page   1
 1                                  Wednesday, 25th November 2009
 2   (10.05 am)
 3              PROFESSOR CHRISTOPHE CHAMPOD, affirmed
 4   THE CHAIRMAN:  It is Professor Christophe Champod.
 5   A.  Yes.
 6   THE CHAIRMAN:  Just so we get it on the record.
 7                     Examined by MR MOYNIHAN
 8   Q.  Good morning, Professor.
 9   A.  Good morning.
10   Q.  Before we come to the main part of your evidence, which
11       is a presentation that you have prepared for us, I just
12       want to go over some of the history of our involvement
13       with you so that it is available to the public because
14       some questions have been asked about us in relation to
15       our working with you and a number of other individuals.
16           First of all, if I begin with the fact that we will
17       end this Inquiry really as I began and that is with a
18       conversation with you just a little over a year ago?
19   A.  Absolutely.
20   Q.  In relation to that, if I just look at things before
21       your involvement with us some questions have been raised
22       about whether you personally had worked with the mark
23       Y7.  Had you before your involvement with the Inquiry
24       done any work in relation to the fingerprint Y7?
25   A.  No, I had seen the marks published on the Internet but I
page   2
 1       haven't done the examination on these.
 2   Q.  When first approached by the Inquiry team you did
 3       confirm that you are not a fingerprint practitioner?
 4   A.  I'm not a Fingerprint Expert in the sense that I do not
 5       practice fingerprint comparison on a daily basis.
 6   Q.  But you do have a function in relation, first of all, to
 7       research?
 8   A.  Yes.
 9   Q.  Do you also teach fingerprint practitioners?
10   A.  I do, both for our university students and also for
11       fingerprint practitioners in Switzerland and elsewhere.
12   Q.  When you say "elsewhere", have you a fairly extensive
13       contact with fingerprint practitioners throughout the
14       world?
15   A.  I have contact with the fingerprint community through
16       the International Association for Identification and I
17       am also a member of SWGFAST which is an FBI sponsored
18       group on fingerprint matters.  I am also a member of a
19       subcommittee for the International Association for
20       Identification called the Standardisation II Committee
21       and through these committees indeed I am in contact with
22       practitioners.
23   Q.  Looking at the question of your involvement with Y7,
24       just for the record when the Inquiry team first had a
25       discussion with you about what your role might be, did
page   3
 1       you offer your own view that it would not be beneficial
 2       for you to study Y7?
 3   A.  Yes, I offered that view, indeed.
 4   Q.  Can you explain why it was that you felt it would not be
 5       beneficial for you to study Y7?
 6   A.  Well, at that time I knew that the mark has led to
 7       numerous reports with opposing opinions and I didn't
 8       felt that it will be of any usefulness to offer another
 9       opinion.
10   Q.  So if I put it in very simple terms, the concern was if
11       a substantial number of people with expertise have
12       already expressed an opinion on Y7, you didn't feel that
13       it would help us to have yet another person express an
14       opinion on Y7?
15   A.  Yes.
16   Q.  If I then look, ultimately, you have provided for us two
17       reports that are now to be superseded, in effect, by the
18       PowerPoint presentation that you will do today.
19           If I begin with the first of those ED0003, that is
20       your main report that you have provided for us and if I
21       go to paragraph 8 of that report, it is on page 3, this
22       is in a section where you are dealing with the first
23       task that you were asked to undertake by the Inquiry
24       team along with some others, which was to look at the
25       preexisting reports and to tell us whether by studying
page   4
 1       those reports you could discern the essential points in
 2       agreement or disagreement among the various experts.  Is
 3       that fair?
 4   A.  Yes, that was the task which was given to me, to assess
 5       whether from the documentation made available I could
 6       identify rooms of agreement and areas of disagreement
 7       between the submitted reports.
 8   Q.  Even with your own sort of expertise and background, in
 9       fact when you looked at the various reports you found it
10       a task that was not ultimately beneficial because of the
11       fact that different experts had studied different
12       images?
13   A.  Yes, that was one of the elements which made the task
14       quite complex, is the difference in material that the
15       experts had been made available with.  The main issue
16       was the fact that, apart from the obvious disagreement
17       in the ultimate conclusion, the documents which was made
18       available I considered this material to be more of a
19       post hoc documentation that any documentation may help
20       to understand what was done at the time of the
21       examination by all the experts.  So I felt that it would
22       not be of any benefit, apart from highlighting what the
23       Inquiry already knew, that there is a difference of
24       opinion between these experts and their conclusions, I
25       didn't feel that we will gain anything by looking in
page   5
 1       detail at each charting and points that had been
 2       prepared for the productions that were made available to
 3       me at the time.
 4   Q.  It will be one of the themes we will return to I think
 5       at various points during your evidence, that in fact
 6       what you were concerned with was the material was
 7       written after the event and what you personally would
 8       have been more interested to have studied would have
 9       been the contemporaneous reasoning of the individuals,
10       what they first observed and how they built up their
11       reasoning prior to their initial conclusion?
12   A.  Yes, that's a key element to -- in an attempt to
13       understand why different examiners came to different
14       observation and then ultimately to different conclusions
15       it is of paramount importance to have the notes that had
16       been taken during the phase, which is the
17       information-gathering phase, which I call the analysis
18       phase of the comparison and not what was left now in my
19       hands at the time was charting that had been prepared,
20       most of them much later in the process.  Hence, it's
21       very difficult to reconstruct the chain of evidence and
22       understand where people differed in their construction
23       of their propositions if you don't have that
24       documentation available to you.
