016) 126 SASR 415 at [78] per Kourakis CJ (with whom Peek and Stanley JJ agreed).
[75] (2016) 126 SASR 415.
[76] At [78] per Kourakis CJ (with whom Peek, Stanley, Lovell and Doyle JJ agreed).
Ordinarily the consideration given by a single Judge to the question whether permission should be granted, refused or referred to the Full Court will be constrained compared to the extent of the consideration given by the Full Court to the question whether the ground or grounds of appeal are established. However, in some cases a more extensive consideration by a single Judge will be appropriate. In R v B, FG,[77] Kourakis CJ said:
My reasons are necessarily long because of the sheer volume of the evidence adduced at trial and the very many grounds of appeal. Permission to appeal is granted only in those cases in which an arguable case is demonstrated. It should not be thought that permission will be given simply because the grounds are many and the case is factually complex. Volume is not necessarily an indication of merit.[78]
[77] [2012] SASC 157, (2012) 114 SASR 170.
[78] At [4].
Expert evidence about DNA adduced at trial
The prosecution adduced expert evidence from eight DNA experts at trial:
·Mr Blankers and Dr Kokshoorn who gave evidence focusing on the NFI analysis of the DNA samples and on LRmix Studio;
·Susan Vintiner of the Institute of Environmental Science & Research (ESR) in New Zealand who gave evidence focusing on her independent review of the NFI analysis of the DNA samples;
·Ms Mitchell who gave evidence focusing on the FSSA analysis of the DNA samples;
·Professor Adrian Linacre of Flinders University who gave evidence focusing on analysis of low template DNA samples;
·Dr Taylor of FSSA and Dr Buckleton of ESR who gave evidence focusing on likelihood ratios and STRmix;
·Professor Gill of the University of Oslo and the Norwegian Institute of Public Health who gave evidence focusing on analysis of low template DNA samples, likelihood ratios and STRmix.
The evidence of each expert DNA witness was admitted without objection. The applicant did not call any expert DNA evidence and relied on cross-examination of the witnesses called by the prosecution.
The Judge summarised the explanation given by the witnesses about DNA profiling in the following non-controversial terms:
It is now well established that DNA (deoxyribonucleic acid) is a long molecule which governs a person’s characteristics. In the human genome (the whole of a person’s DNA), there are something like three billion sites called loci which contain information. Each locus has two pieces of information called alleles. Each allele is contributed by the person’s father or mother. Each allele contains a repeated four-part sequence. It is the number of times the four‑part sequence repeats itself that indicates the size of an allele. That size is given a number representing the number of repeats. The size of a father’s or mother’s allele may be different or may be the same. It is the size of the allele as indicated by the four-part sequence of repeats that allows DNA to discriminate between people. Ignoring identical twins, a person’s genome is believed to be unique. Nevertheless, the alleles at the vast majority of loci in a human being are the same size.
In DNA profiling, a small number of loci are compared between samples. Those chosen for comparison are regarded as the few which are highly discriminating between individuals. The DNA scientist, at each locus, compares the size of the alleles as explained above. By comparing two samples in this way, the scientist is able to see whether the profiles match. If they do not match, one can be excluded as coming from the other. If they do match it does not necessarily mean that both samples come from the same person because this will only be a minor sample of the whole genome. It means that one sample cannot be excluded as coming from the other.
At that stage, the strength of the DNA evidence is dependent upon statistics and approved mathematical systems. Depending upon what system is used, a “weighting” is given to each comparison.
…
The first step in the process was the extraction of DNA by automation...
…
The next stage of the process was to determine how much DNA was in the sample extracted. This process is known as “quantitation”...
…
The next stage was what is known as polymerase chain reaction amplification. The scientist targets a particular region of the DNA and copies those particular regions over and over again so that they can detect what DNA is in that sample at those particular regions. …FSSA in carrying out PCR uses a kit called the “Profiler Plus kit”. This kit targets 10 loci which are considered to be hypervariable between people, allowing for a greater discrimination.[79] The information at those loci is copied over and over again and each time a copy is made it is labelled with a fluorescent tag. …
[79] As explained below, the NFI used Next Generation Multiplex which targeted 16 loci and FSSA now uses GlobalFiler which targets 24 loci.
…
The next stage … was, after the DNA has been copied, to detect what DNA is present at each locus. The detection stage is done by a process called “capillary electrophoresis”. Capillary electrophoresis detects the fluorescent tags added during the previous PCR process. ... Using known standards, the speed at which the DNA passes through the detection window can be converted to a number representing the size of the DNA. The size of that piece of DNA is then represented by a number.
At the next stage, the raw data that is detected is input into software called “Genemapper”. Genemapper takes the raw data and compares it to known standards called an “allelic ladder”. The allelic ladder designates a particular allele number to DNA fragments dependent upon their size. The software generates a graphical representation of the DNA detected, which is known as an “electropherogram”.
…
... It is important at this stage to mention the evidence that was led, undisputed, comparing the development of programs used to interpret DNA profiles.
The method now commonly used is known as the “probabilistic software program”, of which STRmix is an example.
This is to be contrasted with the binary interpretation system. The binary interpretation system … requires scientists to interpret an electropherogram and determine whether it was possible or impossible for a particular explanation to give rise to the evidence profile. Explanations that are deemed possible were all given a probability of one, whereas explanations that are deemed impossible are given a probability of zero. If a scientist is able to rule out all the explanations but one, then the profile can be interpreted. Once a profile could be interpreted, a statistical weighting could be calculated for the result...
The binary method is to be contrasted with probabilistic models of interpretation which use likelihood ratios. It is undisputed that a likelihood ratio is a general statistical principle. It is the weighting of evidence in any investigation considering two propositions. With DNA analysis, the likelihood ratio is the statistical weighting of the DNA evidence given two competing hypotheses. The first is that an individual is a contributor to a DNA profile (the prosecution hypothesis) and the second is that the individual is not a contributor (the defence hypothesis). The likelihood ratio is the probability of the evidence given the prosecution hypothesis over the probability of the evidence given the defence hypothesis.
…
STRmix is a continuous model in interpreting a DNA profile by use of a complicated formula utilising more information in the DNA profile than before. The difference between a continuous model and a semi-continuous model is that peak heights are used as part of the information, in particular to differentiate between true and false contributors to a mixture.
