England and Wales Court of Appeal (Criminal Division) Decisions

IN THE HIGH COURT OF JUSTICE
COURT OF APPEAL (CRIMINAL DIVISION)
ON APPEAL FROM THE CROWN COURT AT OXFORD
HIS HONOUR JUDGE ECCLES QC
T20077265

		Royal Courts of Justice Strand, London, WC2A 2LL
		24/03/2010

Lord Justice Thomas :

1. This is the judgment of the Court.
2. After a retrial of some 12 days, the appellant was convicted at the Crown Court at Oxford on 13 February 2009 before HH Judge Eccles QC and a jury of conspiracy to commit arson. He was subsequently sentenced to 10 years imprisonment less time on remand. He appeals against conviction by leave of the Single Judge and by leave from us on three matters:

i) The admission of Low Template DNA (LTDNA) evidence where the quantity was at the very bottom of the scale where a DNA profile could be reliably analyzed.

ii) The direction the judge gave on the status of the DNA evidence if the jury were to conclude that that they could not accept the evidence given by the Crown in relation to the interpretation of the components of the DNA profile.

iii) The judge's refusal to discharge a juror on the ground of apparent bias.

He seeks leave to appeal on a further matter.

iv) The judge's refusal to exclude evidence in relation to sparklers found at the appellant's premises when he was arrested.

The facts and the respective cases

3. The facts can be briefly summarised. The sports pavilion at Queen's College, Oxford was set on fire in November 2006. An attempt was made to set fire to a portacabin at Templeton College, Oxford in February 2007. Both incidents involved incendiary devices with sparklers used to act as fuses.
4. It was common ground that the attacks were planned and devices planted by animal rights activists as a centre was being built at Oxford University for scientific research using live animals. The attacks were publicised on an American website called Bite Back. The campaign totalled 10 attacks but only in these two were sparklers used as fuses.
5. It was the Crown's case that the appellant had been responsible for planning or causing each of the fires at Queen's and Templeton. They relied upon the following facts:

i) The appellant was a spokesman for an animal rights group known as SPEAK.

ii) In 2000 he had pleaded guilty to an offence of conspiracy to cause an explosion; he had admitted then that he and another planned to plant incendiary devices on lorries contracted to take animals for slaughter.

iii) After his release from prison he had continued to protest in respect of animal rights and was actively engaged in demonstrations in Oxford.

iv) A scenes of crime investigation after the fire at Queen's College uncovered 12 sparkler rods. They had formed part of the incendiary device which had been placed in the roof of the pavilion and the remainder of that device had been destroyed by the fire.

v) At Templeton College, the devices found were bottles filled with petrol with sparklers, matches and firelighters laid on towels to act as fuses.

vi) The devices discovered at Templeton were taken away and examined at Fort Halstead.

a) On the stalks of some of the matches a minute quantity of DNA was found and analysed by LGC Forensics using a LTDNA process shared by them and Cellmark for enhancing the amplified DNA. The Crown contended that the probability of obtaining the match between the profiles so generated and the profile of the appellant if in fact the DNA did not originate from him was less than 1 in 1 billion.

b) A minute quantity of DNA was also found on the caps of the bottles. This was subjected to both enhancement and the Low Copy Number (LCN) process described by this court in Reed & Reed [2009] EWCA Crim 2698. Although this permitted the calculation of a match probability, it was common ground this was insufficient to link the appellant to the attack.

vii) Upon the appellant's arrest and a search of his accommodation, officers found 14 or 15 packets of sparklers in a disused water tank. One packet was missing two sparklers. No tests were carried out to see if they were of the same chemical composition as those found at Templeton; they were of a similar length but when lit they gave off an effect of a different colour to those at Templeton. There were also found at his property lists of companies and individuals linked to animal research.

6. It was the appellant's case at trial that he was not involved, that since his conviction he had decided that he could play a more valuable role as a spokesman and not engage in arson. The DNA evidence was insufficiently reliable to be admitted.
7. Although there was strong suspicion that the appellant had committed these attacks, the LTDNA evidence was central to the Crown's case.