25   Q.  In fact, if I can then just pick out a couple of
page   6
 1       sentences in the bottom area, what you were concluding
 2       when you wrote this report for us was that:
 3           "Especially in a case involving a disputed mark, the
 4       mere fact that the examiners have not worked on the same
 5       material and especially in a case with a complex mark
 6       makes the task of comparing these reports
 7       difficult ...", and indeed you say, "... subject to
 8       endless debates."
 9   A.  Yes.
10   Q.  Then you said:
11           "A better setting for a meaningful comparison would
12       be a situation where the examiners would examine the
13       same set of images."
14   A.  Yes, absolutely.
15   Q.  If we are looking at it as an ideal you would have
16       preferred yourself to have seen contemporaneous notes of
17       the reasoning which you were told by us did not exist.
18       Is that fair?
19   A.  Yes.
20   Q.  Then very much a lesser but nonetheless an alternative
21       would be to get the examiners to look at least at one
22       standard set of images because otherwise there would be
23       an endless debate driven by the differences in the
24       source material or influenced by the differences in the
25       source material?
page   7
 1   A.  Yes, that is especially the case when the mark is
 2       difficult, when the mark is complex.  If I may just add
 3       to complement this sentence, of course what matters is
 4       the result of the examination in the analysis phase so
 5       had we given examiners the same set of images to compare
 6       again, it is important that their documentation
 7       distinguish between the information gathered during the
 8       sole examination of the mark and the analysis to be
 9       distinguished from what they gathered following the
10       comparison with the known print.
11   Q.  That is actually helpful because picking up after that
12       sentence which says it would be helpful to start with a
13       set of standard images, just for the record you yourself
14       were not involved in the design of the comparative
15       exercise which was used in the course of these hearings?
16   A.  Not at all.
17   Q.  If I understand you correctly, if you had been advising
18       on a design of exercise you would have preferred an
19       exercise which looked at the analysis of the mark in
20       isolation rather than a charted comparison between the
21       mark and the print?
22   A.  Yes, that will be my view, yes.
23   Q.  However, would you recognise that where individuals have
24       for many years been debating this mark and this print,
25       it would have been an artificial exercise to have asked
page   8
 1       them to make an untutored analysis of the mark Y7, for
 2       example?
 3   A.  Yes, I agree with you, I think after many years of
 4       debate on the mark an examiner will know it by heart
 5       informed by the whole comparison.
 6   Q.  So, in fact, although it is not your preferred
 7       methodology, in this particular instance to come to a
 8       debate about the mark Y7, or indeed the other mark QI2,
 9       at least by reference to a standard set of images has
10       the benefit that it enables one to compare the
11       differences of opinion on a level playing field?
12   A.  Yes.
13   Q.  The other aspect of this I just want to cover is in
14       relation to committee involvement and I will mention
15       some e-mails with a particular individual.  So far as
16       the fingerprint community is concerned, is it a
17       community in which there is limited academic
18       involvement?
19   A.  There is not a lot of academic department in forensic
20       science who teach and research into fingerprint matters
21       and we have the chance to be one of them, but there is
22       not a lot of these departments in the world, in fact.
23   Q.  Therefore, are you and your colleagues at Lausanne very
24       much called upon by authorities throughout the world to
25       assist in relation to these matters?
page   9
 1   A.  I would not say systematic but the research we produced
 2       over the years and the book we publish has allowed us to
 3       be more and more involved at an international level in
 4       various committees and discussion in relation to
 5       fingerprint evidence.
 6   Q.  We have heard that, at least within the Scottish
 7       fingerprint community at a certain point in time, the
 8       practitioners tended to be isolated in their own
 9       bureaux, taught internally, working in their own narrow
10       area so not attending conferences.
11           Are the number of practitioners who are active in
12       committee work quite a small population, a small
13       community?
14   A.  It is a small community active in committee work and
15       probably 100 or 2 individuals which is quite a small
16       community compared to the number of examiners practising
17       on a daily basis.
18   Q.  Therefore, as someone who is involved because of your
19       academic interest in committee work will we tend to find
20       that you will be, in fact, in contact on committees with
21       a number of the individuals who have, in fact, been
22       active in relation to the dispute, in particular in
23       relation to Y7?
24   A.  Yes, that is true.
25   Q.  If I take it, for example, with Mr Pat Wertheim, have
page  10
 1       you had a professional involvement with Mr Pat Wertheim?
 2   A.  Yes, I'm sitting on the same committee as Pat Wertheim,
 3       on SWGFAST and we meet -- this committee meets two weeks
 4       a year in the United States.
 5   Q.  You have also co-authored an article with Mr Pat
 6       Wertheim?
 7   A.  Yes, I did co-author an article in the Journal of
 8       Forensic Sciences with Pat Wertheim and Glen Langenberg.
 9   Q.  So far as the SWGFAST committee is concerned which
10       organisation sponsors the SWGFAST committee?
11   A.  The SWGFAST, all the scientific working groups in the
12       US, and SWGFAST is the group devoted to friction ridge
13       analysis, all these groups are sponsored by the FBI and
14       the purpose of the group is to have representative from
15       federal and local fingerprint bureaux to meet and
16       discuss standard and technology associated with friction
17       ridge skin every two years, make recommendations, more
18       and more proposed standard and standard operating
19       procedures for practitioners to follow.  It's a US-based
20       organisation but the Committee allows for external
21       representative to be present and I have been asked a few
22       years ago to join as a member of SWGFAST, part of the
23       international representation.
