35.1. The Forensic Science Regulator recognises a need for the regulation of forensic science techniques just as much as regulation of providers and practitioners.1 I agree. The output of the most competent practitioner is no more reliable than the limits of the technique that is being applied. This chapter begins by looking at the foundations of the technique or methodology of fingerprint identification.
35.2. Fingerprint examiners identify marks as having been made by a particular person. Professor Champod and Mr Chamberlain state: "Fingerprint identification in the UK and around the world is understood to mean that a mark has been attributed to a particular individual to the exclusion of all others, although it is seldom articulated in this way. 'Others' refers often to any human in the world, living or dead."2
35.3. Given that examiners 'individualise' marks, the underlying assumption is that fingerprints are unique to an individual. As mentioned in chapter 2, the assumption of uniqueness is the subject of discussion and on-going research but is a premise about friction ridge skin.3 Fingerprint examiners work with impressions.4 The uniqueness of fingerprint detail and its persistence over time mean no more than that fingerprints are capable of providing a reliable basis for identifying an individual. That is not to be confused with the separate question whether evidence given by fingerprint examiners relating to the comparison of a particular mark and a print from a known individual reliably identifies that individual as the maker of the mark:5
"The question is less a matter of whether each person's fingerprints are permanent and unique - uniqueness is commonly assumed - and more a matter of whether one can determine with adequate reliability that the finger that left an imperfect impression at a crime scene is the same finger that left an impression (with different imperfections) in a file of fingerprints."6
Studies of the variability in findings of fingerprint examiners
35.4. Research carried out by the Metropolitan Police7 sought, among other things, to measure the extent of agreement between fingerprint examiners across a series of marks on a scale from good to very poor quality. The research found that there was consensus between the examiners on the easiest and most difficult marks.8 The proportion identifying marks decreased with difficult marks and the proportion deciding that the marks were insufficient increased.9 The research also showed that there was disagreement on the most difficult rotational marks.10 At 75% rotation the experts disagreed 50/50, but there was more consistency for marks with superimposition and reduced pressure.11
35.5. Other research suggests that different examiners will analyse the same complex mark differently. Professor Champod explained that when the quality of the mark is reduced variations between examiners as to the number of points found in agreement become quite large, and he referred to the study by Evett and Williams in this regard. The Evett and Williams report showed that different examiners interpret the same marks and prints in very different ways. In one case one examiner saw 14 and another 56 points based on the same material. A similar test carried out with Swiss examiners, based on the material used in Evett and Williams, produced similar results.12
35.6. Mr Kent carried out two experiments with a colleague in the Home Office Police Scientific Development Branch13 around 1995/1996 designed to ascertain the level of digital image quality required by examiners to make an identification on the introduction of digital photography.14 Examiners were asked to evaluate marks of varying quality but not to compare the marks to prints. They were provided with one mark at a time and asked to indicate how many characteristics they could find that could be used for comparison purposes. The first experiment involved examiners marking high resolution photographs and the second used on-screen images. Data from these experiments was analysed on an examiner by examiner basis. This showed substantial variations in the numbers of characteristics found by examiners. For one scene of crime mark the results were:
(i) one examiner found between 5 and 10 characteristics;
(ii) four found 10 to 15;
(iii) five found 15 to 20;
(iv) three found 20 to 25; and
(v) one found between 25 and 30.15
It was suggested that the problem lay with small bureaux or inexperienced staff but analysis of the results did not support either theory.16
The findings of the Inquiry
35.7. The evidence to the Inquiry also demonstrated variability in the findings of fingerprint examiners. There was a spread among those who agreed the identification of Y7.
35.8. Those who disputed that conclusion reported a lesser number of potentially matching points: Mr Wertheim at most five points, and the NTC reported three23 or five24 depending on the image studied by them.
35.9. The differences in the number of characteristics observed and the interpretation of those that were observed foreshadowed a difference of opinion among examiners as to the result in relation to Y7. Differences in detail do not, however, necessarily lead to differences as to the result. In the case of QI2 Asbury the conclusions of PSNI and SCRO coincided but PSNI found only ten points in sequence and agreement, whereas Mr MacPherson, Ms McBride and Mr Geddes, studying another image, reported 17.25 Where there is clarity in ridge detail in mark and print it ought to be possible to say that one view is more clearly right and the other wrong, but if there is poor clarity of detail (as there was in QI2) there can be understandable differences of opinion at the margins which may not affect the validity of the conclusion on a more limited number of common points. The challenge is to determine whether inconsistencies among examiners are merely incidental (as in QI2 Asbury) or substantive (as in Y7 and QI2 Ross).
35.10. At one level the implications of that challenge could be viewed narrowly as raising only a question about the competence of individual practitioners but on closer examination it does demonstrate uncertainties about the limits of performance of the technique itself that transcend subordinate issues of individual practitioner performance.
Fingerprint evidence as opinion not fact
35.11. The previous studies and the findings of the Inquiry support the conclusion that fingerprint evidence is a matter of opinion not fact but that simple proposition merely gives rise to two consequential issues:
(i) From the practitioner's perspective, what factors are material to the formulation of a reliable opinion?
(ii) From the perspective of a fact-finder, be it a judge or a jury, how is the opinion to be properly assessed?
35.12. In addressing those overlapping questions it is necessary to begin by considering the key variables that are integral to the technique applied in carrying out a fingerprint comparison. These can be focussed in five questions:
(i) Are the materials supplied (mark and print) of suffi cient quality for comparison purposes?
(ii) Can the examiner accurately observe sufficient characteristics in mark and print for a reliable comparison?
(iii) Can the examiner reliably interpret those characteristics in such a manner as to determine which match and which differ?
(iv) Can the examiner ascertain a reliable explanation for the characteristics that differ?
(v) Can the examiner find sufficient matching characteristics to justify the inference that the mark is uniquely identifiable as having been made by a specific person?
35.13. The discussion of these questions emphasises that fingerprint identifications can depend on a complex series of subjective judgments. That is integral to the technique of fingerprint identification. The focus of the discussion in this chapter is on how examiners reach their conclusions but a proper understanding of these matters is equally important to the ability of the fact-finder properly to assess the reliability of the opinion. An important link between the two is the extent to which the examiner is conscious of the separate judgments comprised in his conclusion and his ability to be more transparent in both the report that is provided and ultimately in the evidence presented in court.
Definition of terms
35.14. Judge Edwards, the co-chairman of the committee that produced the report of the National Academy of Sciences, has drawn attention to the failure of forensic scientists generally to use standard terminology.26 This is an acute issue when addressing the process of reasoning applied by fingerprint examiners in reaching a conclusion on a comparison.
35.15. It may broadly be said that an identification can be made when the 'points' in mark and print 'agree' and an exclusion made when they 'disagree' but what is a 'point' and by what criteria is it determined whether there is 'agreement'?
35.16. INTERPOL's European Expert Group on Fingerprint Identification has produced a Method for Fingerprint Identification27 with one set of definitions that could be used in discussing that question and, more recently, during 2011 the Scientific Working Group on Friction Ridge Analysis, Study and Technology (SWGFAST)28 published its 'Standard Terminology of Friction Ridge Examination'.29
35.17. The 2011 OIG report uses the SWGFAST terminology and draws a distinction between a 'difference', a 'dissimilarity' and a 'discrepancy'; 'difference' being used as the generic term and 'dissimilarity' being used for the subset of 'differences' due to the natural variation in the appearance that can occur due to distortion; and 'discrepancy' being used for differences in appearance "potentially requiring exclusion".30
35.18. Any debate about terminology can obscure the more substantive issue raised here. To take the SWGFAST model as an example, the essence of the distinction between the two subsets, a 'dissimilarity' and a 'discrepancy', is not semantic. The boundary between the two is determined by 'tolerance', a term which SWGFAST defines as: "The amount of variation in appearance of friction ridge features to be allowed during a comparison, should a corresponding print be made available." That begs the substantive question: what amount of variation in appearance is permitted in a proper allowance for 'tolerance'? To that there is no finite answer.
35.19. The INTERPOL text also subsumes the concept of 'tolerance' within the definition of its basic term, which is a 'point of agreement': there is said to be a 'point of agreement' if there is "a similarity that meets a specific value and where that similarity falls within the ruling tolerance."31
35.20. Subsuming 'tolerance' within the definition of 'dissimilarity' or 'point of agreement' obscures the fact that there is no objective standard by which to determine the appropriate limit of 'tolerance'. Absent such a standard there is no clear boundary between a 'dissimilarity' and a 'discrepancy' or points in 'agreement' and points in 'disagreement'. The paramount consideration is that 'tolerance' is but one way in which fingerprint examiners seek to respond to the fact that two impressions from the same finger will not necessarily correspond exactly. Depending on the personal threshold for 'tolerance' one examiner might conclude that two impressions are 'similar' because the differences are within tolerance. A second may consider the variation to be beyond the limit for 'tolerance' and hence say that the detail is 'dissimilar'. If that second examiner considers that there is some specific explanation for the 'dissimilarity', such as a double touch, that examiner could conclude that there is no 'discrepancy' and might identify. A third examiner might see no satisfactory explanation for the 'discrepancy' and therefore conclude that mark and print are not identical. The three examiners are engaged in the same task, which is determining whether a conclusion of identification can be reached despite some difference between the impressions and the issue is the same: do they have a satisfactory explanation for the difference?