The preliminary ruling by the judge

8. At the first trial in October and November 2008, the appellant was charged with conspiracy to commit arson, possession of an article with intent to destroy property and possession of an explosive substance with intent - the sparklers. A submission was made on behalf of the appellant that the DNA evidence which the Crown sought to adduce was insufficiently reliable for it to be admitted into evidence. It was agreed that the judge should consider the evidence submitted on paper alone and then should hear oral argument.
9. The essence of the case on behalf of the appellant on reliability and admissibility was that the quantity of DNA recovered from the stalks of the matches was so small, namely less than 100 picograms (pg), and the method of its analysis developed by LGC Forensics insufficiently validated for the profiles to be reliable and admissible. Before the first hearing, the appellant relied primarily upon the evidence of Professor Jamieson who gave evidence before this Court in Reed & Reed, (see paragraphs 104-112) and Dr Scott Bader. The Crown relied upon the evidence of Ms Rosalyn Hammond, the forensic scientist who had carried out the analysis at LGC Forensics, Dr James Walker of LGC, Dr Paul Debenham of LGC, Mr Matthew Greenhalgh of Cellmark Forensic Services (Cellmark) and Dr Linacre of Strathclyde University. Dr Linacre had been a member of the review carried out by Professor Caddy into LTDNA to which reference is made at paragraphs 71 and 72 of the judgment in Reed & Reed and to which we refer at paragraph 33 below.
10. After a full review of the law, the judge concluded in a written ruling on 31 October 2008 that the test that a judge should apply was whether there appeared to be a risk that the evidence might be unreliable so that it would potentially mislead the jury rather than help them. Applying that test to the full review he made of the reports, he concluded that the evidence should be admitted.
11. The trial took place immediately thereafter. The jury were unable to reach a verdict in relation to the count of conspiracy to commit arson or the count of possession of an article with intent to destroy property, but acquitted him on the count relating to the sparklers. A retrial was ordered on the conspiracy to commit arson or possession with intent counts for January 2009.
12. By the time of the commencement of the second trial, the appellant had instructed Dr Daniel Krane, a professor of biological science at Wright State University, Dayton, Ohio who had given evidence in the Northern Ireland case of R v Sean Hoey [2007] NICC 49 and to whom reference was made at paragraph 6 of the judgment in Reed & Reed. The judge was asked to reconsider his ruling on admissibility in the light of the further evidence contained in the report of Dr Krane. The judge concluded that the DNA evidence was admissible.
13. Dr Krane subsequently gave evidence before the jury, as did Professor Jamieson and Dr Bader and those on behalf of the Crown to whom we have referred at paragraph 8 above.
14. Although the evidence called on behalf of the appellant was directed at matters which were the legitimate subject of disagreement for reasons we shall explain, an attack was made on behalf of the appellant on the integrity of LGC Forensics; it was alleged that their commercial interests and influence over their case workers had tainted their professionalism and objectivity. LGC Forensics were underestimating the problems which were associated with LTDNA and promoting its viability for financial reasons.
15. The summary by the judge in his summing up of the respective scientific cases of the appellant and the Crown is accepted by counsel for the appellant to have been accurate. We would ourselves characterise it as clear, concise and very helpful to the jury. The judge also referred to one further matter which is of great importance to this appeal - the fact that Ms Hammond accepted that, if her opinion was wrong on the interpretation of one of the DNA profiles, then her random match probability statistics were also wrong and she was not in a position to put forward any others.

Issues (i) and (ii): The DNA evidence

16. There are two issues before us in relation to the DNA evidence:

i) whether the judge should have excluded the DNA evidence on the basis that the quantities of DNA were so low that no reliable DNA profile could be obtained;

ii) whether the jury should have been directed to disregard the DNA evidence if they concluded that that they could not accept the evidence given by the Crown in relation to the interpretation of the components of the DNA profile, as there were no random match probability statistics in that eventuality.

17. We shall address each of these issues in turn but before doing so we must begin with a brief description of the technique used to carry out the LTDNA profiling analyses relied upon by the Crown (as used by LGC Forensics) which is in some respects different to the LCN process described in Reed & Reed and of the results obtained.