The Judge’s reasons for verdict
The Judge summarised the lay and non-DNA expert evidence. The Judge summarised the expert DNA evidence and referred to two questions identified by the applicant’s counsel:
1. Whose DNA is on the pyjama top?
2. If it is proved that it is the applicant’s DNA, how did it get there?
The Judge summarised the defence submissions concerning asserted infirmities in the expert DNA evidence and the asserted inability of that evidence to prove beyond reasonable doubt that the applicant’s DNA was on the pyjama top. The Judge made the following finding:
Having observed and studied all of this body of the evidence very carefully and considered the arguments of the defence, I find that the results of both FSSA and the NFI have been proved beyond reasonable doubt.
None of the matters raised by Mr Charman, either specific or general, cause me disquiet. I accept beyond reasonable doubt the explanations of the various eminently qualified scientists set out above in relation to his criticisms. Although there is no onus on the defence, those experts were unchallenged by contrary evidence. I find that the reasons for the disparities between the results is as explained by the prosecution witnesses. Indeed, such explanations, putting aside their scientific complexity, really amount to matters of common sense: more sophisticated techniques will lead to different results. However, that does not invalidate those results obtained by using lesser techniques.
It is to be remembered, and I direct myself, that none of these results are evidence of the fact that the accused is the killer of Louise Bell. Nor, indeed is there direct evidence that it is his DNA on Louise Bell’s pyjama top. However, those results are pieces of circumstantial evidence led for the purpose of proving that the DNA on the pyjama top matches that of the accused.
The Judge summarised the defence submissions concerning the reasonable possibility that the applicant’s DNA transferred onto the pyjama top by an innocent process and made the following finding:
I have already found that the pyjama top was purchased as a Christmas present for the deceased. I accept the evidence of Colin Bell on that topic, despite the previous statement of Dianne Bell. I therefore draw the clear inference that the garment came into the deceased’s household sometime shortly before the Christmas of 1982. I also find that the last contact the deceased had with Petra Pfennig would have been, at the latest, in November or December 1982. This was around the time of the basketball pool party. I find it inconceivable that the DNA extracted by the tape lift or contained within the piece of fluff could somehow have been fortuitously transferred from the accused to Petra, then to the deceased who transported that piece of DNA back into the household where it at some time, and somehow, was deposited on the pyjama top. I find such a proposition fanciful. I find it proved beyond reasonable doubt that such innocent transference did not take place.
That being so, transference does not give rise to an explanation which would affect the scientific weightings of the comparison between the DNA of the accused and the DNA samples deposited on the pyjama top. I find those findings proved beyond reasonable doubt and use them as pieces of circumstantial evidence, to be assessed with the other circumstantial evidence in this case.
The Judge considered the whole of the circumstantial evidence, including the DNA evidence, as summarised at [5] above and concluded:
I direct myself that before I can find the accused guilty of the offence, I must be satisfied beyond reasonable doubt that a combination of some or all of those circumstances leads to the conclusion that it was the accused who abducted and killed the deceased. I have to exclude any other reasonable possibility beyond reasonable doubt.
I find that a combination of the above circumstances amounts to proof beyond reasonable doubt that the accused abducted and murdered the deceased. In particular, I find the evidence of the weighting of the DNA comparisons compelling, combined with the fact that I also find that the pyjama top was washed in tap water before being deposited on KD’s lawn. This leads me to the clear inference that there would have been more DNA on the top before washing. I add to that all the other circumstances including the accused’s connection with the Onkaparinga River, the comparison of accents with the voice of the person who telephoned KD and the opportunity the accused had, being alone in his house on the night of the abduction. I can see no other explanation to account for these proven facts other than his guilt.
The pieces of circumstantial evidence paint a compelling and consistent story which amounts to proof of the elements of the offence beyond reasonable doubt to the exclusion of any reasonable possibility of any innocent explanation.
The sole ground of appeal is that the verdict is unreasonable and incapable of being supported having regard to the evidence. There is no ground that the Judge’s reasons are inadequate or that the Judge made any specific error in reaching the Judge’s conclusions: the grounds advanced are that the evidence was incapable of supporting a conclusion that the applicant’s DNA was present on the victim’s pyjama top or excluding as a reasonable possibility the innocent transfer of the applicant’s DNA to the victim’s pyjama top (via secondary transfer). Accordingly the parties directed their submissions to the evidence.
First limb: proof the applicant’s DNA was on the pyjama top
The first limb of the proposed ground of appeal is that the evidence was not capable of proving beyond reasonable doubt that the applicant’s DNA was present on the victim’s pyjama top. There are seven particulars. The applicant relies on each particular as creating a reasonable doubt in itself but also relies on the cumulative force of the seven particulars combined.
The particulars are:
1.The DNA results obtained from FSSA and the NFI in relation to the same exhibits were so different that none of the results could be relied upon to the standard of proof required.
2.In relation to the exhibit named the fluff alternative testing of the Y‑STR produced a result that was capable of being interpreted as excluding the applicant as a potential donor of the DNA found on that exhibit but nevertheless the Forensic Science South Australia proceeded to a likelihood ratio.
3.That the STRmix software program used by FSSA to determine the likelihood ratios was of itself subject to a number of variations.
3.1 A natural variation.
3.2 The Californian code error.
3.3 Variations according to a laboratory’s settings and policies.
4.That if the exhibits had been compared to profiles on the Cold Case Database there may have been a significant number of matches some of which may have been to a likelihood ratio higher than of the Applicant’s.
5.With respect to the NFI the use of four replicants may have led to different alleles being called or different weights being given to alleles called which could have led to a different likelihood ratio.
6.That there was no attempt made to run the results obtained from the kits through the software of the alternate laboratories to determine the reliability of the outcomes.
7.That the more sensitive DNA kit that is now available [the kit being the process that produces the electropherograms], if used in the Applicant’s matter, may have led to a conclusion that it was not possible to proceed to a likelihood ratio.
It is necessary first to consider each particular separately and then to consider their combined weight.
Dutch results
Different results compared to FSSA
Particular 1 is the first and foremost contention by the applicant and is that the DNA results obtained from FSSA and the NFI in relation to the same exhibits were so different that none of the results could be relied upon to the standard of proof required.
Different results for the Tapelift DNA
In respect of the Tapelift DNA, the NFI likelihood ratio was between 9.6 and 36.9 billion to 1. The FSSA likelihood ratio was approximately 6,400 to 1.