35.21. Because examiners can approach the same issue from different perspectives the discussion in this chapter will, for consistency, use the one common term 'difference' without resorting to any subsets. The term 'difference' is used in a generic sense to embrace any lack of exact correspondence between mark and print. It includes ridge characteristics that are different in shape or in type (i.e. having the appearance of a ridge ending in one impression and a bifurcation in the other), ridge characteristics that are in different positions relative to the core or not in the same sequence and ridge characteristics present in one but not in the other.
35.22. A fuller discussion of the potential for variation in the opinions of examiners also requires discussion of the meaning of the term 'point' or 'characteristic', which brings in the significance of third level detail. For simplicity the discussion of the five questions in paragraph 12 will take level 2 detail as its base and, therefore, the term 'characteristic' will be used solely to refer to second level detail. Level 3 detail is discussed later in the chapter.
Question 1: Are the materials supplied (mark and print) of sufficient quality for comparison purposes?
35.23. Differences in the method of detection and recording of impressions can affect the appearance of the mark and complicate the comparison process.32
35.24. A misunderstanding regarding the method of detection can itself lead to differences of opinion regarding an identification. SCRO in their 2006 report on QD233 said "it is a fundamental starting point of any comparison to know how the mark was developed and recorded." It is apparent that some fingerprint examiners had to make assumptions about the methods used in detecting and recording some of the marks considered by the Inquiry. In the case of QD2 the assumption made regarding the possible use of ninhydrin may have been erroneous and may have been a contributory factor in Mr Rokkjaer and Mr Rasmussen disputing the SCRO identification. The labelling of QD2 was also not helpful. The image used in Production 98 had a number of different impressions and it is possible that the one that the Danish experts examined was not the one identified by SCRO.34
35.25. Details regarding the method of detection and the particular impression that is being compared may be considered to be objective facts that should be made clear to all by appropriate clerical recording. There are, however, subjective elements.
35.26. If the quality of the impression is very clear it is much less likely that there will be differences of opinion among examiners and the fact-finder can have much greater confidence in the identification. XF, the mark on the gift tag, is a good example of this.35
35.27. Marks with less clarity can be challenging.
35.28. It might be thought from the discussion of marks Y7 and QI2 Ross that differences of opinion among examiners boil down to competing professional judgments regarding the observability of ridge detail or the interpretation of observed detail. However, there can be a preliminary issue unrelated to a conflict of professional judgment. Image quality can be an important factor. This was seen with QI2 Asbury. The same practitioner can reach opposite conclusions depending on the particular image of the mark that is studied; and different groups of examiners can disagree if studying a single image but agree if a range of images is made available.36
35.29. Judgments about the 'sufficiency' of the quality of the materials used in making a comparison can be personal and hence differences of opinion among fingerprint examiners may have their origins in their personal assessment of the 'suitability' of the material being used. There may not be consensus among examiners regarding the 'best' image of the mark or the 'best' print. Given that image quality is a critical variable the integrity of the image that the examiner is relying upon requires to be carefully monitored.
35.30. At various times the materials used by Mr MacPherson, Mr Mackenzie and Mr Swann have included a variety of secondary copies of images of Y7 produced by photocopying or by printing reproductions of digital copies. It is to be expected that differences of opinion among experts that turn on examination of such copy material will be readily resolved in a case governed by Scots law by the application of the best evidence rule to exclude evidence relative to copy images of doubtful provenance. The quality of the copy is suspect and that undermines the reliability, and hence admissibility, of evidence based on it. 37
35.31. However, the best evidence rule would not have resolved all of the disputes addressed by the Inquiry. There can be variations in lighting, exposure and contrast in photographs taken separately and even in different images reproduced from the same negative.38 Each of these images can have equivalent status in the eyes of the law as a first generation image but no one image is necessarily the most 'suitable' for all examiners. PSNI did not initially identify QI2 Asbury despite access to a number of the original photographic images studied by SCRO. It was only after obtaining the negative and developing photographs that they considered to be 'suitable' that they were able not only to identify the mark but to agree SCRO's identification of it.39
35.32. Variability is not confined to the crime scene mark. Prints taken under controlled circumstances can vary. The prints of Ms McKie's left thumb provide two examples of that. The first is that the point that came to be labelled SCRO 4 looks more like a ridge ending or a bifurcation depending on whether one looks at the plain or the rolled impression taken by the police.40 The second is that the prints taken by the police were considered by some to be inadequate because they did not show the outer edge of the thumb and Mr Wertheim took a number of prints himself in order to capture the ridge detail in that area.41 Mr Swann was critical of even the plain impressions taken by Mr Wertheim and argued that it was necessary to use a rolled impression42 and in his presentation to the Inquiry he used the rolled impression sent to him by Levy & McRae.43 This was one of the points of significance in relation to the debate concerning the Rosetta.44
35.33. The dispute concerning the top section of QI2 Ross, particularly in relation to Mr Swann's points 4 and 10, provides a further illustration of variability in opinion being ultimately referable to the quality of a controlled print. The ridge structure in Miss Ross's print is broken and therefore open to interpretation. Witnesses were able, with equal conviction, to interpret an interruption in the ridge detail in the print as either a matter of substantive significance (a genuine break in the structure of the ridge) or an irrelevance (the product of the incomplete state of the impression).45
Question 2: Can the examiner accurately observe sufficient characteristics in mark and print for a reliable comparison?
35.34. The Metropolitan Police "tipping point" research46 found that fingerprint practitioners apply personal thresholds to both the qualitative and the quantitative variables.47 This can be considered using the generic term 'event' that was discussed by Professor Champod in his evidence to the Inquiry.48 An 'event' is simply a feature in the mark (or print) and it could be a carry through from the underlying substrate (such as the outline of the edge of a groove in wood) or part of the fingerprint itself (a ridge ending or a bifurcation). A fingerprint examiner must firstly consider whether any particular 'event' is observable.
35.35. There can be variations even in the observation of an event. SCRO points 11-14 in Y7 are good examples. Some fingerprint examiners reported observing ridge characteristics at these locations while others said that there was nothing observable.
35.36. There was a debate whether a fingerprint practitioner, by training and experience, can see details that the lay eye would miss.49 That does not account for SCRO points 11-14 because some trained eyes reported seeing them, while others did not. The Inquiry has no reason to doubt that it takes a trained eye to spot a relevant pattern but, once observed, a trained fingerprint practitioner should be able to demonstrate the existence of at least some 'event' to the fact-finder, be that a judge or the members of a jury. In the case of SCRO points 11-14 in Y7 I reached the conclusion that the existence of these points had not been demonstrated.
35.37. It is recommended that the test be adopted that features (or 'events') on which examiners rely should be demonstrable to a lay person with normal eye sight as being observable in the mark. The fact-finder can trust the evidence of his own eyes: either he sees some 'event' in the location indicated or he does not. If not, the evidence of the examiner on that point can be discounted.
35.38. Interpretation of observable events raises separate issues.
Question 3: Can the examiner reliably interpret those characteristics in such a manner as to determine which match and which differ?
35.39. Assuming that the presence of some 'event' can be demonstrated, the next question is the proper interpretation of that event. Is it no more than a reflection of the underlying substrate; or is it a part of the mark itself and, if so, is it a ridge ending or a bifurcation? This calls for the exercise of professional judgment and is dependent on a number of variables including the clarity of the impression, the personal threshold that the practitioner is applying and, ultimately, his personal opinion as to the 'proper' interpretation of the detail.
35.40. SCRO 750 in Y7 is an example of the preliminary issue whether any observable 'event' is a feature of the underlying substrate or part of the mark. It is towards the edge of the impression and views among the experts differed, one of the questions being whether, if there is any observable event at all, it is the bifurcation relied upon by Mr MacPherson and Mr Mackenzie51 or only some "grey-ish looking thing" indistinguishable from the "other noise" in the vicinity, as Mr Zeelenberg argued.52 My conclusion was that the event is so indistinct as to be indecipherable.53
35.41. SCRO points 4-6 in Y7 are useful illustrations of the next issue. Looking to Figure 10 in chapter 25, it may readily be accepted that each of the lines marked A-D is some 'event' forming part of the impression but the detail is incomplete and affected by smudging. Assumptions can be made about how best to read the gaps in ridge detail and, as a consequence of the rival assumptions made, the SCRO examiners (supported by some independent examiners) were able to advance the argument that these points matched the print of Ms McKie while others (notably Mr Wertheim, Mr Zeelenberg and Mr Grigg) argued that there was a difference between mark and print at one or more of these points sufficient to exclude
ie.54 This illustrates how highly subjective fingerprint evidence can be. The rival conclusions were simply the products of a series of conflicting assumptions made regarding the interpretation of incomplete ridge detail in the mark.
35.42. Subsumed within that issue is a more subtle and potentially more fundamental one. Assuming that the correct interpretation is that what appears as a ridge ending in the mark is a bifurcation in the print (or vice versa), what is the significance of this type of difference?
35.43. The INTERPOL 'Method for Fingerprint Identification' focuses on 'events' without insisting on an exact discrimination between ridge endings and bifurcations:
"Due to moisture, pressure or even over (or under) inking a true ending ridge might show as a bifurcation and vice versa. Those differences are within normal tolerances and have no fundamental relevance, i.e. they [do not] conflict possible [donorship] as with all level two detail."55
35.44. That assumes that a difference between a ridge ending and a bifurcation is necessarily within 'normal tolerances'. One of the principal lessons in the Brandon Mayfield case56 is that such a difference may not be within 'normal tolerances' and, indeed, that there can be 'fundamental relevance' in the distinction between these two characteristics. The Office of Inspector General (OIG) report 'A Review of the FBI's Handling of the Brandon Mayfield Case' was published in January 2006 and post-dates the INTERPOL 'Method for Fingerprint Identification'.