(a) LTDNA profiling methodology

18. The work was carried out by or under the supervision of Ms Hammond at LGC Forensics, with the assistance of other forensic scientists at Cellmark. Ms Hammond has a degree in Natural Sciences from Cambridge University and some 20 years experience of DNA profiling. After the stalks of some of the matches around one of the devices (referred to as CDR 4/5) recovered from Templeton College were swabbed at Fort Halstead, the DNA was extracted by conventional means. It was then suspended in about 60 microlitres (µl) of fluid. An attempt was made to quantify the amount of DNA present but this proved unsuccessful because the amount of DNA present was below the limit of detection of the apparatus used. This is a matter of some importance. It revealed the concentration of DNA was no greater (and possibly rather less) than 10 picograms (pg) per µl.
19. Ms Hammond then carried out a series of conventional PCR profiling analyses, each involving a standard 28 cycle amplification of the DNA contained in 10µl of fluid. Ms Hammond confirmed what necessarily followed from the quantification test, namely that each sample therefore contained no more than 100pg of DNA, equating to the DNA contained in about 15 human cells. Four amplifications were carried out; two using the familiar SGM-Plus technique which amplifies 10 Short Tandem Repeat (STR) loci and two using a similar technique called Identifiler which amplifies 15 STR loci (comprising the 10 STR loci amplified by SGM-Plus and five additional ones).
20. The products of these four amplifications were then profiled twice. First a standard profile was prepared. Then, in each case, the product of the amplification was enhanced or, as Ms Hammond put it, "cleaned up" and another profile prepared. As will be seen, this markedly improved the results in that significantly more alleles were detected. These could then be compared to a reference profile derived from the appellant.
21. A total of about 40µl of the original fluid was used in Ms Hammond's experiments, leaving about 20µl which was, we understand, offered to the defence to carry out such repeats or other experiments they considered appropriate. That offer was not accepted.

(b) The profiling results

22. The results of the profiling so obtained were presented to the jury in the form of two tables, providing, in a conventional way, a numerical representation of the number of STRs, which is to say the alleles, measured at each locus.
23. The standard profiles revealed so few alleles that no reliance was placed upon them. The Crown case therefore rested on the profiles produced from the four enhanced samples. These were analysed by Ms Hammond using the consensus approach, that is to say, she only included an allele for the purpose of calculating a random match probability if it appeared twice in separate profiles. This was described by the judge in his summing up as a conservative approach.
24. She initially applied this approach to the profiles generated from the products of each of the amplification techniques. Hence the two profiles generated using the enhanced products of the SGM-Plus amplifications were condensed into one set of consensus profile data and the two profiles generated using the enhanced products of the Identifiler amplifications were condensed into another set of consensus profile data.
25. The enhanced SGM-Plus consensus profile revealed eight alleles, all being alleles in common with the appellant's reference profile, generating a random match probability of approximately 1 in 2,300. The enhanced Identifiler consensus profile revealed 10 alleles, again all being alleles in common with the appellant's reference profile, generating a random match probability of approximately 1 in 32,000.
26. Ms Hammond then produced two sets of additional data, being what she called "combined enhanced results from both Identifiler and SGM-Plus". In the first she drew together all the alleles which appeared in either of the SGM-Plus or the Identifiler consensus profiles. This profile now contained 15 alleles, all in common with the appellant's reference profile, generating a random match probability of approximately 1 in 12 million. In the second, she went back to the original four enhanced profiles and drew together all the alleles which appeared in any two of them. This profile now contained 20 alleles, all in common with the appellant's reference profile, generating a random match probability of less than 1 in 1 billion.
27. As will be apparent from the foregoing, the consensus approach did not reveal any alleles which were different from the alleles in the appellant's reference profile. However, two such alleles did appear in the profile generated from the enhanced product of the first SGM-Plus run. At locus D2, three putative alleles were detected, namely 17 and 23, both in common with the appellant's reference profile, and 25, which was not. Similarly, at locus D18, three putative alleles were again detected, namely 14 and 19, both in common with the appellant's reference profile, and 13, which was not.
28. Ms Hammond attributed the putative allele 25 at locus D2 to contamination and the putative allele 13 at locus D18 to stutter. Importantly, she did not consider either to be evidence of a mixed profile. In relation to putative allele 25 at locus D2, she based her opinion upon three matters – (1) that it does not occur again in any of the other enhanced runs, (2) that there are three other loci which amplify more effectively and where one would expect to find evidence of other alleles in the case of a mixed profile and (3) that generally the overall consistency of the data was such she could not imagine there being any likelihood of them being derived from a mixed profile. As for putative allele 13 at D18, Ms Hammond explained this was in a typical stutter position and the underlying graph revealed an imbalance characteristic of this artefact.

(c) The appellant's case

29. The thrust of the evidence of Professor Jamieson and Dr Krane and Dr Bader was that the consensus approach was flawed in that it was overly subjective; that the quantity of DNA recovered was so low as to be below the stochastic threshold; and that when dealing with such low levels of DNA, stochastic effects become very much more common, as confirmed by the data generated by Ms Hammond and her co-workers in this case which show a significant number of drop-outs from run to run. Further, in these circumstances it was very difficult to establish whether something was a stutter or a genuine allele because the relevant peaks become very difficult to distinguish. Overall, Dr Bader thought that the DNA recovered from CDR 4/5 probably came from two contributors. Dr Krane thought that the safest course was to treat the results as inconclusive.