24   Q.  So, first of all, we should understand that it is a
25       committee, the ultimate sponsorship of which is the FBI,
page  11
 1       yes?
 2   A.  Yes.
 3   Q.  Secondly, it is there to in fact guide and inform daily
 4       practice of fingerprint practitioners primarily in the
 5       United States?
 6   A.  Absolutely.
 7   Q.  It will have a high local, that is to say American,
 8       representation of practitioners?
 9   A.  Most of the SWGFAST members, which is about 50, are
10       Americans and there is three international
11       representative on the Committee.
12   Q.  You are one of the three.  The other members?
13   A.  The other members are Ian Curry(?) from Canada and
14       Cedric Neumann from the UK.
15   Q.  Mr Neumann is someone who also writes fairly extensively
16       in some of the areas that coincide with your areas of
17       interest?
18   A.  Yes, we have done and still do research together.
19   Q.  He is in the United Kingdom?
20   A.  He works for the Forensic Science Service in Birmingham.
21   Q.  In addition to the SWGFAST that has a contact with
22       Mr Pat Wertheim, do you also have on any professional
23       basis an association with Mr Wertheim's son?
24   A.  Yes, through the same group because he's also a member
25       of SWGFAST.
page  12
 1   Q.  Other individuals who are involved in the Inquiry, if we
 2       go through them, Mr Zeelenberg, what is your
 3       professional involvement with Mr Zeelenberg?
 4   A.  I'm currently sitting with him on a committee of the
 5       International Association for Identification and the
 6       subcommittee is called the Standardisation II Committee
 7       and I was asked by the Chairman of that Committee to sit
 8       and Arie Zeelenberg is also a member of that Committee.
 9   Q.  So Mr Zeelenberg is a member and we have also heard
10       Mr Chamberlain of the Forensic Science Service is also a
11       member of that Committee?
12   A.  Yes, indeed.
13   Q.  The role of that Committee?  What is the role of that
14       Committee?
15   A.  This Committee was set up by the IAI, which is the
16       acronym for International Association for
17       Identification, to revisit the soundness and the
18       relevancy of two of their resolution that had been
19       passed, one in 1973 which I will alluded to in my
20       presentation and also the other one which dates back to
21       the '80s, 1980, and these resolutions are giving
22       guidance to examiners on two issues.  One is the concept
23       of what is the minimum number of features required for
24       an identification and the other resolution invited
25       examiners to avoid using terms like possible and
page  13
 1       probable in the context of fingerprint evidence but
 2       restrict their testimonies to identification and
 3       exclusion, otherwise it's termed inconclusive.
 4   Q.  So, in fact, if I understand these two resolutions the
 5       one in 1973, if I understand it correctly, is critical
 6       in the move in some countries from a numerical standard
 7       to a non-numerical approach to fingerprint
 8       identification?
 9   A.  Yes, it is.
10   Q.  There are variations in practice among countries.  Some
11       have moved to the non-numeric and some have remained
12       numeric?
13   A.  Yes, and I hope my presentation will set the scene as
14       far as this duality is concerned.
15   Q.  But at least so far as your own personal position is
16       concerned, would it be fair to say you are very much in
17       favour of a non-numerical approach?
18   A.  Yes, I am.
19   Q.  So therefore you are very much in favour of the approach
20       that is adopted in the various countries within the
21       United Kingdom today?
22   A.  Yes.
23   Q.  The second one you have mentioned we understand and we
24       have had evidence, particularly from Mr Chamberlain,
25       that the IAI currently forbids its members to express
page  14
 1       the conclusions in terms of anything less than
 2       certainty.  Is that correct?
 3   A.  Yes, that's correct and that is the essence of the 1980
 4       resolution which is under review by the Standardisation
 5       II Committee.
 6   Q.  If I understand your own perspective on that, your own
 7       perspective is that you prefer that there be a more
 8       explicit statistical basis to opinion evidence in court?
 9   A.  In relation to the IAI 1980 resolution my view is that
10       it should not be practitioner which restrict themselves
11       as to the strength of the evidence they propose to the
12       court room.  It's a matter for the judiciary to decide
13       if less than certain evidence is useful for the
14       fact-finder.  So my take on the IAI resolution has
15       always been that it may be practical as a possibility to
16       restrict the range of conclusion an examiner may offer
17       but the strength of the evidence is a continuum
18       statistically speaking, indeed, that I would see no harm
19       at all for less than certain evidence, if I may say so,
20       to be presented in a case should the court allow it and
21       find it useful for the fact-finding.
22   Q.  No doubt again, not only because of your role as an
23       academic in fingerprints but also because of the
24       particular research you do in relation to statistics,
25       your expertise is particularly germane to that
page  15
 1       Committee?
 2   A.  Yes.
 3   Q.  We have covered Mr Pat Wertheim, Mr Kasey Wertheim,
 4       Mr Zeelenberg.  Now if I move to some of the others, we
 5       have also covered Mr Chamberlain from whom we have had
 6       evidence, Mr Zeelenberg has been mentioned.
 7       Mr Leadbetter we have also heard evidence from
 8       Mr Leadbetter who mentions that he has also been on a
 9       committee with you.
10   A.  Yes.  When I joined the Forensic Science Service in the
11       year 2000, I was asked to join a UK committee devoted to
12       Level 3 features and that Committee was under the
13       umbrella of ACPO and Martin Leadbetter was the Chair of
14       that Committee and I attended a few meetings to discuss
15       the status and use of Level 3 features in the UK under
16       his presidency.