35.45. In the Mayfield case the FBI misidentified a crime scene mark from the Madrid bombing (LFP 17) as the mark of an American attorney, Mr Mayfield, when it was actually made by an Algerian national Mr Daoud. Subsequent analysis showed that there were ten 'features' (or 'events') in sequence common to the prints of Mr Mayfield and Mr Daoud. Such was the lack of clarity in the mark that the exact type of the 'feature' (i.e. a bifurcation or ridge ending) was ambiguous. The similarity in the 'features' in Mr Mayfield's print and LFP 17 led the FBI examiners into making a misidentification of Mr Mayfield as the donor of the mark before the competing print of Mr Daoud was shown to them. The differences between the prints of the two men lay, in part, in the discrimination of the precise type of these 'features'. What was a ridge ending in one was a bifurcation in the other.57 Each of the ten features had been interpreted by the FBI in such a way as to match Mr Mayfield's print but after suspicion fell on Mr Daoud it was found that an alternative interpretation of the same features matched Mr Daoud. When identifying the mark as that of Mr Mayfield, the FBI had failed to give adequate consideration to the incomplete nature of the agreement between the corresponding ridge details in mark and print. The OIG report concluded that "the 'quality' of the agreement was inadequate to support the conclusion of identification."58
35.46. The same is evident in relation to both Y7 and QI2. Some of the observed 'events' were ambiguous as can be seen from the contrasting characterisations of some of the ridge characteristics by Mr MacPherson and Mr Mackenzie and this was a consequence of the lack of quality of each mark.
35.47. In relation to Y7, SCRO points 4-6 again serve as an example, with Mr MacPherson interpreting SCRO 4 as a bifurcation and Mr Mackenzie seeing it as a ridge ending, each interpretation being possible because the detail in both mark and print is ambiguous.
35.48. The contrast was even more pronounced in relation to QI2 Ross.59 There are two clear examples.
35.49. The findings of the OIG in the Mayfield case and my findings in this Inquiry support the conclusion that it does not necessarily suffice for two or more examiners to agree that there is some 'event' or a series of 'events' in common in mark and print. Determination of the precise type of characteristic may have a bearing on the question whether there is an identity. A difference in the type of the characteristic might support an exclusion and certainly merits careful attention when considering whether an identification has been established. Each examiner must approach incomplete detail in an impression (mark or print) with conspicuous care because, depending on how the ambiguity is resolved, the same ridge detail could be understood to be equally supportive of the opposite conclusions of identity or exclusion. Moreover, where two examiners are proceeding on the basis of conflicting interpretations of ambiguous 'events', they need to give consideration to the question identified in the Brandon Mayfield case, which is whether the 'quality' of the assumed agreement in ridge detail is adequate to support the conclusion of identification.
35.50. There is no hard and fast rule. Impressions from the same finger can vary and, in particular, a ridge characteristic in the skin can reproduce inconsistently as a ridge ending or bifurcation in separate impressions. SCRO 4 in the print of Ms McKie's left thumb is a case in point. Without examining her finger no one is to know which is the 'true' characterisation. For that reason, it may be open to fingerprint examiners in appropriate circumstances to make a finding of identification even when there is a difference in the characterisation of the corresponding 'events' in mark and print. Whether or not that is 'appropriate' in any particular case is another facet of the broader issue of the tolerances being applied in the comparison. In the case of Y7 and QI2 Ross my finding is that the ridge detail in the marks did not have sufficient quality to justify a finding of identification. The fact that the SCRO examiners held conflicting views regarding the interpretation of ambiguous 'events' in Y7 and QI2 was not only indicative of that lack of quality but also demonstrated that they had applied an inappropriate degree of tolerance in the comparison of these marks and had, as a result, fallen into error. 60 For the future, where examiners differ in their interpretation of ridge detail in a complex mark I would expect this to be carefully considered in the technical review that I recommend for complex marks.61
35.51. The degree of tolerance applied is an integral part of the next question.
Question 4: Can the examiner ascertain a reliable explanation for the characteristics that differ?
35.52. Uniqueness of friction ridge detail in skin depends on the precise configuration of the ridge patterns. It might be thought that it ought to follow that even a single difference between two impressions ought to require a finding of exclusion. A single difference between two impressions may be indicative of an exclusion but an identification can be made where there are differences. It depends on the significance of the difference. This is yet another example of the fundamental proposition advanced by Ashbaugh regarding the distinction between ridge detail in the skin, as opposed to detail reproduced in impressions: "Statements about the two media are not necessarily interchangeable."62
35.53. Given the propensity for the precise appearance of ridges to alter depending on the manner of deposition, it is not realistic to expect that the detail in two impressions made by the same finger will coincide exactly. In fingerprint comparison allowance has to be made for inevitable variations in detail and this leads to a distinction between what Professor Champod called 'within source' variations (i.e. the type of difference that can occur when any individual deposits his print) and 'between source' variations (i.e. any difference indicative of the fact that the prints come from two different individuals). The presence of some differences between mark and print would not preclude a finding of identification if the difference is properly to be regarded as a within source variation. On the other hand, if the difference is a between source variation there is an exclusion.
35.54. The first task of the fingerprint examiner is to ascertain whether any difference is capable of reliable explanation as a within source variation. This involves two issues. The first issue is whether the difference is within tolerance. Mr McGinnies described tolerance aptly as "a clarity bridge".63 Tolerance enables an examiner to bridge across differences. The second issue is whether differences that exceed the limits of tolerance are capable of satisfactory explanation consistent with identity. The residual, or third, issue is whether a conclusion of identification can be reached despite the presence of an unexplained difference.
35.55. These three issues shade into each other because examiners can approach the same point of difference from a variety of perspectives and the same issue of principle arises under whichever heading the discussion proceeds: in what circumstances can a finding of identity be made when there is a difference (or differences) between mark and print?
35.56. The degree of tolerance that it is appropriate to apply is a subjective judgment: "There is no manual of tolerance levels like you would have for a machine."64 In the absence of any objective measure, examiners work to personal tolerances.
35.57. Professor Champod explained that varying levels of tolerance can be justified depending on the manner in which the mark has been made.65 In some cases tolerances will be lower, e.g. where there are clear characteristics. In other cases the tolerances may be wider, the example given being a mark on sticky tape: such a mark may be distorted by the skin stretching as it lifts from the tape and this may affect the shape and angle of the core. An examiner will allow for such distortion to have occurred and will consequently apply quite a large tolerance when looking for correspondence.66
35.58. The issue of the appropriate level of tolerance is controversial when it comes to marks which are of poor quality.
35.59. Ashbaugh discusses "tolerance for discrepancy" and states that "The idea of allowing more tolerance with poor prints appears to be the reverse of what it should be."67 It is counterintuitive because, as Professor Champod stressed, the larger the degree of tolerance applied the greater the chance of an adventitious match between mark and print. It follows that applying larger tolerances to marks of poorer quality increases the chances of a finding of identity in relation to marks that are less reliable because of their reduced quality.68 Mr Zeelenberg drew attention to the cautionary advice in INTERPOL's 'Method for Fingerprint Identification' that "the paradox [is] that one may be inclined to accept more differences in bad prints under the umbrella of distortion than one would accept in better quality prints."69 The INTERPOL European Expert Group has argued that there should be a rule that tolerances should not vary dependent on the quality of the impression.70
35.60. The point is made in chapter 28 that a number of factors ought to have alerted SCRO to the inappropriate tolerances being applied in the identification of Y7 and QI2 Ross. One was the fact that Mr Geddes could not agree the full complement of 16 points in Y7 despite Mr MacPherson's demonstration, to which can be added the similar inability of a number of other examiners, including the senior examiner and the quality assurance officer, to find 16 points. The second is that, on closer analysis, though Mr MacPherson and Mr Mackenzie agreed in the result there were variations between them on the interpretation of 'matching' points in Y7 and QI2, as already discussed in the context of question 3 above. Notwithstanding the move to the non-numeric approach, which on the face of it reduces the relevance of the number of points considered to be in agreement, conflicting estimations of the number of 'matching' characteristics, as much as differences of opinion as to the interpretation of the same ridge detail, can raise issues concerning the degrees of tolerance being applied.
35.61. The current approach to training can be seen in the evidence of Mr McGinnies.71 Starting from the premise that tolerances are built into the comparison process "not to explain things away but to realise why things may look different",72 trainees learn how different factors such as deposition pressure and movement can affect (or distort) a mark. They learn this by experience and by studying development techniques and demonstration marks identified by more experienced examiners, including unusual and difficult marks that have been discussed at the national trainers' forum and, therefore, have broad acceptance. The inclusion of examples which have acceptance at a national level does at least guard against an insular approach and the risk that one bureau may be operating to tolerances which are inconsistent with others. It remains the case that levels of tolerance are informed by the conventional wisdom of fingerprint practitioners, rather than scientifically validated experiments.