(d) Admissibility of the DNA evidence

30. The appellant's contention was that the judge erred in declining to exclude the DNA evidence altogether, alternatively that he erred in leaving to the jury the existence or otherwise of the stochastic threshold, and that he insufficiently emphasised the unreliability of DNA profiling techniques when dealing with DNA below quantifiable levels. It was argued that in the light of the decision of this court in Reed & Reed which, it is said, recognises the existence of a stochastic threshold of between about 100 and 200pg of DNA and, by implication, the inherent unreliability, and hence inadmissibility, of profiling evidence derived from the analysis of any smaller quantity of DNA.
31. The appellant's submission is, we conclude, founded upon a misunderstanding of the decision in Reed & Reed. This court recognised that in the current state of technology there is a stochastic threshold between 100 and 200 pg above which LTDNA techniques, including the LCN process used by the Forensic Science Service (FSS), can be used to obtain profiles capable of reliable interpretation. Specifically, the court observed that above this threshold a challenge to the validity of the method of analysing LTDNA by the LCN process should not be permitted in the absence of new scientific evidence. However, the court did not hold or make any observation to the effect that below the stochastic threshold DNA evidence is not admissible. To the contrary, the court explained at paragraph 48:

"… Above that threshold (often called the stochastic threshold), the stochastic effect should not affect the reliability of the DNA profile obtained. Below the stochastic threshold the electrophoretograms may be capable of producing a reliable profile, if for example there is reproducibility between the two runs."

32. It was therefore necessary to apply the relevant principles relating to the admissibility of expert evidence in cases of this kind which are set out in Reed & Reed at paragraphs 111 to 113. Although the Courts of England and Wales have adopted a more flexible approach in admitting expert evidence than some jurisdictions, a court must consider whether the subject matter of the evidence is part of a body of knowledge or experience which is sufficiently well organised or recognised to be accepted as a reliable body of knowledge or experience. If the field is sufficiently well established to pass the ordinary tests of reliability and relevance, then that is sufficient. The weight of the evidence should then be established by our familiar adversarial forensic techniques. As we have set out at paragraph 9 above, the judge set out a different test; it was too low.
33. We must therefore examine the evidence in the light of the correct principles. We consider the following points to be material.

i) Professor Caddy's review of April 2008 (conducted with Dr Linacre and Dr Taylor of Cancer Research UK), which is referred to extensively in Reed & Reed, observes in its executive summary that the science supporting the delivery of LTDNA analysis is sound and that the three companies (the FSS, LGC Forensics and Cellmark) providing this service have validated their processes in accordance with accepted scientific principles concerning both 28 and 34 PCR cycles for extracts containing less than 200pg of DNA. Professor Caddy recognised that, at these levels, stochastic and inhibition effects have an impact on the DNA profiles produced but continued that all those involved have established guidelines for profile interpretation. At paragraph 1.7 of his report, Professor Caddy makes clear that his review has encompassed the particular techniques employed in this case:

"Standard DNA profiling which uses 28 cycles works effectively with identifiable traces of body fluids but there are times when no identifiable body fluid is present. The amount of DNA in these samples may be present at very low levels perhaps corresponding to one or more human cells. Some of these samples are sometimes referred to as 'touch DNA' and may be present at levels similar to incidental DNA or that of low level contamination that would not normally be detected using standard DNA profiling. Modifications to obtain an STR profile from less than 200 picograms (pg) include: optimisation of the electrophoresis system, increasing the number of amplification steps from 28 to 34 cycles [6] and/or purification of the PCR product from a 28 cycles process. Any of these modifications results in an increase in the sensitivity of the test but may also increase the stochastic effects and the opportunity for detecting DNA not related to the alleged incident (either incidental or due to contamination). The stochastic effects include allelic "drop out", random allelic "drop-in" and an increase in stutter products. These processes confuse the outcome of such DNA profiling and are usually dealt with by repeating the process a small number of times, usually twice is sufficient. The stochastic effects are not limited to increased cycle number but occur even with 28 cycles when using low template DNA…"

ii) As this court also noted in Reed & Reed, the Forensic Science Regulator broadly accepted the conclusions of Professor Caddy's review in his response dated 7 May 2008. He concluded:

"4.1.1 Having considered the Review and discussed its conclusions with the FSAC and stakeholders. I am content that the science underpinning the LTDNA analytical services, as provided to the CJS, is sound and that the three forensic science suppliers offering such services have properly validated their processes. There is no flaw inherent in the process which prevents its use within the CJS.