17   Q.  One other more specific matter was at one stage we asked
18       you about some contact, e-mail contact, you had had with
19       a particular individual, Mr Horn.  If I bring up the
20       material we have relating to that, FI0202.
21           This you will see starts with an e-mail that was in
22       fact sent to a Mr Flinn and Miss Gilpin who were members
23       of the Inquiry team.  If I give you just a chance to
24       read, I don't think you will have seen this particular
25       e-mail before?
page  16
 1   A.  I don't think I've seen this e-mail before.
 2   Q.  Exactly.  (Pause)
 3   A.  Would you like me to comment on this?
 4   Q.  If I just take it in stages, first of all, you will see
 5       that the thrust of this is the suggestion that you had
 6       in fact taken a view on the case of HMA v McKie, in
 7       other words you had taken a view on Y7 and other
 8       features of the case.  Is that --
 9   A.  No, I didn't.
10   Q.  So far as what is attributed to you, you will see in the
11       second paragraph it says:
12           "Professor Champod is quoted as saying ..." and then
13       there is a lengthy quotation.
14           Is that in fact an incorrect representation of the
15       correspondence you had with Mr Horn?
16   A.  I have no recollection of having written or sent these
17       two paragraphs.
18   Q.  That is fine because, in fact, what we will now see is
19       that though you were not shown the particular e-mail
20       that is on the screen just now, you were in fact asked
21       by the Inquiry team about your contact with Mr Horn?
22   A.  Yes, I was.
23   Q.  You provided, in fact, the exchange of e-mails with
24       Mr Horn?
25   A.  Yes, I did.
page  17
 1   Q.  If I can take you, if I understand, if I bring up
 2       page 3, there is in fact an Internet posting by someone
 3       called Outsider that I will come back to just in a
 4       moment.
 5           If I proceed then to page 4 and in fact if I go to
 6       the end we will come back through.  Because they are
 7       e-mails it is in reverse order.  So if we go back to
 8       page 7 and 8, we will see an e-mail on 11th December at
 9       the foot from Mrs Ann Nelson to you indicating that the
10       Chairman had received some submissions.  We will see at
11       the top of the right-hand page that one matter raised in
12       the submission concerns your possible connection to
13       Mr Steven Horn.  You had been told that Mr Horn, using
14       the name Outsider, made a posting on CLPEX website some
15       time ago quoting correspondence he had with you about
16       the McKie case and you were then asked -- you were given
17       the reference to that posting:
18           "Can you help us, please, as to whether this is in
19       fact correct and as to any connection that you may have
20       had with Mr Horn?"
21           So you were asked that.  If I then bring up pages 6
22       and 7, we will go back, pages 6 and 7 of this particular
23       document, is there then set out what you, in fact,
24       copied to the Inquiry team as the initial approach from
25       Mr Horn to you?
page  18
 1   A.  Yes, it was.
 2   Q.  If I take it from the beginning where it says:
 3           "Sorry I have to write to you in English.  Your name
 4       came up on a Google search I did recently ..."
 5           Was this exchange with Mr Horn, so far as you were
 6       concerned, an unsolicited e-mail?
 7   A.  I never heard before that day, the e-mail I received, I
 8       never heard of this person before.
 9   Q.  But you did, however, respond to his e-mail?
10   A.  I did, yes.
11   Q.  His particular interest -- and we will perhaps touch on
12       this very lightly for obvious reasons -- we see on the
13       right-hand page about halfway down was something in
14       relation to Bayes Theorum and the Shirley McKie case.
15           Is Bayes Theorum a particular statistical theory?
16   A.  Yes, the purpose of his e-mail was to alert me on a
17       discussion he has published on the Internet on the
18       application of a probabilistic theorem, which is called
19       Bayes Theorum, to the special circumstances an element
20       of this case.
21   Q.  If we bring up at page 4 and 5, do we see at the bottom
22       of page 5, it is at the bottom of the right-hand side,
23       the extent of your response to Mr Horn?
24   A.  Yes, that was my response to Mr Horn.
25   Q.  If I can keep page 5 on the screen, please, and then go
page  19
 1       back to what I think will be page 3 so keep page 5 and
 2       go back to page 3.
 3           Page 3 now has the posting from the Internet, the
 4       Outsider posting.  We will see, in fact, on this
 5       particular copy it has a passage indented in it and is
 6       the indented passage, in fact, a quotation from your
 7       response?
 8   A.  As far as I can just check, yes, it seems to me that it
 9       is the entire passage from my response, yes.
10   Q.  So, in fact, what you said was:
11           "Thanks for this information.  I followed your
12       messages on CLPEX including your recent exchange with
13       Cedric Neumann on the LR concept."
14           Is that the likelihood ratio?
15   A.  Yes, it is.
16   Q.  "I view your analysis as a global analysis of the case
17       as a whole (addressing both the value of the fingerprint
18       evidence and the assessment of the strength of the case
19       arising from other elements)."
20           If I just stop there, if I understand that, we will
21       come later to ask you about probability and likelihood
22       ratios.  I understand when you are looking at likelihood
23       ratios you are looking at the likelihood of a match in
24       relation to the fingerprint evidence alone.  Is that
25       correct?
page  20
 1   A.  Yes.  When forensic scientists use the term likelihood
 2       ratio they tend to refer to the weight associated with
 3       the findings, the forensic findings, alone so in that
 4       case it will be the fingerprint evidence alone.
 5   Q.  However, what Mr Horn is arguing for on Bayes Theorum is
 6       to take into account other variables, in fact variables
 7       apart from the fingerprint evidence and that would
 8       include evidence about whether Ms McKie had been
 9       observed to enter the house or not.  Is that what he is
10       talking about?