35.62. Professor Champod suggested a number of ways to guard against the risks associated with tolerance. The first is for the examiner to apply a distinct analysis stage before the comparison is made in order to be clear in advance if large tolerances are being applied to certain characteristics. Those tolerances must then be properly considered at the evaluation stage by the individual examiner.73 Secondly, Professor Champod said that he took the view that if a feature was reliable it should be reliable to a set of examiners looking at the same mark.74 Consequently, he advocated that at the verification stage complex marks should be subject to an additional check. Examiners should not only be concerned with the question whether they agree in the conclusion, they should also conduct a technical review of the features used in the individual comparisons and if there are divergences of view among the examiners as to the characteristics relied upon that should trigger a discussion about the tolerances applied and hence the reliability of the conclusion.75 The 'multiple procedure' applied in the Netherlands incorporates such a technical review at the verification stage.76
35.63. The need for separate procedures to be applied to the comparison of 'complex' marks is discussed in chapter 39.
Explanations for differences that are outside the limits of tolerance
35.64. A finding of identity can be made if any differences are capable of satisfactory explanation. The question is how to determine if any particular explanation is 'satisfactory'.
35.65. A feature of evidence to the Inquiry is that most explanations about differences were couched in generalities and supported by 'common sense' and 'experience' alone. Mr Zeelenberg referred to a tendency to "explain differences away" as distortion and he said this was a distinct warning sign.77 The same can be said of explanations that simply cite 'movement' without specifying the particular pattern of movement and demonstrating that the assumed pattern of movement can account for both the different distribution of characteristics in mark and print and the physical appearance of mark and print.
35.66. The evidence of Mr McGinnies pointed to the relevance of experience and indicated that trainees may be being taught what might be regarded as conventional wisdom.
35.67. The limitations of experience and conventional wisdom are discussed by the National Academy of Sciences:
"Currently, distortion and quality issues are typically based on 'common sense' explanations or on information that is passed down through oral tradition from examiner to examiner. A criticism of the latent print community is that the examiners can too easily explain a 'difference' as an 'acceptable distortion' in order to make an identification."
The National Academy of Sciences has recommended formal research in this context.81
35.68. Professor Champod agreed that there is a requirement for empirical evidence. When an expert invokes physical movement of a finger to explain a perceived difference he would prefer to see evidence that what is suggested is a possibility. If movement is invoked and there is evidence that such movement can be reproduced through controlled experiments it becomes a possible explanation.82
35.69. Mr Zeelenberg drew the attention of the Inquiry to a research paper by Alice V. Maceo entitled 'Qualitative Assessment of Skin Deformation: A Pilot Study'83 reporting on the impact of various sources of distortion (including differential pressure and torque - i.e. turning movement) on prints. The conclusion is that "'Distortion' is not a wild card to be played to dismiss unexplained regions of concern in a latent print" (page 438). Different types of distortion can affect prints in different ways and even different types of print (a whorl or a loop) can be affected differently under the same source of distortion, with the result that "some distortions may be subject to interpretation or may be unintelligible" (page 436).
35.70. One of the specific conclusions in the Maceo article is helpful to the argument that turning movement may explain Y7 and that is the finding that "Torque ... created significant deformation, affecting the position of minutiae on the periphery of the finger relative to the core" (page 435). That is balanced by the observation that "Latent prints created under torque contain visible clues that allow an analyst to determine the type of stress experienced by the skin" (page 432).
35.71. The first of those findings is relevant to Mr Swann's argument about the points at the tip of Ms McKie's left thumb shown in his chart M.84 Mr Swann accepts that those points are not in sequence with the Rosetta (chart N) or points in the lower part of the mark (chart O) but torque can produce differential movement of characteristics, particularly on the periphery.
35.72. The second point, regarding 'visible clues' of movement, is relevant to the competing arguments concerning the upper part of the mark Y7 summarised in chapter 25.85 Mr Swann's explanation of the eight points at the tip of the mark shown in his chart M assumes that a merging of ridges can have occurred without any criss-crossing and the evidence of Mr Zeelenberg, which highlights the relatively small size of the mark, invites the conclusion that such an outcome is unlikely and I have accepted that argument.86
35.73. Again there are parallels with the Mayfield case. When the mark was first identified to Mr Mayfield the FBI had explained differences in appearance between the mark and his print on the basis that the former was the product of a double touch. They did so despite the absence of any physical evidence of a double touch, such as the presence of crossovers, misalignment of ridges or protruding ridge ends. The OIG observed that the proposition that there had been a double touch leaving no physical evidence required the FBI to assume "an extraordinary set of coincidences" and that was judged to be unsatisfactory. The OIG report discussed the need for cogent explanations for differences and recommended that the FBI's Standard Operating Procedures "should be revised to explicitly require that the examiner must achieve a degree of certainty with respect to each 'explanation for differences'."87
35.74. Mr Dunbar said in evidence:
"Fingerprint experts do talk in terms of generalities and we at SCRO try to train our students that you can only talk in generalities because the next one that comes up to contradict your theory has just proven you wrong."88
Thinking in terms of generalities is to be avoided because it avoids the necessary challenge of considering whether a cogent explanation can be given for the specific differences that are observed between the mark and the print. Mr Mackenzie, Mr Berry and Mr Swann are to be commended for having taken on that challenge. The very fact that they produced chartings enabled their theories to be put to the test and has enabled me to reach an informed conclusion. Fingerprint examiners should not resort to generalities for fear of contradiction. They must be willing to test the robustness of their own assumptions and should expect that this will be a point on which they may be questioned when giving evidence in court.
35.75. In chapter 28 it is seen that SCRO examiners, working to the 16-point standard, engaged in circular reasoning in relation to otherwise unexplained differences. Sixteen points in agreement were understood to be indicative of unique identity and, therefore, because sixteen matching points were found the examiners assumed that any difference must be explicable, even though no specific reason could be advanced for the differences.
35.76. Echoes of that reasoning were heard in the evidence of some witnesses describing current practice under the non-numeric system. Mr McGinnies said it is possible to make an identification when a mark has many points in sequence and agreement even though the examiner does not have an explanation for a difference or is unsure about the explanation for a difference.89 Evidence was given to the same effect by Mr Geddes who spoke of carrying out a balancing exercise but if there was 'sufficient' volume of points in agreement he could decide that there was an identification even without an explanation for an observed difference.90
35.77. In theory Mr Chamberlain might identify a mark even though there was an unexplained difference but he could not recall having done so. Any unexplained difference would be documented in the report so that it was clear to the defence and to the court.91 The nature of the 'discrepancy'92 would be important. Some 'discrepancies' might not prevent an identification being made. Others would; for example, a bifurcation appearing in the mark that did not appear in the print.93
35.78. Other practitioners took a different view. The Metropolitan Police said there must be "no unexplained features in disagreement".94 Mrs Tierney's position was that if an examiner discovered differences that could not be easily and obviously accounted for, the volume in agreement is immaterial. She also made the broader point, consistent with Mr Grigg's approach to Y7,95 that the presence of an unexplained difference ought to cause the examiner to start again and consider whether the characteristics initially perceived to be in agreement are in agreement.96
35.79. Two separate matters require discussion at this point.
35.80. The first is the proposition that friction ridge detail is unique not just in the fingerprint as a whole but also in a small part of the fingerprint. It is the first of those propositions that underpins the ability of fingerprint examiners to identify partial impressions. It was one of the propositions that Mr Dunbar emphasised in his presentation at Tulliallan,97 sourced from the book by Wentworth and Wilder on Personal Identification, written in 1932: "there is never the slightest doubt of the impossibility of the duplication of a finger print, or even the small part of one."98
35.81. Care has to be taken when considering the possible extrapolation of that proposition to infer that, if friction ridge detail is 'sufficiently' unique in a small part of a mark, a difference elsewhere is immaterial. The context in which it occurs in Wentworth and Wilder is discussion of the broader proposition that under sufficient magnification no two natural objects are duplicates: "No two heads of clover, no two ears of corn, can be exactly alike."99 Ashbaugh incorporates a variant of the Wentworth and Wilder proposition in his third premise of friction ridge identification:100 friction ridge patterns and the details in small areas of friction ridges are unique and never repeated.101 However, consistent with his general thesis that statements that may be true in relation to the skin do not necessarily apply directly to impressions, he observes:
"when discussing friction ridge skin, it can be said that friction skin is unique in a very small area. The statement, however, only applies to a friction skin print if clarity is present. When clarity is absent it may be an incorrect statement."102
35.82. Ashbaugh proceeds to discuss the presence of any indication of distortion as a red flag necessitating careful analysis of the mark103 and stresses the significance of clarity:
"When there is less clarity, there is room for some tolerance of discrepancy. However, discrepancies must be consistent with the factors found in the substrate, matrix, development medium, deposition pressure, pressure distortion, anatomical factors. Also, due to these discrepancies each area in agreement will have less evaluative weight. Therefore, a greater volume of unique and accidental details in agreement will be required to satisfy an opinion of individualisation."104
35.83. That passage emphasises the need for an explanation for any difference consistent with a recognised source of distortion and in addition requires the mark as a whole to be assessed. It coincides with the reasoning of Mrs Tierney, who said that the presence of a difference that she could not explain would cause her to reconsider whether the characteristics that she otherwise had thought were in agreement truly were.105
35.84. This leads to consideration of the second matter which relates to the so-called 'one-discrepancy rule' and the 'non-discrepancy rule'. While both rules have a bearing on the relevance of a 'difference' there is a significant distinction between them that can be seen by recalling that fingerprint examiners can reach one of three possible conclusions at the end of a comparison: (1) identification, (2) exclusion or (3) inconclusive.