Given that the NFI tested for alleles at 15 variable loci and FSSA only at nine, the NFI performed 34 cycles and FSSA only 28 cycles of magnification and the NFI detected alleles at 15 variable loci and FSSA only at five variable loci, it was inevitable that the NFI likelihood ratio would be multiple orders of magnitude larger than the FSSA likelihood ratio. Indeed, if they had been of a similar order of magnitude, that would have cast doubt on the reliability of one or other of the calculations.
The expert witnesses explained the different results in succinct and consistent terms.
Dr Taylor gave the following evidence:
Q.... Can you assist as to why there is that sort of difference between the two likelihood ratios generated.
A.Yes. That is what you would expect for a number of reasons. One is that they are using a different profiling kit, so they are targeting more loci and they are going to be getting more information from those more loci. They are also amplifying their sample using more PCR cycles, which, again, increases the sensitivity of the whole system and, again, you would expect more information as a result of that. So as you get more information you are going to get a different evaluation at the end. That is sort of common sense. If you have a certain number of say matching loci, if you have more matching loci, that evaluation is going to change. Then also they use a different statistical model. They use a semi-continuous system within LRmix Studio and STRmix is a fully continuous system. So, again, you would expect a slight difference from that statistical model as well.
Dr Kokshoorn gave the following evidence:
Q.Are you able to explain why a continuous model has a statistical weighting of 6,400 when your semicontinuous model has a range of 9 billion to 36 billion.
A.Yes. I have seen the DNA profiles that were generated by Forensic Science South Australia from this particular sample and they were very incomplete. We were asked to do DNA analysis of the same sample and obtained our own DNA profiles, which contained a lot more information. We performed our calculation on DNA profiles as we generate a DNA profile in the Netherlands and came up with the number that was just mentioned. I do believe that Forensic Science South Australia performed a calculation with STRmix on the DNA profiles that were created here at Forensic Science South Australia, very incomplete DNA profiles, and apparently came up with the number that you just mentioned. My explanation for that would be that the DNA profiles generated in the Netherlands contained a lot more information, a lot more alleles were observed, than the DNA profiles obtained here in Forensic Science South Australia.
Professor Gill gave the following evidence:
Q.... What I’m asking you is in a general sense what is the explanation if any for the disparity between those two likelihood ratios.
…
A.First all it does not surprise me that we have differences in the results. It’s a bit like comparing apples and pears I think because effectively the NFI are using a method which is more sensitive and that means they will visualise more information in the stain. If they are visualising more alleles which match the defendant then the likelihood ratio will increase. There will be a direct correlation with the number of alleles in the crime stain that match the defendant. For example, if the NFI system matched 12 alleles for example and the Australian system only matched six then that would account for the discrepancy. So I think it’s important to realise that this isn’t - this discrepancy is not just the software. The software might be part of it but you have to look at the entirety of the case and the way in which it was analysed and the biochemical method is crucial to this. One way to resolve this possibly - maybe not resolve it but the samples could be analysed by swapping the softwares - if we’re concerned about the software - there could be an analysis of the Australian result using the LRmix Studio for example and possibly vice versa. To summarise, I’m not surprised that there is this difference given that the biochemistries used were given.
…
Q.So, depending upon the different methodology in terms of the number of cycles, the number of loci and the software, a laboratory can deliver a likelihood ratio that is 6,400 and another laboratory can give a figure of 936 billion; is that correct.
A.Yes. In this example that seems to be the position. But you have to remember you are using low template analysis. It’s a bit like turning up the radio. You can hear a lot more as you turn the volume up. The Australian laboratory would have the volume turned down very low so they are just not picking up as much information.
Professor Linacre gave the following evidence:
Q.Now the point has been made that there's a big difference between the likelihood ratio arrived at of 6,400 and one in the billions, given both were run using material that came from the same sample. Why is it that such different likelihood ratios are generated in those circumstances.
A.Because two different kits - as you said yourself, one was using Profiler Plus and one was using NGM. In the first instance, that's one example. Where Profiler Plus looks at only nine and NGM looks at 15, the more you look at, if you generate data for those alleles, for those loci the higher that discrimination power is. So by using more loci, if you can generate data at the loci, then that increases that number automatically. The other event, the other reason is, just looking at the profiles I'm aware of, clearly the profile done with Profiler Plus, the DNA profile I'm looking at, this was done 28 cycles, as we say, and you can see it's typical of very low level DNA where we have generated results at six loci. But at other loci you can see there are potential peaks but very close to the baseline. Now that indicates to us that there might well be something there but given 28 cycles it's just not possible to see it. I'm only seeing alleles at six loci and one of them is called the 13 there's an allele there called 16, it's tiny, it's just above one of our thresholds you might call. So it could well have a partner, being a what we call heterozygote but it's below the base line, you can't see it but you know that there's something there.
Q.How does that differ from what occurred at the NFI.
A.By doing 34 cycles, which is the whole basis behind it, then there's a much better chance of those products, those things which are below baseline we are now looking at 28 cycles, appearing.
On the expert evidence given at trial, the difference in results did not throw any doubt on the NFI results or opinions expressed by Mr Blankers and Dr Kokshoorn.
Different results for the Fluff DNA
In respect of the Fluff DNA, the NFI could not under its protocols proceed to calculate a random match probability. The FSSA likelihood ratio was approximately 17,000 to 1.
Mr Blankers gave evidence that in 2012 the NFI had a protocol that it would not proceed to determine a random match probability unless it had either a single source or a clear major DNA profile.[80] The NFI detected more than one source of the Fluff DNA and there was no clear major profile.[81] The protocol did not permit the NFI to proceed to calculate a random match probability.
[80] Evidence of Mr Blankers T2896/3-10 and 2897/10-29.
[81] Evidence of Mr Blankers T2897/19-22.
Mr Blankers gave evidence that the limitation to a single source or clear major DNA profile was specific to random match probability. He said that probabilistic systems producing likelihood ratios (such as STRmix used by FSSA) are not so limited.[82]
[82] Evidence of Mr Blankers T2897/10-37.
By contrast, FSSA using less sensitive techniques and examining fewer loci detected only alleles from a single source and had a clear major DNA profile, albeit substantially incomplete. FSSA’s protocols permitted it to calculate a likelihood ratio. However, given the limited number of alleles detected, the calculated ratio was the relatively small figure of approximately17,000 to 1.[83]
[83] Evidence of Ms Mitchell T1492/8-13.