11   A.  I haven't checked his website before coming here but
12       from the memory I have from it indeed he was offering an
13       analysis which encompasses all the elements of outside
14       or in addition to forensic science in the case and
15       combining them in using Bayes Theorum.  That was a
16       proposal made by Mr Horn.
17   Q.  In fact, when we come later to evidence that wider range
18       of an extrinsic variables is not part of the methodology
19       that you are advocating?
20   A.  No, as I was mentioning in the bracket in the response
21       to Mr Horn, the Court of Appeal had taken the view that
22       asking a jury, for example, to combine various pieces of
23       evidence being of forensic nature, like DNA in the case
24       at the time and eye witnesses, for example, was using
25       Bayes Theorum was outside what we can reasonably ask
page  21
 1       from a jury and I think this is a fair ruling to say.
 2       Hence there is a tendency to restrict the expression of
 3       the weight of the evidence on the scientific element
 4       alone and leave to the court and the fact-finder the
 5       duty of combining these elements together with the rest
 6       of the other evidence at hand.
 7   Q.  So if, for example, a statistician were to suggest there
 8       could be a mathematical formula to weigh in the balance,
 9       eye witness evidence such as someone being observed to
10       go into a house or not, and the fingerprint evidence,
11       that an application of Bayes Theorum, that statistical
12       approach of measuring these different pieces of
13       evidence?
14   A.  If we want to combine, as an academic exercise, to
15       combine various pieces of evidence together, all of them
16       weighted by a numerical value, then in that process
17       somewhere we will use Bayes Theorum.
18   Q.  The Court of Appeal ruling that you understand
19       discourages that approach, is that the case of Doheny
20       and Adams?
21   A.  Yes, that's the case of Doheny and Adams and I
22       understand the court discouraged the numerical exercise
23       of combining an eye witness evidence which will be
24       transformed into some sort of statistical value with DNA
25       evidence which is already presented in statistical form.
page  22
 1       The Court of Appeal discouraged the mathematic exercise
 2       of combining these two numbers together.
 3   Q.  That, in fact, will give us a fuller explanation of what
 4       you now write to Mr Horn because you say:
 5           "I view your analysis as a global analysis of the
 6       case as a whole (addressing both the value of the
 7       fingerprint evidence and the assessment of the strength
 8       of the case arising from other elements).  Forensic
 9       scientists (and especially in the UK, following appeal
10       courts rulings pertaining to DNA) ..." so that would be
11       Doheny and Adams and indeed there is another Adams case
12       to referred to in it?
13   A.  Absolutely.
14   Q.  "... tends to restrict their testimony to the forensic
15       findings alone (without commenting on the prior or
16       posterior probabilities on the issue itself)."
17           Again, without getting too much into statistical
18       modelling, the prior or posterior probabilities on the
19       issue itself is that what brings in extrinsic factors
20       such as eye witness evidence?
21   A.  Yes.
22   Q.  "Hence while reading your material, forensic scientists
23       may have difficulties to accept the point of view you
24       adopted.  When read by the trier of fact, it make full
25       sense of course."
page  23
 1           Then you go on to make some comments about the OJ
 2       Simpson case and you sign off by saying:
 3           "Thanks for sharing that with me and I hope to have
 4       an opportunity to meet you at some point.  Please
 5       continue to contribute to the fingerprint/forensic field
 6       as debate always promotes critical thinking and
 7       understanding."
 8           That was the end, as far as you were concerned?
 9   A.  Yes.
10   Q.  Did you have any further contact with Mr Horn, as far as
11       you are aware?
12   A.  I received follow-up e-mails from Mr Horn informing me
13       of messages that had been posted on CLPEX.  I had also
14       roughly around the same time a request to publish on the
15       Internet two of my papers which unfortunately it was not
16       possible because they were under copyright, hence I did
17       not authorise him to publish these papers on the
18       Internet.
19   Q.  Did you go on in relation to this particular debate to
20       entertain yourself a consideration of the proposition
21       that Mr Horn had advanced?
22   A.  Not at all.  I never discussed with Mr Horn any of the
23       specificities of the model he suggested.
24   Q.  If we, therefore, take that now back to what was posted
25       on the Internet, the quotation in the Outsider article
page  24
 1       that I can now highlight is the thrust of what you sent
 2       to Mr Horn now probably explained, about the
 3       significance of Doheny and Adams?
 4   A.  Yes.
 5   Q.  Then what comes ultimately in the e-mail that you have
 6       not seen -- if I take that down just now, sorry -- what
 7       ultimately then comes to be attributed to you in an
 8       e-mail sent to the Inquiry which was written:
 9           "I was wondering if anyone can remember if Shirley
10       McKie's defence at the perjury trial ..." et cetera.
11       This is not written by you?
12   A.  No, not at all.
13   Q.  If I take it back to the original, we can in fact see
14       that the formatting of the Internet copy here correctly
15       indicates that all that Mr Horn was doing was, in the
16       indented passage, quoting the limited and, indeed,
17       guarded response that had come from you?
18   A.  Yes and, indeed, he quoted that with my permission.  He
19       asked me for permission to quote these few lines.
20   Q.  But then beyond that, it is the personal observations of
21       Mr Horn?
22   A.  Yes.
23   Q.  Thank you, Professor Champod.  So if I can take that
24       then, that is the background, I will then just update
25       us.  We have looked at your first report, that is for us
page  25
 1       ED0003, and I just simply looked at the extent to which
 2       you contributed to the exercise, the work of the team
 3       preceding the comparative exercise.  I take it you had
 4       not been involved in the design of the comparative
 5       exercise.