35.85. The 'non-discrepancy rule' can relate to identification and is to be seen in the formulation that a finding of identification can be reached when the examiner observes sufficient coincident characteristics in sequence and the absence of any 'dissimilar' characteristics.106 The 'one discrepancy rule' can be applied to an exclusion: "The presence of one discrepancy is sufficient to exclude."107 The two rules are sometimes conflated but must be kept separate because they can lead to distinct conclusions and, of course, the examiner is not facing an either or decision - either identification or exclusion - because of the possibility of a finding of inconclusive. A 'difference' that precludes an identification does not necessarily warrant a finding of exclusion because it may be inconclusive either way.
35.86. There is a preliminary question common to both 'rules' and that relates to the meaning of 'discrepancy'. This is best seen in the SWGFAST 'Standards for Conclusions' which qualifies the 'one discrepancy rule' by this statement: "Distortion is not a discrepancy and is not a basis for exclusion."108 The inability to set any parameters on the extent to which 'distortion' can produce 'differences' means that the practical application of both rules is uncertain. Professor Champod's comment was that "a one-unexplained-discrepancy rule is not a standard. In fact this rule is ill-defined by a circular argument109... The whole concept of sufficiency to exclude is left to the skilled judgment of the examiner."110
35.87. In this report I am primarily concerned with two marks that were identified and the fourth question that is presently being discussed (Can the examiner ascertain a reliable explanation for the characteristics that differ?) relates to the circumstances in which a conclusion of identification can be justified when some 'difference' is present. Support can be found in the literature for the circular argument that if there are sufficient characteristics in agreement then any differences must be explicable. Mr Swann111 cited Wendell W. Clements on The Study of Latent Fingerprints. Clements, in effect, inverts the reasoning of Mrs Tierney, that the presence of a difference that she could not explain would cause her to reconsider the characteristics she had thought to be in agreement. Clements' position is that if sufficient characteristics in agreement are present it is the dissimilarities that require to be reconsidered. The discussion in that text is apropos the question whether a fingerprint examiner can be convinced of an identification if there are 12 matching characteristics and three 'dissimilarities', to which the answer is: "Yes, because the 3 dissimilarities would have to be explainable." The explanation given is in these terms: "... it is not possible to have a print with 12 points of similarity and 3 unexplainable dissimilarities. It has been determined by the Los Angeles Police Department, and many police agencies, that 10 matching ridge characteristics are ample to prove a positive identification¿"112
35.88. The contention that as few as ten 'matching' characteristics can provide proof of identity, even with some 'dissimilarities' being present, provides a direct link to the Brandon Mayfield case, which post-dates the Clements text. An error was made by the FBI for the overlapping reasons that (a) they failed to give due weight to the lack of quality in ten characteristics that they believed were matching and (b) they failed to give a satisfactory explanation for certain differences. Applying the reasoning of Mrs Tierney, the two failures were inter-related.
35.89. The official response to the Mayfield case has led to a reformulation of both the 'one discrepancy rule' as applying to exclusions and the 'non-discrepancy rule' as applying to identifications. The rationale for the reforms can be considered in light of a practical example. Take the case of a fingerprint found on a knife used in a murder that it is suspected belongs to the accused. A fingerprint examiner observes a difference between the print on the knife (i.e. the mark) and the accused's print. If the examiner attributes that difference to 'distortion' the examiner might conclude that there is an identification, which would be a highly incriminating finding. Alternatively, if the examiner were to conclude that it is not due to distortion, and therefore is a 'discrepancy', he could make a finding of exclusion which could be exculpatory. That suggests that the examiner must be equally careful before arriving at either positive conclusion. The examiner should not make an identification without good cause to regard the 'difference' as the product of 'distortion'. Equally, the examiner should not, without good cause, conclude that the 'difference' cannot be attributable to 'distortion' and hence make a finding of exclusion. If there is reason for doubt, the appropriate finding ought to be neutral, i.e. inconclusive.
35.90. The 2011 OIG report discloses that, in relation to findings of exclusion, the FBI's standard operating procedures have moved away from the proposition that one discrepancy is sufficient for an exclusion and replaced it with the criterion that exclusion occurs where "there are sufficient friction ridge details in disagreement to conclude that two friction ridge prints did not originate from the same source." It should not be inferred that the FBI would make a finding of identification despite some unexplained difference. On the contrary, the FBI has moved away from the one discrepancy rule to avoid erroneous exclusions.113
35.91. Turning to identification, the OIG believed that "accepting plausible or reasonable explanations supported by mixed evidence was inconsistent with the certainty claimed for identifications" and the FBI's standard operating procedures were also modified in relation to that finding, the revised statement being: "An examiner must be confident that any apparent difference between two prints is due to distortion, and not an actual difference in friction ridge detail. This level of confidence must be consistent with the degree of confidence an examiner must have in order to render an identification decision."114 Even where an examiner sees many similarities between mark and print, if the examiner is unable to determine with adequate confidence whether one or more differences are "distortions (consistent with identification) or discrepancies (consistent with exclusion)" the examiner will not make an identification.115 There is a separate question whether the FBI would classify such an indeterminate comparison as either 'not of value for identification'116 or 'inconclusive'117 but for present purposes the critical fact is that they would not make a finding of identification in such circumstances.
35.92. Writing in 2000, INTERPOL'S European Expert Group on Fingerprint Identification observed:
"With identifications proven to be mistaken, it became clear that the involved experts have ignored the differences. Evaluation of those comparisons often contain a long list of excuses why the print does not look like how it should, disguised as demonstration of the skill and experience of the expert¿ The rule is therefore that: 'Tolerances should not vary dependent on the quality of the impression'."118
35.93. That is consistent with not only the OIG's findings in relation to the Mayfield case but also my findings in the present Inquiry.
Comment on the approach to differences
35.94. Prior to the Mayfield case it could be argued that the question whether an identification could be made despite an unexplained difference was more apparent than real because explanations for differences could be tenuous generalisations based on no more than 'experience'.
35.95. The 'Rosetta' in Y7 may be taken as an example.119 At a level of generality some examiners argued that it was 'explained' on the basis that the mark was the product of either (a) more than a single touch or (b) some other movement. At that level it could be argued that Y7 was not a case where a mark was identified despite some unexplained difference.
35.96. In light of my findings and the Mayfield case it can now be concluded:
35.97. To take Y7 as an example, if the assumed source of distortion is that the mark is the product of multiple touches, how many touches are required to explain the whole constellation of points in the mark? Is it likely that the assumed sequence of movements will have produced the distribution of characteristics observed in the mark as highlighted in Figure 13 in chapter 25; or does this hypothesis require the examiners "to believe a remarkable set of coincidences"?120 I rejected the explanation of the distribution of characteristics in Y7, including the Rosetta, as a product of multiple touches or some other movement because it lacked cogency.
35.98. Advancing a test of 'cogency' demands additional rigour at the evaluation stage of a comparison, and during the technical review recommended for complex marks121 but the fact remains that in the absence of any objective standards 'cogency' is itself a subjective criterion. There is a need for further research to inform decision making.
Question 5: Can the examiner find sufficient matching characteristics to justify the inference that the mark is uniquely identifiable as having been made by a specific person?
35.99. Under the 16-point approach the general position was that examiners had to find sixteen characteristics in sequence and agreement, with no unexplained differences, in order to make an identification.122 Since the introduction of the non-numeric approach or 'standard' in England and Wales in 2001 and Scotland in 2006 there has been no set threshold as to the number of characteristics required to declare an identification. There is no objective test of 'sufficiency' for a conclusion that mark and print match and the judgment is subjective. As Mr Zeelenberg put it, sufficiency is left to the discretion of the examiner.123 No individual examiner who gave evidence to the Inquiry gave a definitive reply to the question what was his or her personal minimum threshold; it would depend on the examiner's assessment of the particular pattern in the mark.124
35.100. Examiners are influenced by the perceived 'rarity' of particular combinations of level 2 detail. As discussed in chapter 41 some work is being undertaken to establish the statistical incidence of combinations of ridge detail but the current practice of fingerprint examiners is little informed by statistics. Mr Geddes summed it up: "I can only go by my own experience. I can only go by the empirical experience of the fingerprint community as a whole."125 This is a further example of opinion depending on personal experience informed by conventional wisdom and not scientific experiment.
The use and reliability of third level detail
35.101. In discussing the five questions the focus has been primarily on traditional level 2 features, ridge endings and bifurcations. With the change to the non-numeric system consideration has now also to be given to third level detail. Third level detail can be described briefly as "extremely tiny variations in the ridges themselves, such as the shape of ridge edges, the width of the ridges, and the shape and relative location of pores along the ridges."126 The fact that third level detail can be 'extremely tiny' places considerable emphasis on the clarity of the impression.