4.1.2. The recommendations set out in the Review, and points raised by members of the FSAC and stakeholder organisations, demonstrate that there are areas where the current processes can be improved. LTDNA services can be separated into three sections: collection, analysis and interpretation. I believe the key areas where improvements can be made are the collection of and, probably most importantly, the interpretation of the evidence. I have, within this Response, set out the way in which I wish to achieve these improvements.

4.1.3. The ability to improve on the current approach does not mean that the approach should not be employed within the CJS. As long as the scientist reporting the results of LTDNA analysis complies with the duties and obligations placed on expert witnesses the CJS will appreciate the nature and value of the evidence provided."

iii) Dr Linacre, who had some 15 years experience in the field of forensic DNA analysis, explained in his evidence that in conducting the Caddy review he and his colleagues carried out an extensive investigation of the processes of protocol design and review used by the three principal service providers and of the data they had generated using their LTDNA procedures. He also observed that all three organisations are accredited (or in the process of securing accreditation) under ISO 17025, the standard to which analytical science laboratories the world over seek to adhere, and which requires their procedures to have been tested and verified. He was firmly of the view that they conduct their LTDNA procedures in a reliable and robust manner.

iv) These matters were confirmed by Dr Matthew Greenhalgh and Dr Paul Debenham.

v) Dr Krane expressed concern that, as of early 2007, LTDNA techniques had not been adequately validated and that, at least in the USA, LTDNA results are treated with great caution. Overall, he believed that the results derived from the very low quantities of DNA analysed by Ms Hammond should be treated as inconclusive. We have no doubt that Dr Krane's opinions were honestly held. But it is to be noted that in expressing his concerns about adequate validation, Dr Krane was considering the position at a time before the Caddy review and the response to it from the Forensic Science Regulator.

34. It is apparent from the foregoing that there is now a considerable body of opinion from respected independent scientists and the Forensic Science Regulator that LTDNA techniques, including those used to generate the profiles relied upon by the Crown in this case, are well understood, have been properly validated and are accepted to be capable of generating reliable and valuable evidence. At these very low levels of DNA, the dangers presented by the possibility of stochastic effects, including allelic drop-out, drop-in and stutter are very real and must be fully appreciated, but they may often be addressed by repeating the process a number of times, as Professor Caddy recognised.
35. There will of course be occasions where profiles generated from less than 200pg are wholly and obviously unreliable. We anticipate that the Crown would never seek to adduce such profiles in evidence. If it put forward such a profile, then the unreliability would be pointed out in the report of the defence expert and, if not accepted by the Crown's expert in the exchange that must take place under Part 34 of the Criminal Procedure Rules, the judge would have to consider the dispute; if they were unreliable, he would exclude them.
36. There will be other occasions where the probative value of the profiles is more debatable. In such cases the evidence may properly be adduced and it must then be addressed and its weight established by adversarial forensic techniques. But we do not accept that these are reasons for ruling out LTDNA evidence altogether. In our judgment, the science of LTDNA is sufficiently well established to pass the ordinary tests of reliability and relevance and it would be wrong wholly to deprive the justice system of the benefits to be gained from the new techniques and advances which it embodies, in cases where there is clear evidence (adduced in the manner discussed) that the profiles are sufficiently reliable.
37. In the context of the present appeal, we have also reached the conclusion, applying a higher test than that applied by the judge, that the judge's conclusion was correct and they were admissible in evidence as sufficiently reliable. We recognise that the profiles were derived from unquantified samples of DNA of less than 100pg and that this raised entirely legitimate grounds for scientific dispute which the appellant was right in testing before the judge. However, the Identifiler and SGM-Plus techniques used by Ms Hammond and her co-workers are well established and were used in an entirely conventional way with a standard 28 cycle amplification followed by purification of the PCR product. The process was repeated four times, which may be considered rather generous in the light of Professor Caddy's review. The profiles which were generated revealed a large number of alleles which for the most part did not require subjective interpretation (being above 50 relative fluorescence units). It is right to record that there is evidence of drop-out from run to run, but this is hardly unexpected given the low quantity of DNA analysed. On the other hand, the consensus data relied upon by the Crown show what may be considered a remarkable consistency in that all of the consensus alleles match those in the appellant's profile. In other words, the consensus profiles do not suggest the procedures suffered from drop-in or stutter such as to render the results inherently unreliable. Indeed, this is reflected in the statistics derived from the consensus profiles to which we have referred and about which there was no dispute. At their most powerful and when derived from all duplicated components, these give rise to the match probability of less than 1 in 1 billion. We believe that these were all matters properly admitted in evidence.
38. We cannot leave this aspect of the case without commenting on the attack made on the integrity of the LGC Forensics and, by implication, Cellmark. Whatever may be the position in other jurisdictions, it is the duty of an advocate and an expert in this jurisdiction not to embark upon an attack on the integrity of other experts unless there is an evidential basis for doing so. There was none in this case. The attack made on the integrity of LGC Forensics and Cellmark was without foundation and should never have been made. As we made clear in Reed & Reed at paragraph 74(v) there can well be a difference of opinion between experts on LTDNA, but there should be no question of the good faith of those involved in LTDNA being put in issue. This is a case where there is a proper disagreement between experts but the course taken by those giving evidence on behalf of the appellant went into matters for which there was no foundation. Not only was the attack on the good faith of the Crown's witness wholly deplorable and unwarranted, but it also was a great disservice to the appellant's case.