 6           So far as the results of the comparative exercise
 7       when we received in chartings from various individuals,
 8       you were not yourself asked to analyse those or have any
 9       input in relation to them?
10   A.  No, I have not been asked to look at any of these
11       chartings.
12   Q.  However, you were asked by the Inquiry team to provide a
13       report which you have which is your first report which
14       we brought up ED0003?
15   A.  Yes.
16   Q.  That relates to general matters relating to fingerprint
17       practice and ultimately to probability studies?
18   A.  Yes.
19   Q.  At a later stage, were you also asked to expand to
20       include some observations on Third Level Detail?
21   A.  Yes.
22   Q.  If I bring up ED0004, that is just the cover page.  At
23       least for me, it ends on page 7 of, 9 which may
24       electronically be page 8.  We will just bring up page 7
25       of 9.  It is just simply so people can see the date of
page  26
 1       this report was 21st October 2009?
 2   A.  Yes.
 3   Q.  So that you have given some assistance to the Inquiry in
 4       relation to Third Level Detail and an understanding of
 5       that?
 6   A.  Yes.
 7   Q.  Rather than take you through all of these reports in
 8       detail, what you in fact have been asked to do is to
 9       prepare a presentation?
10   A.  Yes, indeed.
11   Q.  That takes us back, in fact, to the position of just a
12       little over a year ago where in a meeting in Edinburgh
13       in Drumsheugh Gardens you made a presentation to the
14       Inquiry team, including the Chairman?
15   A.  Yes, indeed.
16   Q.  So while you will, in what you are about to do,
17       reproduce the entirety of what was there what you have
18       been asked to do, for the benefit of the public and the
19       other lawyers who are involved, is to give them the
20       essence of what you told the Inquiry team and the
21       Chairman a little over a year ago?
22   A.  Yes.
23   Q.  So that everyone can understand the information and the
24       input you had at that time?
25   A.  Yes.
page  27
 1   Q.  You have also expanded it now to include the benefit of
 2       the two later reports so that your contribution to the
 3       Inquiry is now distilled into the presentation you are
 4       just about to give?
 5   A.  Yes.
 6   Q.  What I will do, Professor, is therefore just hand over
 7       control to you to enable you to present your
 8       presentation.  I trust not to interrupt you unless I
 9       feel a desperate urge to do so and I will save my
10       questions until the end.  What you may want to do, you
11       will have seen the difficulties of working with the
12       microphone, you may want to adjust your microphone to
13       make it easier so that we can hear you as you are making
14       your presentation.
15   A.  Thank you.
16               [Inquiry document reference ED0005]
17           So, my Lord, the objective of this presentation is
18       multiple.  First, it is to bring back to our memories
19       some of the discussion that we had at the Inquiry office
20       last year.  In addition, I have structured the
21       presentation with chapters according to the main themes
22       of the reports.  I took the view that instead of going
23       into all details I will try to summarise as much as I
24       can and in doing this I have been helped through the
25       conversation we had last week over the telephone to
page  28
 1       focus on elements which was mentioned to me to be
 2       elements of still debate or discussion or complexity for
 3       this Inquiry.
 4           I will end up with some perspective on the case and
 5       the original material for this presentation is of course
 6       the two reports which have been served; the presentation
 7       which I gave to the Inquiry in November 2008.  I have
 8       distilled -- I have removed some of the slides from this
 9       earlier presentation either because at the time we
10       hadn't a chance to go through everything I had prepared
11       for that meeting or because I felt that some of the
12       slides were not bringing any additional information for
13       the benefit of today.
14           I started the presentation in November last year by
15       going back to the very basic of fingerprint examination
16       and the basis, of course, is the morphogenesis of
17       papillary ridges.  To help to identify the slide which I
18       have used with the Inquiry, my Lord, last year I have
19       indicated the date down here, that is the slides which
20       indicate November 13, 2008 are exactly the same slides,
21       sometimes with slight modification, mainly to remove the
22       English typo, I managed to find some, but these are
23       exactly the same material which had been discussed with
24       the Inquiry a year ago and you will see later there are
25       slides indicating the date of today.  This is the
page  29
 1       material which I prepared last week in preparation to
 2       this hearing.
 3           There is two basic premises which are used and
 4       accepted as true to developed fingerprint evidence in
 5       forensic practice.  One is the fact that the friction
 6       ridge arrangement are persistent.  I will come back to,
 7       the next slide will explore this concept of persistency
 8       further.  But at the time of roughly 25th week of the
 9       foetal development, all our papillary surfaces are
10       covered with ridges which will remain in that state
11       throughout life until death.
12           Of course the organ grow but the intrinsic
13       relationship between the ridges remain the same so we
14       have a (inaudible) growth of pattern as we grow up but
15       the intrinsic relationship between the ridges will
16       remain stable throughout life.  This is a very important
17       aspect.  That is the one which allow police officers to
18       identify a person, people on the basis of a record which
19       has been taken maybe years before the incident of
20       interest.
21           The other element which help or support the use of
22       fingerprinting for identification issues is the fact
23       that skin arrangements are extremely discriminating and
24       they do vary from person to person.  The reason why it
25       is extremely discriminative lies in the very
page  30
 1       morphogenesis of the friction ridge skin.  Then I will
 2       refer sometimes during this talk I will refer to the
 3       concept of between sources viability and by that we mean
 4       that if we looked at different fingerprints from
 5       different people, we will identify very easily
 6       differences which allow to discriminate between these
 7       impressions from one person to the other.