35.102. Some of the evidence to the Inquiry made use of third level detail. For example, it was the inclusion of third level detail that enabled Mr Mackenzie to augment his tally of matching points in Y7 to a total of 45 in his Tulliallan presentation.127 Although the use of first and second level detail in the identification of fingerprints is long-standing, the conscious use of third level detail as a factor in decision making is a more recent development. Mr MacPherson observed that the relevance of third level detail (particularly pores) can be traced back to the work of Locard in 1912128 but Ashbaugh explains that while numerical standards prevailed third level detail was not to the fore because identification depended on the number of level two characteristics in agreement.129 Under the non-numeric approach comparison is based on the aggregate of all three levels of detail.130
35.103. Mr Kent, who had sat on a 'third level detail' working group for over a year, said that it proved difficult even to define what third level detail is, let alone quantify it.131 Professor Champod noted that there was no single definition of third level detail. Examiners classify things as second level which others would classify as third level and vice-versa.132
35.104. Some examiners spoke of the possibility of identifying marks relying on third level detail without any second level detail. Mr McGinnies said that trainees are made aware that identification of a mark that has no second level detail is theoretically possible.133 Mr Wertheim referred to a fingerprint published in the Journal of Forensic Identification where the examiner used only first and third level detail. All of the third level detail was clear and Mr Wertheim agreed with the identification. He had never seen such a mark in his own casework.134
35.105. Mr Zeelenberg's approach differed. He observed that only good quality prints can show third level detail. Third level detail cannot stand on its own but rather has to be checked in correlation with second level detail.135 In his view an examiner can use third level detail as a factor in assessing confidence in relation to the second level patterns that are observed and may therefore be a factor that leads to an identification being made on ten points rather than twelve.136
35.106. Professor Champod referred to a survey in which examiners were asked about the strength of a given set of third level detail. The judgment as to the strength varied significantly from one examiner to the other.137 Professor Champod explained that a good starting point to assess the strength of third level detail is to explore how it is reproduced from one impression to the other as reproducibility is one of the key components in robustness. He stated: "we conducted experiments as to the reproducibility of these features from mark to mark, we noted that the only features which have a good level of reproducibility - by this term I mean that you can expect to find these features in correspondence should a corresponding print be available....are the relative pore positions and the specific shape of minutiae. The form of the ridge edges, the specific forms of the pores are not well reproduced from one impression to the other."138
35.107. Professor Champod's opinion is that the lack of reproducibility gives a good indication that such features are not reliable in the identification process and that third level detail is quite limited in terms of weight. Accordingly, his opinion is that it is not good practice to rely solely on third level detail to individualise.139
35.108. The OIG report expressed concern at the reproducibility of third level detail and the risks of 'cherry picking' helpful level 3 detail. The OIG recommended that standard operating procedures (SOPs) should be refined to: "define the circumstances under which the clarity of a latent fingerprint is sufficient to support the utilization of level 3 details to support an identification. The SOPs should also require that the examiner consult all versions of the available known prints of the subject to determine whether any level 3 details utilised are reliably and repeatedly reproduced. The SOPs should require that the examiner apply 'fair reasoning' in utilising level 3 details that support the identification so as to avoid the selective use of supporting level 3 details."140
35.109. The review published in June 2011 by the OIG reports on ongoing research which indicates that the appearance of level 3 detail is not consistently reproduced in different friction ridge impressions and can vary depending on the method of capture, pressure applied and the resulting image quality.141 Maceo's experiments showed, for example, that the prominence of incipient ridges was affected by the degree of pressure applied during deposition of the print.142 FBI Laboratory training manuals have been revised to emphasise that level 3 detail should only be used when the mark is very clear and where there has been similar deposition pressure in mark and print.143 In practice FBI examiners are now extremely cautious in relying on level 3 detail and would not let it be decisive if there was insufficient level 2 detail.144
35.110. The evidence to the Inquiry is consistent with that review.
The risk of contextual bias
35.111. Subjectivity may be influenced by more than personal traits. Dr Dror and Mr Charlton suggest that contextual bias or circumstantial knowledge can have an effect on decision making by fingerprint examiners.145 Some of the work carried out at the analysis stage of the ACE-V procedure requires contextual information but there is a limit to the information that is necessary.
35.112. The urgency of a situation may create bias, and urgency is a particular feature of criminal investigations. As the National Academy of Science notes: "Another potential bias is illustrated by the erroneous fingerprint identification of Brandon Mayfield as someone involved with the Madrid train bombing in 2004. The FBI investigation determined that once the fingerprint examiner had declared a match, both he and other examiners who were aware of this finding were influenced by the urgency of the investigation to affirm repeatedly this erroneous decision."146 The possibility of this type of bias is readily apparent in the context of a criminal investigation. A request to compare marks against a limited set of prints may make perfect sense from the point of view of efficiency; but such a request creates the risk of such bias.
35.113. A further way in which bias can arise is by virtue of the way the question for the examiner is framed. The National Academy of Sciences notes: "A common cognitive bias is the tendency for conclusions to be affected by how a question is framed or how data are presented...A series of studies has shown that judges can be subject to errors in judgment resulting from similar cognitive biases. Forensic scientists also can be affected by this cognitive bias if, for example, they are asked to compare two particular hairs, shoeprints, fingerprints - one from the crime scene and one from a suspect - rather than comparing the crime scene exemplar with a pool of counterparts."147 A risk of this type in the context of fingerprint examination arises when an examiner is provided with only one mark and print to compare, in the knowledge that the mark has been identified, which is the normal approach to verification.
35.114. The research is not all in one direction. The Metropolitan Police has carried out work on the emotional context to fingerprint work. The research involved presenting a mark said to be from a murder scene to create a high emotional context and another from a forgery scene to create a low emotional context. The mark was from a known source, was of poor quality and was at the boundaries of practitioner interpretation. Seventy experts volunteered. Thirty five were given the high emotional context mark and thirty five the low emotional context mark. Personality and demographic data was recorded. There was no significant difference between the decisions in either emotional context and no relationship between the emotional context and the decisions. Analysis showed there was no difference between gender, years of experience and age.148
35.115. The National Academy of Sciences refers to the fact that research "has been sparse on the important topic of cognitive bias in forensic science - both regarding their effects and methods for minimizing them."149 It notes that "forensic science disciplines are just beginning to become aware of contextual bias and the dangers it poses. The traps created by such biases can be very subtle, and typically one is not aware that his or her judgment is being affected."150
35.116. The training programmes for fingerprint personnel at local and national level incorporate elements designed to develop and test an individual's ability to remain professionally objective when undertaking fingerprint examinations. A fundamental aspect of the training programmes Mrs Tierney has undertaken and implemented is the encouragement of personnel to challenge constantly their own conclusions, the conclusions of their colleagues and also to be open and receptive to challenges from others.151
35.117. The results of research into contextual bias are divided, and, as the National Academy of Sciences noted, research has not been extensive. That some studies at least point to a risk of forensic scientific opinion being influenced by contextual bias, and the FBI finding that the error in the Mayfield case was contributed to by the context of urgency provide a basis for concern that contextual bias may give rise to error in fingerprint identification. Unless the provision of contextual information is absolutely necessary, it should be avoided.
Numeric versus non-numeric approach
35.118. There remains some debate as to whether the non-numeric approach is appropriate. Not least this is because it could be argued that a numeric standard introduces objectivity at least in relation to the fifth question above: can the examiner find sufficient matching characteristics to justify the inference that the mark is uniquely identifiable as having been made by a specific person?
35.119. Mr Chamberlain said that the majority of the members of the EU apply a numerical standard, generally around 12.152
35.120. Mr Kent believed that a standard is required for identification.153 As not enough is known about the statistical distribution of characteristics and the probability of fortuitous or adventitious matches, it is undesirable to have a situation where experts do not have clear guidelines. In particular, he argued that it is unrealistic to expect an expert to know, intuitively, how many other people in the population have a particular subset of characteristics.154 Significantly, though, he added that a numeric standard is not enough. One must consider image quality and quality control of the processes followed in fingerprint comparison work.155
35.121. Mr Zeelenberg was supportive of a numeric approach.156 A numeric standard is applied in the Netherlands varying from ten to twelve depending on the clarity of the mark.157 Mr Zeelenberg explained that the 12-point rule is based on the gut feeling of early fingerprint experts including Locard and Galton leading to the view that reliability started somewhere around 12 and that a drop-off in reliability occurred quickly below ten.158 He pointed to recent statistical surveys which confirmed this, for example the findings of Pankanti, Prabhakar and Jain.159 They found that if an examiner was wrong with one or two points below 12 there was a significant drop in reliability and that reliability dropped down dramatically below ten.160
35.122. Other witnesses were not supportive of a numeric approach. Mr Nelson thought that a problem with the 16-point standard was that whenever someone found 16 points they believed they could almost create the impression that it was a 'gold standard' and, therefore, could not be challenged. A numeric standard leads to the belief that one can stop when the relevant number is reached and that it is infallible.161
35.123. Mr Pugh said that the 16-point standard led to the "hunting out" of points and was thought to be depriving the courts of useful evidence.162 An examiner could still be 100% certain with less than 16 points163 but if the 16-point standard applied an examiner could not report it, even if he was confident that a mark could be individualised. The non-numeric approach allows a greater number of identifications to be reported.
35.124. Professor Champod's opinion is that there is no scientific justification for a numeric standard.164 A strict numeric approach gives each point the same weight when it is known that weight may vary depending on the type and position on the fingerprint.165 He explained that the prevalence of different types of level 1 features in the population is well documented.166 Statistical data in relation to level 2 features had been gathered through various research initiatives, and he provided the Inquiry with results from one study, which indicated that some characteristics are much rarer than others, conditional on location on the fingerprint pattern.167 He argued that a rule which gives equal weight to each point is an inadequate model from a statistical perspective.168 In addition, he argued that it is restrictive to consider only level 2 detail because level 3 features such as pore position can be taken into account in forming a view.
35.125. A return to a numeric standard would introduce the appearance of objectivity only in relation to the fifth question, with the conclusion remaining overwhelmingly subjective given the first four questions. What is more, it is open to doubt whether a numeric approach would truly add objectivity even in relation to the fifth question given the temptation for examiners to 'hunt out' or 'tease out' points to reach the required number. A numeric standard afforded no guarantee of accuracy; QI2 and Y7 were identified to a 16-point standard.