(e) The direction to the jury: their course of action if they did not accept the Crown's case on the interpretation of the profile

39. We turn therefore to the second ground of the appeal – whether the jury should have been directed to disregard the DNA evidence if they concluded that they could not accept the evidence given by the Crown in relation to the interpretation of the components of the DNA profile.
40. This ground of appeal is founded upon the detection in the first run of the putative alleles 25 at locus D2 and 13 at D16, neither of which matched the profile of the appellant. It will be recalled (as we have set out at paragraph 27 above) that Ms Hammond attributed these to contamination and stutter respectively. The defence experts, on the other hand, considered they were inconclusive or, in the case of Dr Bader, evidence of a mixed profile. If the latter were correct, there was no dispute that the match probability statistics could no longer be relied upon.
41. The judge first directed the jury in relation to this issue as follows:

"So far as the consensus approach is concerned in general terms, members of the jury, Rosalyn Hammond explained it to you, but the important thing that she was saying, and the prosecution invite you to consider in this way, is that there are two parts to the consensus approach.

The first is that you decide from all the results that you have whether you have a single or a mixed profile or whether you have a single profile plus contamination. She says that somebody with experience is best placed to decide that on all the information that you have. So that includes whether part of the information is looking at three particular components which you were told about that she says would amplify more readily than others and so would be expected to show if there was a mixed profile, and you will remember the debate about those three components at D3, 8 and 19.

So once you have done that, that where a component appears twice at least then you include that for statistical purposes in working out the match probability otherwise you leave it out on a conservative basis. I am sure you understand by now the meaning of the match probability.

She agrees that if the judgment about the components is wrong then the statistics are wrong, and so in this case if it was your conclusion that any of her evidence about a specific component was wrong or may be wrong then of course that would affect the statistics that appear in the right hand column of the match probability and there you don't have alternative statistics on that basis. You are still invited to look at the evidence as a whole but you must bear in mind then you would have to be cautious because you would not have any precise figure to put on the match probability.

She accepted that there is an element of subjective judgment but no more than anywhere else in forensic science. She agrees that there is, as I say, subjective judgment and what she agreed was that somebody else might hold a contrary view to hers and that you may not be able to say who was right and who was wrong. In other words, there is no, as it were, answer at the back of the book. There is no independent machine if people hold contrary views to tell you in these circumstances who is right and who is wrong. It is a question of expert evidence and scientific judgment, and indeed that is, the prosecution say, why forensic scientists give evidence and other experts give evidence."

42. Importantly, the judge recognised and directed the jury that if Ms Hammond's judgment about the components of the profile was wrong then the statistics were wrong and no alternatives were provided. Nevertheless, the jury were still invited to look at the evidence as a whole, but cautiously because they would not have any precise figure to put upon any match probability.
43. Shortly afterwards, and in the absence of the jury, the appellant submitted that, absent reliable statistics, the jury should be directed to disregard the DNA evidence altogether. The judge rejected that submission, but gave the jury the following further direction:

"Finally this, members of the jury, that I said that if it should be your conclusion that in calling some component Miss Hammond may have been wrong in the conclusion that she arrived at about a particular component, that that would therefore destroy the statistical figure that has been given as a match probability, and I should emphasise that if you reach that point, although the evidence is available for you to reach your own conclusions about looking at the evidence as a whole and looking at the range of match probabilities which are given at lower levels, depending on where you say an error might have occurred if it occurred, you would have to be very careful indeed to arrive at firm conclusions and have to exercise caution because you would not have a statistical figure to put on it and you could not put your own figure on it because you are not experts. So that is a word of caution. But I have identified for you I hope sufficiently the various areas where there is need for caution in the interpretation of this evidence."