Professor Linacre gave evidence that the fact that the NFI could not under its protocol calculate a random match probability did not impinge upon the validity of the FSSA result.[84]
[84] Evidence of Professor Linacre T2687/7-9 and Dr Taylor T3046/23-33.
On the expert evidence given at trial, the difference in results did not throw any doubt on the NFI results or opinions expressed by Mr Blankers and Dr Kokshoorn.
Running results through alternate software
Particular 6 is that there was no attempt made to run the results obtained from the kits through the software of the alternate laboratories to determine the reliability of the outcomes. The applicant submits that, if this had been done, there may have been a more accurate basis to determine the reliability of the differing results. The applicant points to two short passages in the cross-examinations of Dr Kokshoorn and Professor Gill.
It was put to Dr Kokshoorn in cross-examination that, all other things being equal, a continuous model such as STRmix may be expected to give a higher likelihood ratio than a semi-continuous model such as LRmix Studio, a general proposition with which he agreed. He was then asked to explain why for the Tapelift a continuous model (used by FSSA) had a statistical weighting of approximately 6,400 to 1 when the NFI’s semi-continuous model had a range of 9 to 36 billion to 1. His answer was that the NFI looked at more alleles and used greater multiplication. This question and answer is set out at [58] above. The following further questions and answers were then asked and given:
Q.But do you accept that LRmix Studio is less sensitive in terms of the analysis of the profiles than STRmix.
A.Yes, because it uses less information.
Q.And I come back to the point that you've never been asked to look at the Forensic Science profiles and run it through your LRmix Studio.
A.No, we haven't.
Q.That would be the only way to compare the likelihood ratio from the profile obtained from Forensic Science using the same software; is that correct.
A.There's either one of two comparisons that you could do. You could have the DNA profiles obtained here in Australia and have them run through LRmix Studio, or the other way round, we could have the DNA profiles obtained at the NFI and have them run through STRmix.
It was not suggested that the NFI should have asked FSSA to run the DNA profiles obtained at the NFI through STRmix or that the absence of its having done so invalidated or threw doubt on the NFI’s results (or vice versa in respect of the FSSA profiles). The questions were limited to the context of explaining the difference of six orders of magnitude between the NFI and FSSA likelihood ratios which has been addressed above. Indeed, if FSSA had run the DNA profiles obtained at the NFI through STRmix, theoretically the likelihood ratio would only have increased.
Professor Gill was also asked in cross-examination to explain why for the Tapelift DNA the analysis by FSSA had a statistical weighting of 6,400 to 1 but the NFI’s analysis had a range of 9 billion to 36 billion to 1 and his answer was also that the NFI looked at more alleles and used greater multiplication. This question and answer is set out at [59] above. In the course of his explanation, he said:
One way to resolve this possibly - maybe not resolve it but the samples could be analysed by swapping the softwares - if we’re concerned about the software - there could be an analysis of the Australian result using the LRmix Studio for example and possibly vice versa.
This sentence was limited to the context of explaining the difference between the NFI and FSSA likelihood ratios which has been addressed above.
It is always possible to postulate more extensive or intensive testing and analysis and to speculate that this might have altered the results.
It was not suggested in cross-examination of any expert witness that the fact that there was no attempt made to run the results obtained from the kits through the software of the alternate laboratories impeached the results obtained by either laboratory. On the expert evidence given at trial, the fact that such an attempt was not made does not throw doubt on the NFI results or opinions expressed by Mr Blankers and Dr Kokshoorn.
Detecting at additional loci
Particular 7 is that the more sensitive DNA kit now available and which produces the electropherograms, if used in the applicant’s matter, may have led to a conclusion that it was not possible to proceed to a likelihood ratio. This particular applies specifically to the FSSA analysis by reference to GlobalFiler but might be considered also to have application to the NFI analysis.
Dr Taylor gave evidence that FSSA now uses GlobalFiler which detects alleles at 23 human variable loci plus amelogenin for gender. He said that, as a rule of thumb, where Profiler Plus examining ten loci had produced a likelihood ratio supporting inclusion of a suspect, GlobalFiler examining 24 loci could be expected to produce a likelihood ratio supporting inclusion of a suspect many orders of magnitude higher. Likewise, where Profiler Plus supported exclusion, GlobalFiler could be expected to support exclusion even more strongly. In both cases the original results would be confirmed and strengthened. However, in a minority of cases, the use of GlobalFiler examining 24 loci might show that there were six or more contributors, in which case a likelihood ratio could not be calculated. He said that the same applied (although obviously to a lesser extent) if one moved from examining 16 loci to 24 loci.
Dr Taylor was asked in cross-examination some general questions whether subsequent testing using a kit that examines more loci might adversely affect results produced by a kit that had examined fewer loci. His evidence included the following passages:
Q.The new kit or another kit, does it show you either more or less information, does it, or it changes the information you see. I mean, I want to know what the difference can be, an example of what the difference can be.
A.If you are moving to a more sensitive kit, it should show you the same information but then might also show you additional information.
Q.So it won't change the information in the sense of what is there, it might have added stuff. Is that the situation.
A.Correct, and I will give you an example of what can typically occur in cases where we have upgraded Profiler Plus results to GlobalFiler results.
Q.Yes.
A.We have had instances where in the Profiler Plus result you have an original strong profile originating from one contributor, then you update to GlobalFiler and the profile you then get is the same at the overlapping regions but then because it's more sensitive you also detect at a much lower level a secondary contributor.
Q.It does change it but it doesn't really change the results of seeing what you see but adds information to it.
A.Yes.
…
Q.So if you were to put an old sample into GlobalFiler 24 loci and it indicated that there were six contributors, just by way of example, then you wouldn't, under your guidelines, be able to proceed to using STRmix to obtain a likelihood ratio.
A.Correct.
…
HIS HONOUR
Q.See, what I am, and what Mr Charman quite properly is cross-examining you on, is what is going to concern me of course, and I want to hear some evidence and information on this, this was done with Profiler Plus, you got a certain result, the system has improved in the sense that it's more sensitive, more information, how does that affect the Profiler Plus results that I've got. Does it negate them, does it say 'Look, they might be wrong because we've got a better system now'. Where do you go. I mean originally there was only four loci they used to look at right back in the binary period.