35.126. There is no merit in returning to a numeric standard. Rather there needs to be greater appreciation, both among fingerprint practitioners and those involved in the criminal justice system, that a conclusion on identity is a subjective opinion reflecting the variables discussed in this chapter.
The limitations of fingerprint methodology
35.127. The issues discussed in this chapter transcend narrow questions about the competence or performance of individual practitioners. Recognising that fingerprint identification is a matter of opinion, not fact, inevitably leads to the need to identify the variables that are relevant to that opinion but more fundamentally ought to focus attention on the limits of the technique itself.
35.128. Returning to the need to maintain perspective,169 there is no reason to suggest that fingerprint comparison is an inherently unreliable form of identification but practitioners and fact-finders alike require to give due consideration to the limits of the discipline that flow from the significance of sources of variability and the absence of objective standards. '100% certainty'170 is not a legal requirement for the admissibility of evidence and fact-finders (judge or jury) can readily accept sources of evidence that are known to suffer from limitations (eyewitness evidence being a classic example) and also subjective conclusions. Acknowledging that fingerprint evidence is not '100% certain' need not detract from the true value of this type of evidence but, in any event, what matters is that fact-finders be given a fair presentation of the evidence, its strengths and weaknesses, so that they can determine its weight in any particular case.
35.129. The appropriate manner in which fingerprint examiners should express their opinions is discussed more fully in chapter 38 where it is observed that practitioners who have been accustomed to thinking in terms of '100% certainty' may not readily adjust to 'conditions of uncertainty'171 and the need to explain their findings in terms of subjective judgments. Since this entails an adjustment to the way that fingerprint examiners practise, the implications reach back to the training, as is discussed in chapter 40. Adjustment begins with recognition that there is a need for more research to explore the sources of variability and to obtain a greater appreciation of the limits of the methodology of fingerprint identification. Meantime, an emphasis needs to be placed on the importance of trainees and qualified examiners not only learning or applying the methodology of fingerprint work, but also being conscious of its limitations and to that end there is a need to engage with members of the academic community working in the field. That engagement will assist fingerprint examiners to be aware of any research of relevance to their work. It may also assist them to be able to articulate their reasoning in support of the subjective judgments that they have made in arriving at their conclusions.
Introduction to recommendations
35.130. The principal recommendations that derive from the discussion in this chapter are the need for fingerprint practitioners and the legal community to acknowledge and adjust to fingerprint evidence as opinion evidence. The full ramifications of that adjustment cannot be foreseen because one of the main consequences is recognition of the need for further research in part to fathom the limits of the technique as currently practised and also to suggest improvements.
35.131. Some immediate improvements to practice can be recommended. Firstly, the potential for contextual bias has to be addressed. Secondly, the methodology has to be strengthened. In carrying out analysis fingerprint examiners require to consider whether features in the mark are observable to a lay person with normal eye sight. In the analysis and comparison stages of the ACE-V procedure172 examiners must think beyond 'events' and give due consideration to the specific type of observed characteristics, especially where more than one type of characteristic cannot be specified. In particular, they must give due consideration to the quality of the similarity between characteristics in corresponding areas of mark and print. In carrying out their evaluation they should pay particular regard to any lack of correspondence between characteristics and the degree of tolerance that would have to be applied to achieve a match. They have to give proper consideration to the degree of tolerance that they are applying in that regard and more generally in the comparison. They should not identify if there is an unexplained difference between mark and print and must carefully evaluate the cogency of any explanations for differences. While a conclusion of identity can be arrived at where there is a lack of precise correspondence in ridge detail in mark and print which is within the proper limits of tolerance or is otherwise capable of cogent explanation, examiners should consider whether the clarity of the mark is sufficient to support a confident conclusion of identity or exclusion, as the case may be. Care should also be taken when relying on third level detail and practitioners should pay close attention to research on the reproducibility of such detail.
The subjective nature of fingerprint evidence
35.132. Fingerprint evidence should be recognised as opinion evidence, not fact, and those involved in the criminal justice system need to assess it as such on its merits.
35.133. Examiners should receive training which emphasises that their findings (as to whether the mark is fragmentary and insufficient or of comparable quality; and, if of comparable quality, whether a comparison with a particular print justifies a conclusion of identification or exclusion or is inconclusive) are based on their personal opinion; and that this opinion is influenced by the quality of the materials that are examined, their ability to observe detail in mark and print reliably; the subjective interpretation of observed characteristics, the cogency of explanations for any differences and the subjective view of 'sufficiency'.
Research and development
35.134. Research should be undertaken into:
(i) distortion and the effect of movement;
(ii) the weight to be given to third level detail, and as to its reliability;
(iii) the frequency of particular characteristics or combinations of characteristics in fingerprints;
(iv) the specific factors that may cause variations among examiners; and
(v) contextual bias.
Engaging with the academic community
35.135. An emphasis needs to be placed on the importance not only of learning and practising the methodology of fingerprint work, but also of engaging with members of the academic community working in the field.
35.136. Fingerprint examiners need to be provided with training to enable them to articulate their reasoning. The SPSA, in conjunction with members of the academic community as appropriate, should determine how best to explain the process of reasoning in arriving at a non-numeric conclusion.
35.137. The SPSA should review its procedures to reduce the risk of contextual bias.
35.138. The SPSA should ensure that examiners are trained to be conscious of the risk of contextual bias.
35.139. The SPSA should consider what limited information is required from the police or other sources for fingerprint examiners to carry out their work, only such information should be provided to examiners, and the information provided should be recorded.
35.140. Features on which examiners rely should be demonstrable to a lay person with normal eye sight as observable in the mark.
35.141. In comparing a mark and print fingerprint practitioners should pay close attention to the precise type of the characteristics and carefully evaluate differences in the type of characteristic.
35.142. The SPSA, in conjunction with members of the academic community as appropriate, should design a practical system for examiners to assess and evaluate (a) tolerances and (b) any reverse reasoning.
35.143. Explanations for any differences between mark and print require to be cogent if a finding of identification is to be made.
35.144. A finding of identification should not be made if there is an unexplained difference between mark and print.
35.145. Examiners should consider whether the clarity of the mark is sufficient to support a confident conclusion of identity or exclusion.
35.146. Care should be taken when relying on third level detail in arriving at a finding and practitioners should pay close attention to research on the reproducibility of such detail.
35.147. Where third level detail is relied upon in making a comparison this should be included in any note of the examination.
1. Mr Rennison 8 July pages 80-81
2. Champod C. and Chamberlain P. Fingerprints, in: Fraser J. and Williams R. (eds) Handbook of Forensic Science, Willan Publishing, 2009, page 72
3. See chapter 2 paras 10 and 30-33
4. Ashbaugh D. Quantitative-Qualitative Friction Ridge Analysis: An Introduction to Basic and Advanced Ridgeology. Boca Raton, Florida: CRC Press, 1999, page 93 and Mr Kent 7 July pages 62-64
5. See chapter 2 paras 30-33
6. Committee on Identifying the Needs of the Forensic Sciences Community, Committee on Science, Technology and Law Policy and Global Affairs, Committee on Applied and Theoretical Statistics Division on Engineering and Physical Sciences, National Research Council. Strengthening Forensic Science in the United States: A Path Forward, Washington, D.C.: National Academies Press, 2009, page 43 (and also pages 143-144). Also Professor Champod 25 November pages 40-41.
7. MP_0008 section 5. This research was not peer reviewed as at November 2009 - Mr Pugh, Miss Hall 24 November page 56ff at page 63.
8. Miss Hall 24 November page 61
9. MP_0008 para 5.4
10. Miss Hall 24 November page 62
11. Miss Hall 24 November page 63
12. Professor Champod 25 November pages 71-72
13. Now the Home Office Scientific Development Branch
14. FI_0052 para 47ff Inquiry Witness Statement of Mr Kent and HO_0032
15. FI_0052 para 57 Inquiry Witness Statement of Mr Kent
16. FI_0052 para 58 Inquiry Witness Statement of Mr Kent
17. FI_0031 paras 103, 106 Inquiry Witness Statement of Mr Geddes
18. CO_0059 pdf pages 14-15 (the characteristics numbered in red)
19. FI_0089 para 36 Inquiry Witness Statement of Mr Graham
20. Mr Graham 9 July pages 59-60
22. Mr Swann 21 October pages 51-52
23. CO_1065 para 4.3
24. CO_2003 para 2.7
25. See chapter 27 para 44ff
26. Edwards, Judge H.T. Solving the Problems That Plague the Forensic Science Community - The National Academy of Sciences, 3 April 2009, Arizona, US page 2. URL: http://lst.law.asu.edu/FS09/pdfs/H.T.%20Edwards,%20Solving%20the%20Problems%20That%20Plague%20Forensic%20Science.pdf
27. INTERPOL European Expert Group on Fingerprint Identification - IEEGFI II (2004) Method for Fingerprint Identification Part II, URL: http://www.interpol.int/Public/Forensic/fingerprints/WorkingParties/IEEGFI2/default.asp
28. SWGFAST is sponsored by the US Institute of Justice and the FBI. It aims to establish consensus guidelines and standards for the forensic examination of fingerprints, palm prints and foot prints. Members include up to fifty representatives from international, federal, state, and local forensic science laboratories, as well as from academia and private practice.