44. The appellant submitted that this too was an inadequate direction, as it gave the jury no proper guidance as to the weight they should put upon the DNA evidence in the event that they did not accept Ms Hammond's evidence as to the components of the DNA profile. It was submitted that the judge ought to have directed to the jury to disregard the DNA evidence entirely
45. In R v Doheny; R v Adams [1997] 1 Cr App R 369, Phillips L.J., giving the judgment of the court, said in relation to the summing up of DNA evidence:

"When the judge comes to sum-up, the jury are likely to need careful directions in respect of any issues of expert evidence and guidance to dispel any obfuscation that may have been engendered in relation to areas of expert evidence where no real issue exists. The judge should explain to the jury the relevance of the random occurrence ratio in arriving at their verdict and draw attention to the extraneous evidence which provides the context which gives that ratio its significance, and that which conflicts with the conclusion that the defendant was responsible for the crime stain."

46. Similarly, in R v Bates [2006] EWCA Crim 1395, this court was concerned with the problem of partial profiles and, in particular, whether DNA evidence should be excluded on the basis that it is impossible to ascribe any statistical value to the potential exculpatory effect of the voids in a partial profile. In the particular circumstances of that case the trial judge declined to exclude the evidence and, on appeal, the court concluded that he was entirely right to have taken that course. It explained at paragraphs 29 and 30 that such evidence nevertheless remained probative and admissible provided that the jury are given a sufficient explanation to enable them to evaluate it:

"29. …It is important to understand that the results of the testing procedure and the statistical analysis based on them indicate what proportion of the population has the reported alleles at the relevant loci. In the case of the samples taken from areas 2 and 4 it is one person in 610,000 in each case, or a total of roughly one hundred persons in a population of 60 million. That would remain the case even though there might be an allele in one of the voids which exculpated the appellant. If, on the other hand, a "missing" allele matched the appellant's profile, the match probability would be reduced and the chances that the sample had been contributed by the appellant increased accordingly.

30. We consider that the judge's approach to the question was entirely correct. We can see no reason why partial profile DNA evidence should not be admissible provided that the jury are made aware of its inherent limitations and are given a sufficient explanation to enable them to evaluate it….... In many cases there is a possibility (at least in theory) that evidence exists which would assist the accused and perhaps even exculpate him altogether, but that does not provide grounds for excluding relevant evidence that is available and otherwise admissible, though it does make it important to ensure that the jury are given sufficient information to enable them to evaluate that evidence properly."

47. Applying the approach in Doheny and Bates, the dangers inherent in evidence founded upon the analysis of less than 100 to 200pg of DNA make it particularly important that the jury are given sufficient guidance to enable them fully and properly to evaluate the evidence in relation to the components of the DNA profile where there is a disagreement about them. In this case the judge properly directed the jury that if they did not accept Ms Hammond's evidence as to the putative alleles 25 at locus D2 and 13 at D16, then that would, as he put it, destroy the match probability statistics relied upon by the Crown.
48. However, in our judgment the judge then fell into error in directing the jury that, in those circumstances, they could reach their own conclusions on the DNA evidence. It is fair to say that the judge urged the jury to exercise caution and be very careful in arriving at firm conclusions because they were not experts in statistics. However, we believe that only served to emphasise the void in which they were left. They had no guidance from the experts and no guidance from the court to enable them to conduct an evaluation of the evidence for themselves. In this court, counsel for the Crown put the position graphically; if the jury rejected the interpretation of the components of the profile put forward by Ms Hammond, "the statistics provided went out of the window". But although the Crown appreciated this consequence, the Crown had not provided any alternative statistics in the event the jury did not accept Ms Hammond's evidence. It followed in our view, that if the jury did not accept her evidence on the interpretation of the components of the profile, then the jury should have been told to acquit, as there was no basis on which they could assess the match probabilities themselves. Of course, if there had been alternative statistics, then these would have been left to the jury and the jury been directed accordingly.
49. We reach this conclusion with considerable regret. In his otherwise admirable summing up, the judge expertly addressed all the evidence and the complex issues in clear terms about which no complaint has (or could possibly be made) been made. However, the judge ought to have directed the jury that if Ms Hammond was wrong in her conclusion that the DNA profiles were single rather than mixed, then on the only evidence before the court at the trial the DNA evidence must be disregarded. The judge having failed to do so, the jury may well have embarked upon a task of evaluation for which they were not equipped. This means their verdict cannot be regarded as safe.
50. In the light of that conclusion we can deal with the remaining two grounds of the appeal more shortly.