…
A.Sure. Firstly I would just say that the information that we have obtained in our earlier system is not wrong. It might not be as bountiful as additional kits would provide but there is nothing wrong in the information that has been detected. It will be less and it will tend to be weaker which means that the evaluations or the likelihood ratios will tend to be weaker or lower. But in a general sense if someone has contributed DNA to a sample and you profile it in say Profiler Plus you would generate a likelihood ratio. If you then upgrade that to GlobalFiler you are obtaining more information, you should get a larger likelihood ratio favouring their inclusion. If that person hasn't contributed DNA to that sample then further DNA would tend to strengthen their exclusion from the profile. So what you would hope is the further information you provide allows you to distinguish more so a true from false contributor.
Q.So what I think Mr Charman, I am sure he is getting at, is that if you get more information and that more information shows more contributors could that lessen it.
A.It could lessen it. However, if that person is a contributor you would expect it to increase the likelihood ratio, even with some additional contributors. I guess the other just general point to make is that you can't carry out interpretations or second guess your information based on information that you haven't got or that you might have got in some other guise. You can only work with the information that you have and if it's inclusionary it's inclusionary and if it's exclusionary it's exclusionary.
Q.So, putting it another way, are you saying that trying to guess what another system would bring about is almost impossible, I suppose.
A.Correct.
As observed above more extensive or intensive testing and analysis can always be postulated and the subject of speculation as to the results.
It was not suggested to Dr Taylor, and it was not put to Mr Blankers or Dr Kokshoorn of the NFI, that the possibility that more information and more contributors might be found using a kit examining 24 loci invalidates or throws doubt on the results obtained at the NFI for the Tapelift DNA. The cross-examination of Dr Taylor did not throw doubt on the opinions expressed by Mr Blankers or Dr Kokshoorn.
Number of replicates
Particular 5 is that the use of four replicates may have led to different alleles being called or different weights being given to alleles called which could have led to a different likelihood ratio.
Mr Blankers gave evidence in chief that, when using random match probability with replicates, the NFI adopted a policy of generally using three replicates and using uniform/majority results. In 2011 Corina Benschop published an article in the Forensic Science International Genetics journal about the experimental use of different numbers of replicates which concluded that in general three replicates was appropriate but for very poor samples there would be an added benefit in running a fourth replicate.[85]
[85] Evidence of Mr Blankers T1688/17-1690/4.
In cross-examination, Mr Blankers said that using four replicates may have been advisable given that the Tapelift DNA was a low template sample, but there was insufficient material left to perform a fourth analysis in any event. Mr Blankers said that there was a clear major contributor based on three replicates and this was unlikely to change if a fourth replicate had been analysed. While he accepted the theoretical possibility that a fourth replicate might have precluded probabilistic match analysis, it is clear from his answers and from the results of the analysis of the three replicates that this was very unlikely.
Mr Blankers gave evidence that, when using random match probability, the purpose of using replicates was to deal with stochastic effects. It was realised at the time that the best way of dealing with stochastic effects would be through a probabilistic model, but at that time probabilistic models had not been implemented in regular casework. Later NFI adopted a probabilistic model in LRmix Studio.[86]
[86] Evidence of Mr Blankers T1688/4-16.
Again, it might be postulated that more extensive or intensive testing might be undertaken. It was not suggested in cross-examination of Mr Blankers that the fact that four replicates were not used in 2012 when using random match probability impeached the results obtained using three replicates or any reason why it should do so. In any event, this was superseded by the subsequent use of LRmix Studio and the calculation of a likelihood ratio in 2015 and it was not suggested that the fact that four replicates were not used impeached those results. The cross-examination of Mr Blankers did not throw doubt on the ultimate opinions expressed by him.
Conclusion concerning Dutch results
Considering the applicant’s contentions collectively, it was open to the Judge on the evidence to accept that the results of the NFI testing in relation to the Tapelift were proved beyond reasonable doubt. It is not reasonably arguable otherwise.
South Australian results
Variation with NFI results, alternate laboratories and additional alleles
Particulars 1, 6 and 7 have already been considered in the context of the NFI’s results.
Variations between laboratories
Particular 3.3 is that the STRmix software program used by FSSA to determine the likelihood ratios was of itself subject to variations according to a laboratory’s settings and policies.
Ms Mitchell gave evidence in cross-examination that different laboratories may have different policies and use different settings in detecting DNA in samples.[87] This could lead to the generation of a different DNA profile by two laboratories in respect of the same underlying sample.[88] There would be no difference for a good sample but would be differences for a low level sample.[89] For a low level sample, the potential differences would be “slight”.[90]
[87] Evidence of Ms Mitchell T1464/18-36.
[88] Evidence of Ms Mitchell T2280/4-2282/38.
[89] Evidence of Ms Mitchell T2280/31-2281/1.
[90] Evidence of Ms Mitchell T2282/18-35.
Ms Mitchell said that the possibility of another laboratory obtaining a different result did not affect the validity of the result obtained by FSSA using its policies and settings.[91]
[91] Evidence of Ms Mitchell T2545/26-2546/13.
On the expert evidence given at trial, the possibility of another laboratory obtaining a different result using different policies and settings does not throw doubt on the FSSA results or opinions expressed by Ms Mitchell.
Result of Y-STR testing
Particular 2 is that, in relation to the Fluff DNA, alternative testing of the Y-STR produced a result that was capable of being interpreted as excluding the applicant as a potential donor of the DNA but nevertheless FSSA proceeded to calculate a likelihood ratio.
Y-STR targets regions on the Y-chromosome and not on the X-chromosome. -STR information is transferred to paternal offspring, so that in a given family all biological males that are related will have the same Y-STR profile. There is only one allele at each locus.[92] Generally, Y-STR profiling is undertaken merely to show that a person is included or excluded and a likelihood ratio is not calculated.[93]
[92] Evidence of Ms Mitchell T1374/34-37 and Dr Taylor T1519/6-24.
[93] Evidence of Ms Mitchell T1494/12-19.
FSSA used a Y-STR test on the Fluff DNA in 2012 because it was more sensitive than Profiler Plus, which had only detected DNA at three variable loci, for the purpose of determining whether there was one or more than one contributor to the DNA.[94]
[94] Evidence of Ms Mitchell T2176/35-2177/3.