 8           This discrimination is extreme.  When I say extreme
 9       is that to my knowledge, nowadays we haven't found two
10       persons showing the same arrangement in their
11       fingerprints.
12           These are the two premises and the permanence is the
13       first.  The skin is composed of two layers: one is
14       called the dermis and the outer layer is called the
15       epidermis and here on the screen you have a
16       cross-transversal cut of a layer of the epidermis.
17           The epidermis is composed of multiple layers of
18       cells from a base layer, called basal layer, which is
19       just at the interface between the dermis, which is not
20       represented on this image which will be lower, and the
21       epidermis.  Up to the last layer, the one we are in
22       contact with the outside is called the horny layer.
23           The cells of your epidermis are produced
24       systematically and in concert throughout the time on the
25       basal layer and the cells migrate from the basal layer
page  31
 1       up to the horny layer to be lost because the cells are
 2       dead and we have a constant regeneration of cells on the
 3       epidermis starting from the basal layer and progressing
 4       to the horny layer.
 5           The fingerprint pattern which we are interested in
 6       is formed on the dermis and I will show you later slides
 7       of the initial formation of this.  So the blueprint is
 8       on the dermis.  The epidermis information, so the one we
 9       see when we ink a friction ridge skin on a surface, this
10       is an exact replication, there is some mirror image
11       issue but it's an exact replication of information on
12       the dermis.  That's explained why fingerprints are
13       persistent.
14           Because of these constant regeneration local
15       injuries, which does not cut the dermis, will just be
16       removal of outer cells of the epidermis.  It will make a
17       visible scar on the finger which for a limited amount of
18       time will be visible of course but because of the
19       regeneration process it will come back and regenerate ab
20       initio as a function of the blueprint on the dermis.
21       Hence throughout life, even though we are losing a lot
22       of cells, this generative process from the basal layer
23       up to the horny layer ensures that we have a persistence
24       of these features throughout time.
25           Of course, if the basal layer is cut, through a
page  32
 1       scar, for example, then it will heal and the blueprint
 2       will be modified accordingly.  So the scar will be
 3       visible and the scar itself will gain persistency and
 4       will then the cells will be regenerated again not to the
 5       exact initial information but to the information that
 6       had been modified by the scar, but the persistency of
 7       scars, when we are talking about scars at the dermis
 8       level, are as of the same nature as the persistency of
 9       friction ridge skin.  So this is the first premise to
10       the use of fingerprinting.
11           The development I mentioned it is ended roughly at
12       week 25 of the foetal life.  The early start of the
13       friction ridge skin development is around 10 to 12 weeks
14       of development, roughly at between 12 and 16 weeks we
15       have all the development on the hand, which is what we
16       have here (indicated) and a little bit later you have
17       development on the feet.
18           The development is illustrated here by this sketch
19       (indicated).  It happened on the dermis.  This is the
20       dermis and this little layer here is the epidermis.  So
21       the waving aspect, the creation of friction ridges
22       happens at the dermis at these very first weeks.
23           If we look at fingers, the dermis, not the
24       epidermis, the dermis of fingers from aborted foetuses
25       throughout the weeks this is what you can see at week 10
page  33
 1       and here we can see the very early development of
 2       friction ridges on the finger.  This is a little bit
 3       later, of course it is not the same finger for obvious
 4       reasons, it's a little bit later in the process at 11
 5       weeks.  The general flow is dictated by the shape of the
 6       pad on which the friction ridges will develop.  The
 7       exact positioning of what we will call later ridge
 8       ending and bifurcations, we know that there is a lot of
 9       stochastic element involved in the creation of a ridge
10       ending at one position or another.  One way to view this
11       is to say that in fact we have multiple waves of
12       development or points of development on the surface
13       which at some point will cover the whole surface.  Hence
14       there is a lot of uncertainty as to exactly how these
15       ridge flow will connect to each other and in these
16       operations then it will create ridge endings and
17       bifurcations but they will be distributed in a very
18       stochastic way on the surface.
19           The ridge flow, in general terms, is dictated
20       through -- not well-defined but we know that there is a
21       genetic component to it.  The way the minutiae will
22       arrange themselves on this pattern are much more subject
23       to the stochastic element of the growth more than
24       genetic conditioning.  That is for this reason, that if
25       you look at fingerprints taken from twins you most of
page  34
 1       the time have no difficulty whatsoever to distinguish
 2       them as soon as you look at the configuration of these
 3       endings and bifurcation on the ridges.
 4           This is a little bit later, always on the dermis
 5       (indicated).  I will have one slide later but just to
 6       pre-empt on the concept I am going to use this morning.
 7       This is what we call the general pattern or the general
 8       flow of the ridges and I'm referring here to the ridge
 9       ending, a ridge ending and sometime there are
10       bifurcations like here (indicated) which are the major
11       deviation in the path of these ridges.  This is all set
12       on week 14th.
13           Then later in the process, so roughly at that stage
14       here we are at week 14 for the hand, later the ridges
15       tend to mature and by mature I mean they will grow in
16       depth.  The concept is that we want an epidermis which
17       really stick to the dermis.  We want to have a very,
18       very firm interface between the dermis in the epidermis
19       otherwise we may lose the epidermis.  This intimate
20       relationship is gained by having quite a convoluted
21       surface of contact between the dermis and the epidermis.
22           So what happen is from quite a slick ridged surface
23       the grow will continue and it will grow down in the
24       dermis with quite some depth and the reason for the
25       depth is to increase the surface of contact between the
page  35
 1       two, hence increase the gluing, if I may say so, between
 2       the dermis and the epidermis.