29. Scientific Working Group on Friction Ridge Analysis, Study and Technology (2001) Standard Terminology of Friction Ridge Examination, (SWGFAST). URL: http://www.swgfast.org/documents/terminology/110323_Standard-Terminology_3.0.pdf
30. US Department of Justice. Office of the Inspector General (2011) A Review of the FBI's Progress in Responding to the Recommendations in the Office of the Inspector General Report on the Fingerprint Misidentification in the Brandon Mayfield Case, URL: http://www.latent-prints.com/images/FBI%20Mayfield%20Progress%20062011.pdf page 11
31. INTERPOL (2004) Method for Fingerprint Identification Part II para 8.3.1
32. See chapter 19
34. See chapter 27
35. See chapter 27 para 3ff
36. See chapter 27 para 39ff
37. See chapter 29
38. See chapter 19
39. See chapter 27
40. See chapter 25 para 45
41. See chapter 12 para 48
42. FI_0149 para 3.2 (page 30) Inquiry Witness Statement of Mr Swann
44. See chapter 25
45. See chapter 26
46. MP_0008 section 5
47. See para 4
48. Professor Champod 25 November pages 48ff, 104ff
49. See also the New Zealand Court of Appeal in R v Buisson  2 NZLR 542 at pages 549-550
50. FI_0167A SCRO Phase 1 Comparative Exercise Enlargement of Y7
51. See chapter 25 para 84
52. Mr Zeelenberg 7 October page 44
53. See chapter 25 para 85
54. See chapter 25 paras 45ff and 88
55. INTERPOL (2004) Method for Fingerprint Identification Part II, para 7.8.3
56. See chapter 33 para 56ff
57. US Department of Justice, Office of the Inspector General (2006) Review of the FBI's Handling of the Brandon Mayfield Case (Unclassified and Redacted) (US Department of Justice) URL: http://www.justice.gov/oig/special/s0601/final.pdf Table 2 (pdf page 182) and chapter 1 (pdf page 318)
58. OIG (2006) pdf pages 181-183 at 183
59. FI_0166A SCRO Phase 1 Comparative Exercise Enlargement of QI2
61. See chapter 39
62. Ashbaugh D. Quantitative-Qualitative Friction Ridge Analysis, 1999, page 93
63. Mr McGinnies 4 November page 66
64. Mr McGinnies 4 November page 80
65. Professor Champod 25 November page 113
66. Professor Champod 25 November pages 45-53
67. Ashbaugh D. Quantitative-Qualitative Friction Ridge Analysis, 1999, page 94
68. Professor Champod 25 November pages 109-111
69. Full passage quoted in chapter 28 para 48
70. See para 92 below; INTERPOL (2000) Method for Fingerprint Identification
71. Mr McGinnies 4 November from page 65
72. Mr McGinnies 4 November page 67
73. Professor Champod 25 November page 110
74. Professor Champod 25 November page 51
75. Professor Champod 25 November pages 110-112
76. See chapter 39 para 11ff
77. Mr Zeelenberg 7 October pages 16-19
78. Mr McGinnies 4 November page 69
79. Mr McGinnies 4 November pages 71-72
80. Mr McGinnies 4 November page 74
81. NAS, Strengthening Forensic Science in the United States: A Path Forward, 2009, page 145
82. Professor Champod 25 November page 145
83. Maceo A. V. Qualitative Assessment of Skin Deformation: A Pilot Study. Journal of Forensic Identification, 2009; 59 (4); 390
85. Paras 227 and 246-247
86. See chapter 25 para 276
87. OIG (2006) pdf pages 164-177 and 210-217
88. Mr Dunbar 6 October page 67
89. Mr McGinnies 4 November pages 64-65
90. Mr Geddes 1 July pages 19-22
91. Mr Chamberlain 18 November pages 28-29, 86
92. The term used by Mr Chamberlain: see page 91
93. Mr Chamberlain 18 November pages 91-93
94. MP_0008 page 22
95. See chapter 25
96. Mrs Tierney 12 November pages 68-70
97. CO_0050 pdf page 2
98. Wilder, H. H. and Wentworth, B. (1932) Personal Identification - Methods for the Identification of Individuals Living or Dead (2nd edition) Chicago: The Fingerprint Publishing Association, page 325
99. Wilder, H. H. and Wentworth, B. (1932) Personal Identification - Methods for the Identification of Individuals Living or Dead, page 321
100. In this Report this proposition is described as the second premise: see chapter 2 para 9.
101. Ashbaugh D. Quantitative-Qualitative Friction Ridge Analysis, 1999, pages 91-92
102. Ashbaugh D. Quantitative-Qualitative Friction Ridge Analysis, 1999, page 93
103. Ashbaugh D. Quantitative-Qualitative Friction Ridge Analysis, 1999, page 127
104. Ashbaugh D. Quantitative-Qualitative Friction Ridge Analysis, 1999, page 131
105. Mrs Tierney 12 November pages 69-70
106. Mr Zeelenberg 7 October pages 18-19
107. Scientific Working Group on Friction Ridge Analysis, Study and Terminology (2003) Standards for Conclusions, (SWGFAST), URL:
108. SWGFAST (2003) Standards for Conclusions, para 2.2.2.
109. Professor Champod here refers to Thornton J.I. The One-Dissimilarity Doctrine in Fingerprint Identification. International Criminal Police Review, 1977; vol 32, pages 89-95
110. ED_0003 para 30
111. Mr Swann 22 October pages 45-47
112. Wendell W. Clements on The Study of Latent Fingerprints, Charles C. Thomas Publisher, 1987 page 108ff at page 113 and see also 124
113. OIG (2011) pdf pages 11-12
114. OIG (2011) pdf pages 30-32 and see also 37-40
115. OIG (2011) pdf pages 11-12 and 30-32
116. Perhaps akin to the Scottish categorisation as 'fragmentary and insufficient'.
117. OIG (2011) pdf pages 37-40
118. INTERPOL (2000) Method for Fingerprint identification; Definitions, para 12
119. See chapter 25 figure 12
120. OIG (2006) pdf pages 175-177 at pdf page 176
121. See chapter 39
122. See chapter 32
123. Mr Zeelenberg 8 October page 18
124. See discussion at paragraphs 44-48 of chapter 33
125. Mr Geddes 26 June pages 27-30 at page 29
126. OIG (2011) pdf page 6
128. Mr MacPherson 27 October page 137; see also Ashbaugh D. Quantitative-Qualitative Friction Ridge Analysis: An Introduction to Basic and Advanced Ridgeology, 1999, Chapter V
129. Ashbaugh D. Quantitative-Qualitative Friction Ridge Analysis, 1999, pages 2-4
130. Ashbaugh D. Quantitative-Qualitative Friction Ridge Analysis, 1999, pages 93-94
131. FI_0052 para 56 Inquiry Witness Statement of Mr Kent
132. Professor Champod 25 November page 95
133. FI_0193 para 155 Inquiry Witness Statement of Mr McGinnies
134. Mr Wertheim 22 September pages 33-36
135. Mr Zeelenberg 7 October page 53
136. Mr Zeelenberg 8 October pages 44-45
137. Professor Champod 25 November page 96
138. Professor Champod 25 November page 97
139. Professor Champod 25 November pages 97, 99-100
140. OIG (2006) pdf page 210
141. OIG (2011) pdf page 25
142. Maceo A. V. Qualitative Assessment of Skin Deformation: A Pilot Study. Journal of Forensic Identification, 2009; 59 (4); page 410
143. OIG (2011) pdf page 25
144. OIG (2011) pdf pages 33-34
145. Dror I,E. and Charlton D. Why Experts Make Errors. Journal of Forensic Identification, 2006; 56: 600; I. E. Dror and others, Contextual information renders experts vulnerable to making erroneous identifications. Forensic Science International 156 (2006) 74-78; but see also Hall, Player, Will the introduction of an emotional context affect fingerprint analysis and decision-making? Forensic Science International 181 (2008) 36-39
146. NAS, Strengthening Forensic Science in the United States: A Path Forward, 2009, page 123
147. NAS, Strengthening Forensic Science in the United States: A Path Forward, 2009, pages 122-123
148. MP_0008 paras 5.5-5.7
149. NAS, Strengthening Forensic Science in the United States: A Path Forward, 2009, page 124
150. NAS, Strengthening Forensic Science in the United States: A Path Forward, 2009, page 185
151. FI_0152 paras 112-113 Inquiry Witness Statement of Mrs Tierney
152. Mr Chamberlain 18 November page 4
153. FI_0052 para 61 Inquiry Witness Statement of Mr Kent
154. Mr Kent 7 July pages 52-53
155. Mr Kent 7 July pages 58-59
156. Mr Zeelenberg 8 October page 17ff
157. FI_0115 paras 127-135 Inquiry Witness Statement of Mr Zeelenberg
158. See also Professor Champod 25 November page 67ff for a summary of Locard's three rules.
159. Pankanti S., Prabhakar S. and Jain A. On the Individuality of Fingerprints. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002; 24(8)
160. Mr Zeelenberg 8 October pages 23-24
161. Mr Nelson 13 November pages 6-8
162. Mr Pugh 24 November page 23
163. Miss Hall 24 November page 24
164. Professor Champod 25 November page 70-71. Re studies on the variability of minutiae see also EC_0001 (the literature review prepared for the Inquiry) page 2.
165. Professor Champod 25 November page 70
166. Arches, loops, whorls etc - ED_0003 para 50-54
167. ED_0003 para 58
168. ED_0003 para 59
169. See chapter 34
170. See chapter 28 para 26ff
171. NAS, Strengthening Forensic Science in the United States: A Path Forward, 2009, page 217; and see Chapter 38 para 41ff
172. See chapter 36