Ms Mitchell gave evidence that the Y-STR test resulted in the following profiles:
Locus Applicant DNA Fluff DNA DYS 456 14 14, 15 DYS 389I 13 - DYS 390 23 - DYS 389II 28 - DYS 458 18 - DYS 19 17 - DYS 385 12 - DYS 393 13 13 DYS 391 11 11 DYS 439 14 11 DYS 635 21 - DYS 392 12 - GATA H4 12 - DYS 437 15 - DYS 438 10 10 DYS 448 22 -
It was put to Ms Mitchell in cross-examination that the applicant could not be a contributor because his allele at DYS 439 was a 14 whereas the allele detected in the Fluff DNA was an 11. Ms Mitchell disagreed, pointing out that the presence of two alleles at DSY 456 showed that there were two contributors; there were many loci at which at least one if not two alleles had dropped out; the applicant’s alleles were present at the other four loci at which any alleles were detected; and hence the applicant’s allele may well have dropped out.
Ms Mitchell characterised the result as inconclusive, being neither inclusionary nor exclusionary. The cross-examination of Ms Mitchell included the following passages:
Q.And what I'm suggesting is that the presence of the 11 at D198 as the only allele must exclude Mr Pfennig as being a contributor to that fluff and debris from the pyjamas.
A.If that DNA profile, p.997, the evidence profile, originated from one contributor only and there was one difference, then we would exclude, but given that this is originating from two contributors and that some of the alleles from both contributors is not present, that does make that interpretation more difficult to determine whether he is excluded.
HIS HONOUR
Q.So there could have been a 14 that dropped out.
A.Yes.
…
HIS HONOUR
Q.You say the 11 could belong to the 15, Mr Pfennig's could have dropped out, is that right.
A.It's possible that at that location, that two individuals would have an 11 or that one individual has an 11 or the other one has dropped out because it's stochastically affected, so I can't say for absolute certainty whether or not that's happened and so I report that as it's inconclusive as to whether or not Dieter Pfennig is a contributor to that DNA profile.
Q.Therefore you can't weight it but you also can't exclude it either.
A.No, and I can't include it either.
…
HIS HONOUR
Q.So you are saying that exclusion is a possibility.
A.Yes, it's just I can't say one way or the other.
Professor Linacre gave similar evidence. The cross-examination of Professor Linacre included the following passages:
Q.If there is no 14 present, that suggests that he's excluded as a contributor to the fluff and debris from the pyjamas which are shown in D198.
A.Well, that is one interpretation. If you look at the profile, clearly there is the 14 that matches at DYS485, and 13 is in common at DYS393, 11 in is in common at 391, I agree there is no evidence of this profile present.
…
HIS HONOUR
Q.Is it consistent with exclusion. Could it be excluded. Could that be -
A.That is one interpretation.
Q.It is consistent with that.
A.I understand why you use the term 'consistent'. It is not a term I use because that would indicate one opportunity along with any other. It doesn't indicate whether it is the most likely.
Q.Exclusion can't be ruled out.
A.Exclusion can't be ruled out.
Q.That will do me, thank you. The other way you want to ask the question, does it necessarily mean it is excluded.
A.The answer to that would be no because of the nature of the low-level profile.
XXN
Q.If it's not excluded, just dealing with that as a second possibility, can that only be on the assumption that at 439 there must have been dropout of other alleles.
A.Yes.
Q.Are you able to then say that, given that at 439 on the fluff and debris electropherogram there has, as an alternative, been a dropout, that that other allele would be a minor contributor at that -
A.I think it is all minor, it is all low level. I can't say in this profile if anything is a major or a minor profile because most of it has dropped out.
The applicant contends that, on the evidence of Ms Mitchell and Professor Linacre, he was potentially excluded as a donor of the DNA. However, the use of the adjective “potentially” renders the “exclusion” meaningless because both Ms Mitchell and Professor Linacre said that the result was inconclusive: it may be that his DNA dropped out given that the DNA of both contributors had dropped out at the majority of the 16 loci examined.
The applicant contends that FSSA should not have proceeded to calculate a likelihood ratio using the complete chromosomes but this does not follow given that the result of the Y-STR was inconclusive.
The applicant contends that use by Ms Mitchell of the information that there were two contributors shown in the Y-STR test when using STRmix to calculate a likelihood ratio was inconsistent with her not using the information from the NFI testing that the profile was too complex to calculate a random match probability. However, the fact that the criteria contained in the NFI’s protocols in 2012 were not met for the NFI to calculate a random match probability does not invalidate or throw doubt on the FSSA results for the reasons given above. The Y-STR result has no impact on this.
FSSA likelihood ratio subject to variations
Particulars 3.1 and 3.2 are that the STRmix software program used by FSSA to determine the likelihood ratios was of itself subject to a natural variation and the ‘Californian code error’.
Dr Taylor explained that STRmix uses a mathematical statistical method known as Markov Chain Monte Carlo (MCMC) that is used in a broad number of applications including weather forecasting and social studies.[95] He said in cross-examination that there is a degree of variation in the results when STRmix is run on a given sample that is inherent in the MCMC method. The outer limit of the variation is one order of magnitude (for example, between 10,000 and 100,000 or between 100,000 and 1,000,000) but typically it would be much closer than one order of magnitude. The differences are larger with larger ratios. The smaller the ratio, the smaller the variation one tends to see.[96] When it was put to Dr Taylor that, if one started with 100,000 there might be a natural variation from 10,000 to 1,000,000, he disagreed, saying that this would involve a variation of two orders of magnitude.[97]
[95] Evidence of Dr Taylor T1572/23-1574/13.
[96] Evidence of Dr Taylor T3075/27-3076/24.
[97] Evidence of Dr Taylor T3075/34-37.
Dr Taylor said that, after STRmix had been in use for some time, a Mr Myers in California discovered an error in its software code (termed in the evidence “the California error”) which had a small effect on the likelihood ratio.[98] FSSA reviewed the difference resulting from the error and found that the differences were so small that a review of the likelihood ratio calculated in previous cases was not warranted.[99] The effect was so small that FSSA continued to use that version of STRmix for some months until the next version rectified the error.[100] The combined effect of the MCMC natural variation and the California error was less than one order of magnitude.[101]
[98] Evidence of Dr Taylor T3090/12-19.
[99] Evidence of Dr Taylor T3090/25-33.
[100] Evidence of Dr Taylor T3020/25-34.
[101] Evidence of Dr Taylor T3091/22-28.