Issue (iii): Should the judge have discharged a juror on the ground of apparent bias?

51. In accordance with practice for a case of this kind, the jury were asked at the outset of the trial whether they were in any way connected with Oxford University or animal rights activists. The jury were thereafter sworn in.
52. As we have explained, as part of the evidence called to validate the enhancement process developed by LGC Forensics for Low Template DNA, the Crown called Mr Matthew Greenhalgh, Director of Forensic Science at Cellmark. Prior to his giving evidence and while Ms Hammond of LGC Forensics was giving her evidence, a juror sent a note to the judge in which she disclosed that she worked for Cellmark.
53. The juror in question was asked into Court and she confirmed that she had worked as a clerical officer in the Returns Department of Cellmark since January 2009. She confirmed that she had not heard of Mr Greenhalgh.
54. The appellant applied for the discharge of that juror, arguing that a critical issue for the defence was the science of LTDNA and, in particular, the nature of the work being done by companies such as Cellmark and LGC Forensics who were prioritising commercial interests over evidential reliability.
55. In a very clear ruling, the judge rejected the submission. He concluded that the juror had a clerical job involving no scientific knowledge. It was accepted that she was not personally biased. The issue related to the question as to whether a fair minded and fair observer would conclude that there was a real possibility that she would be perceived as being subject to a level of partiality in deciding whether or not to accept the evidence of Mr Greenhalgh. He concluded that he had to have regard to the fact that she did not know Mr Greenhalgh and had a very low position, that an informed observer could not and would not anticipate that she would do anything other than carry out her duties conscientiously.
56. The legal principles are not in dispute. They are set out in Re Medicaments and Related Classes of Goods (No. 2) [2001] 1 WLR 700 and R v Khan [2008] EWCA Crim 531, [2008] 2 Cr App R 13. It was submitted on behalf of the appellant that the judge's decision was wrong. The attack being made on Cellmark was an attack on its good faith which, if it succeeded, would mean that Cellmark might suffer commercially; in a time of recession, a person who had secured even a clerical job, would wish to be sure that the company did not suffer adversely. It was also important to note that Cellmark was a relatively small company and, although she did not know Mr Greenhalgh, she would be very concerned for the company's future. On behalf of the Crown it was submitted that the judge had reached the correct decision in view of the fact that she had a clerical job, was not concerned with the science, did not know Mr Greenhalgh and his evidence was only one part of that relied upon by the Crown.
57. We are entirely sure that the judge was right in the conclusion he reached. The judge carefully weighed the competing considerations but was right to attach decisive importance to the fact that the juror had a clerical position, did not know the director concerned and that the director concerned was one of many experts being called by the Crown. This ground of appeal fails.

Issue (iv): The judge's exercise of his discretion in respect of the evidence about sparklers

58. As we have explained, the jury at the first trial in October and November 2008 had acquitted the appellant on the count of keeping explosives with the intent to endanger life or property which related to the sparklers found hidden in the water tank. At the retrial objection was made to the admission of the evidence in relation to sparklers on the basis that its probative value was considerably outweighed by the prejudice that would be introduced. It was asserted that there was no evidence that the sparklers had been purchased prior to the fires at Queen's and Templeton, the evidence to support the assertion that the missing sparklers were used in the Templeton devices was thin and there was no evidence to support the theory that he had used the missing sparklers to test the fuse mechanism used in the devices. On the other hand the prejudice was great. The judge refused to exercise his discretion under s.78 of PACE to exclude the evidence.
59. In our judgement he was entirely right to do so. The evidence did have a clear probative value and it would have been wrong to exclude it from the evidence put before the jury. The Single Judge refused leave to appeal on this point and the matter was not argued before us in the oral argument. We agree with the decision of the Single Judge and Mr Bentley was right not to argue the point before us. We refuse leave on this ground.