The evidence of Dr Taylor showed that the likelihood ratio calculated by FSSA using STRmix for the Tapelift DNA of approximately 6,400 to 1 is not a precise probability but is a probability of the order of magnitude of 6,400 to 1. It is not unusual for scientific analysis to produce a result that is not precise but is accurate within a tolerance, such as being accurate within an order of magnitude. When the Judge reached the conclusion that the results of FSSA had been proved beyond reasonable doubt, such proof obviously was within the degree of accuracy described by the experts called by the prosecution. In any event, the relatively small probability calculated by FSSA using less sophisticated analysis was engulfed and superseded by the NFI results in relation to the question whether the DNA on the pyjama top belonged to the applicant.
The evidence of Dr Taylor showed that the likelihood ratio calculated by FSSA using STRmix for the Fluff DNA of approximately 17,000 to 1 is not a precise probability but is a probability of the order of magnitude of 17,000 to 1. When the Judge reached the conclusion that the results of FSSA had been proved beyond reasonable doubt, such proof was within the degree of accuracy described by the experts called by the prosecution. In any event, the relatively small probability calculated by FSSA in respect of the Fluff DNA was engulfed and superseded by the NFI results for the Tapelift DNA in relation to the question whether the DNA on the pyjama top belonged to the applicant.
On the expert evidence given at trial, the fact that the FSSA results were only accurate within one order of magnitude does not throw doubt on the results or opinions expressed by Ms Mitchell.
Potential matches in the Database
Particular 4 is that if the exhibits had been compared to profiles on the Cold Case Database there may have been a significant number of matches, some of which may have been to a likelihood ratio higher than of the applicant’s.
Dr Taylor gave evidence that FSSA’s testing of the Fluff detected eight alleles at only five loci (there were only six alleles detected at four human variable loci, leaving aside the amelogenin).[102] This is very low compared to the ten loci that are examined and accounts for the relatively low likelihood ratio of 17,000 to 1.[103] Dr Taylor expressed the opinion that a profile involving eight alleles at five loci was likely to match the profile of a number of persons whose profiles have been uploaded into the Database.[104]
[102] Evidence of Dr Taylor T3043/15-22. See table at [94] above.
[103] Evidence of Dr Taylor T3038/27-3039/4.
[104] Evidence of Dr Taylor T3117/2-22.
The likely matches are simply the consequence of the fact that eight alleles were detected at five loci. The fact that only five alleles at four human variable loci matched those of the applicant is already reflected in the low likelihood ratio of only approximately 17,000 to 1. It was not suggested to Dr Taylor, or any other expert witness, that the fact that other matches were likely with profiles in the Database threw doubt on this figure. The possibility of other matches with profiles in the Database illustrates why the prosecution relied on the NFI results for the Tapelift DNA to prove beyond reasonable doubt that the DNA on the pyjama top belonged to the applicant rather than the FSSA results for the Tapelift or Fluff DNA.
On the expert evidence given at trial, the likelihood of other matches to the Database does not throw doubt on the FSSA results or opinions expressed by Ms Mitchell.
Conclusion concerning FSSA results
Considering the applicant’s contentions collectively, it was open to the Judge on the evidence to accept that the results of the FSSA testing in relation to the Tapelift and Fluff DNA were proved beyond reasonable doubt. It is not reasonably arguable otherwise.
Conclusion concerning NFI and FSSA results
Considering the applicant’s contentions collectively in respect of the NFI and FSSA testing combined, it was open to the Judge on the evidence to accept that the results of the NFI and FSSA testing in relation to the Tapelift and Fluff were proved beyond reasonable doubt. It is not reasonably arguable otherwise.
It was open to the Judge on the evidence to accept that the results of the NFI testing in relation to the Tapelift proved beyond reasonable doubt that the applicant’s DNA was on the pyjama top. It is not reasonably arguable otherwise.
The Judge was entitled to take into account all of the circumstantial evidence summarised at [5] above in finding it proved beyond reasonable doubt that the applicant’s DNA was on the pyjama top. Thus, in R v Doheny and Adams,[105] Phillips LJ, Jowitt and Keene JJ said:
The significance of the DNA evidence will depend critically upon what else is known about the suspect. If he has a convincing alibi at the other end of England at the time of the crime, it will appear highly improbable that he can have been responsible for the crime, despite his matching DNA profile. If, however, he was near the scene of the crime when it was committed, or has been identified as a suspect because of other evidence which suggests that he may have been responsible for the crime, the DNA evidence becomes very significant. The possibility that two of the only 26 men in the United Kingdom with the matching DNA should have been in the vicinity of the crime will seem almost incredible and a comparatively slight nexus between the defendant and the crime, independent of the DNA, is likely to suffice to present an overall picture to the jury that satisfies them of the defendant's guilt.[106]
[105] [1997] 1 Crim App R 369.
[106] At 373.
The additional circumstantial evidence confirmed the conclusion based on the NFI testing that it was proved beyond reasonable doubt that the applicant’s DNA was on the pyjama top.
The first limb of the proposed ground of appeal is not reasonably arguable. I would not grant permission to appeal by reason of the first limb.
Second limb: possibility of innocent secondary transfer
The second limb of the proposed ground of appeal is that there was no evidence adduced by the prosecution capable of excluding as a reasonable possibility the innocent transfer of the applicant’s DNA to the victim’s pyjama top. There are four particulars:
1.The prosecution adduced evidence of possible factual scenarios when transfer could have occurred.
2.The scientific evidence could not preclude transfer of the cells or the material in those circumstances.
3.The likelihood ratio obtained with respect to the answer to question 1 is irrelevant in relation to how and when DNA is deposited and how long it remains.
4.There was evidence adduced that:
4.1 there were two unknown depositors of DNA upon a tapelift of the pyjama top (see P162); and
4.2 there was no other DNA said to link the applicant to the crime scene or the pyjama top.
Expert scientific evidence could not preclude transfer
Particulars 1, 2, 3 and 4.2 can be considered together and are non-controversial.
The prosecution adduced evidence from the DNA experts about the mechanisms for secondary (as opposed to direct) transfer via another person or object. That evidence was that secondary transfer occurs on occasions: its likelihood and extent depend on several variables including the nature, extent and length of the contact; nature of the relevant surfaces; length of time the DNA needs to persist on a surface; events occurring to the surface (including movement and washing); and how many successive transfers are involved.[107]
[107] Evidence of Ms Mitchell T2128/26, Dr Kokshoorn T2423/20-36 and 2424/1-20, and Dr Taylor T3040/3-38 and 3041/1-14.
The prosecution case, accepted by the Judge and not challenged on appeal, was that the pyjama top was only given to Louise Bell at Christmas and that the Pfennig family left Adelaide on the morning of Christmas day. It follows that potential mechanisms for innocent transfer of the applicant’s DNA to the pyjama top involved multiple steps.
Ms Mitchell was cross-examined about the possibility that the applicant’s DNA was transferred to Petra or her clothing; its being transferred in turn to Louise Bell or her clothing, and its being transferred in turn within the Bell household to the pyjama top. Ms Mitchell accepted these were possibilities, the likelihood of which would be affected by the factors referred to above.[108] Ms Mitchell said that she could not quantify the likelihood of each step of the posited transfer. Professor Gill said that a DNA scientist cannot quantify the likelihood of a particular posited secondary transfer: it is a question of fact.[109]
[108] Evidence of Ms Mitchell T2140/22-38.
[109] Evidence of Professor Gill T2991/2-6.
It was common ground at trial that the likelihood ratio that was relevant to the identity of the contributor of the DNA found on the pyjama top was irrelevant to the question how and when it came to be on the pyjama top.
It was common ground at trial that there was no other DNA found at the scene of Louise Bell’s abduction that was analysed and linked to the applicant. The evidence was, and judicial notice could be taken of the fact, that there was no DNA identification science practised in South Australia in 1983. Leaving aside the Tapelift and the Fluff, it was common ground at trial that no other material associated with the pyjama top that was analysed was linked to the applicant.
These particulars do not in and of themselves give rise to a reasonable doubt whether the applicant’s DNA was deposited on the pyjama top upon or after Louise Bell’s abduction. Rather, they form part of the background against which consideration is to be given to that question.
Two unknown contributors to tapelifts
Ms Mitchell gave evidence that DNA from tapelift 686-74.3.1.A was found to have a mixed DNA profile of at least two contributors excluding the applicant. The applicant contends that this rebuts the prosecution contention that any DNA deposited on the pyjama top before Louise Bell’s abduction would have been eliminated by the immersion in the Onkaparinga River and subsequent rinsing in tap water. These two contentions are but part of the overall question whether, on all of the evidence, a trier of fact must have had a reasonable doubt whether the applicant’s DNA came to be deposited on the pyjama top by secondary transfer, ie innocently.
Holistic assessment of possibility of secondary transfer
The prosecution contended at trial, and contends on the application for permission to appeal, that the possibility that there was a sequence of transfers beginning with the applicant’s DNA and ending with the pyjama top with the DNA subsisting at each stage is so remote as to be fanciful. The prosecution contended that, while the applicant’s DNA may have been transferred to Petra or her clothing as a first stage, on the innocence hypothesis sufficient quantities would need to have persisted during the time that elapsed before Petra or her clothing with that DNA may have come into contact with Louise Bell or her clothing, notwithstanding washing and other activities that tend to diminish and remove DNA from one’s person or clothing over time. Similarly, if the applicant’s DNA were transferred to Louise Bell or her clothing as a second stage, sufficient quantities would need to have persisted during the time that elapsed before Louise Bell or her clothing with that DNA came into contact with her pyjama top as a third stage, notwithstanding washing and other activities that tend to diminish and remove DNA from one’s person or clothing over time. Finally, a detectable quantity of the applicant’s DNA would need to have persisted on the pyjama top notwithstanding its immersion in the Onkaparinga River and subsequent rinsing with tap water.
The applicant contends that, in the absence of evidence to assess the likelihood of persistence at each of these stages, it is not possible to conclude beyond reasonable doubt that the applicant’s DNA was not transferred innocently by a secondary process.
In relation to persistence of DNA on the pyjama top notwithstanding its immersion in the Onkaparinga River and subsequent rinsing with tap water, the applicant also points to the fact that the DNA of at least two other individuals was found on tapelift 686-74.3.1.A from the pyjama top and their DNA appears to have persisted notwithstanding such immersion and washing. The prosecution responds that this DNA may have been deposited after the pyjama top was removed from the River and rinsed; for example by KD. The applicant responds in turn that this was not proved and it remains at least a reasonable possibility that the DNA had been deposited before Louise Bell was abducted.
In Fitzgerald v The Queen,[110] Sumner and Fitzgerald were charged with joint enterprise murder and aggravated causing serious harm with intent to cause serious harm. There was no direct evidence that Fitzgerald was present at the house at which the two victims were attacked at about 6 am. Six eyewitnesses were shown his photograph and failed to identify him. However, his DNA was found on a didgeridoo at the house, on which were also found apparent bloodstains containing the DNA of the two victims. Evidence was adduced that Sumner had attended a boxing match at which he twice shook hands with Fitzgerald at about 10.30 pm earlier that night. Sumner’s DNA was not found on the didgeridoo. The High Court concluded that the possibility that Fitzgerald’s DNA was transferred via Sumner to the didgeridoo or had been deposited directly by Fitzgerald onto the didgeridoo on a prior occasion had not been excluded beyond reasonable doubt and acquitted Fitzgerald.
[110] [2014] HCA 28, (2014) 88 ALJR 779.
There are points of distinction between the facts in Fitzgerald and the case at hand. There was no other evidence linking Fitzgerald to the crimes whereas in the present case the prosecution relied on the other items of circumstantial evidence summarised at [5] above. Six eyewitnesses at the scene failed to identify Fitzgerald whereas in the present case there were no eyewitnesses. The posited secondary transfer in Fitzgerald involved only two stages and a period of less than eight hours whereas in the present case the posited secondary transfer involved four stages over at least three months.
The Court of Criminal Appeal has not since the High Court’s decision in Fitzgerald considered whether a verdict is unreasonable or incapable of being supported having regard to the evidence in a case in which a major issue of secondary transfer has arisen. The question whether the verdict in the present case is unreasonable or incapable of being supported having regard to the evidence by reference to the possibility of secondary transfer is not a question that can or should be determined by a single Judge but rather is a question to be determined by the Court of Criminal Appeal. For this reason, permission to appeal should be granted.
Conclusion
Given my conclusion on the second limb, I grant permission to the applicant to appeal on the ground that the verdict is unreasonable or incapable of being supported having regard to the evidence.
0
11
0