 3           When this what we call secondary development
 4       occurred, additional shapes are added to the picture.
 5       They are shown on this image.  This image is taken by a
 6       scanning electron-microscope of the top surface of the
 7       dermis and it's later in the process, we are talking
 8       about the week 20.  These are the early development we
 9       talked about and these holes are the positioning of the
10       pores.  The pores are connected to glands in the dermis
11       and it is through the pores that we produce eccrine
12       secretion which are produced on our papillary surfaces.
13           The side of the ridges are these shapes (indicated).
14       So we can see that it is a bit later in the development
15       and the shapes are not -- the shape of the ridges are
16       not straight rails but they are curved and have very
17       specific forms.
18           If we move later in the process and look at higher
19       magnification, the shapes of the ridges are taking
20       various forms.  These forms will translate themselves on
21       what we have on the epidermis.  So not only our surfaces
22       are made of a flow of ridges on the epidermis, these
23       ridges may stop or bifurcate on the epidermis.  In
24       addition, if we look at a higher magnification, the
25       edges of the shapes and the forms taken around the pores
page  36
 1       are -- vary a lot.  They are not straight lines.  There
 2       is a large variation in the shapes you can observe.
 3       This is because of the secondary development, some sort
 4       of a maturation of the ridges which then leads to
 5       specific shapes.
 6           So that leads me to introduce something which I'm
 7       sure the Inquiry have been exposed to numerous times.
 8       It's a standard way to communicate about features.  We
 9       tend to use three levels of detail but before
10       introducing the level of details, there is two focal
11       points which examiners tend to use a lot for mainly
12       classification purposes in the past.  One is core and
13       here on this loop I have indicated this point
14       (indicated) where the ridges just recurve with the most
15       angle or slope, if I may say so, that will be the core
16       of the pattern.  Generally, it will be at the centre of
17       a fingertip if we talk about fingerprints.
18           The other standard point which we call focal point
19       is the delta and the delta is where the free flows of
20       the ridges come together in this area and the meeting
21       point is called a delta.  That will help to discuss what
22       we name level 1 features and level 1 is the general flow
23       of the ridges.
24           Now, practitioners in the past have, because of the
25       use of classification systems, put labels on categories
page  37
 1       and the standard labels to distinguish between various
 2       flow we observed on the fingers is to use the concept of
 3       loops to the right, to the left, whorl or arch and
 4       that's where the concept of focal points help.  A loop
 5       has only one delta, either to the left like on a right
 6       loop or to the right like on a left loop.  The left and
 7       right to the loop refer to where the ridges flow going
 8       outside of the pattern.
 9           A whorl has two deltas, generally speaking.  An arch
10       has no delta.  So the focal point allowed the early
11       researchers in fingerprint science to classify
12       fingerprints and it is because fingerprints could be
13       classified so you can take a ten-print card of an
14       individual and assign some sort of a classification code
15       and if you multiply this with quite a large population
16       if the classification is efficient that offer a
17       possibility for you to come back to the classification
18       with an unknown and find the potential correspondence
19       without having to look at the entire collection.  So the
20       level 1 has been used traditionally as a sorting device
21       to classify fingerprints.
22           Even though I have restricted the discussion to four
23       classes there is a continuum of classes and this image
24       shows that although that will be described as a whorl
25       and this will be described as an arch (indicated) two
page  38
 1       deltas and no deltas, this old classification was
 2       proposing 39 classes, which, if we look at what is
 3       happening, it is a progression in terms of shape between
 4       a whorl down to an arch and nature in the various
 5       patterns we can display on a finger, although we will
 6       say it is an arch or a whorl the reality of the
 7       production is quite a continuum of shapes between what
 8       we will classify as a whorl and what we would classify
 9       as an arch.
10           So, yes, although we tend for classification
11       purposes to restrict the concept of level 1 to a tag,
12       it's a whorl, it's a loop, it's an arch.  The reality of
13       the flows nature can produce on the tip of our fingers
14       here, the same applies to different degrees to the palm
15       and sole on the tip of our fingers, it's a continuum of
16       shapes.  That's what we refer to Level 1, general flow.
17           The concept of Level 2 is where the major events
18       ridges will take if you follow them and the two main
19       events are ridge ending and bifurcation and they are
20       represented here.  This is a ridge ending and this is a
21       bifurcation (indicated).
22           Of course you can observe on fingerprints a
23       combination of these two which by the coincidence of
24       putting a ridge ending close to a bifurcation you may
25       create a specific type of minutiae which some
page  39
 1       practitioners may have given a name to, for example, a
 2       spur generally referred to a ridge ending which is
 3       connected just to the nearby bifurcation or an island
 4       will refer to two ridge endings connected through this
 5       relationship on the ridge, whereas a lake will be two
 6       bifurcations which just happen to be connected together
 7       to form a lake and other combined minutiae may occur.
 8           To ease communication sometimes we may use these
 9       terms to point on a image on a spur because it helps to
10       focus the attention but at the end of the day I always
11       find it easier to decompose the combined minutiae and
12       talk about one ridge ending and a bifurcation or two
13       bifurcations or two ridge ending respectively for the
14       spur, the lake and the island.
15           The third level of detail -- and these categories
16       again, I insist they are not dictated by biology they
17       are just common agreement in terms of language we're
18       going to use to designate something on an image.
19           Level 3 features refer to features which need higher
20       magnification to be visualised easily and the
21       tendency -- and I will come back later to this -- the
22       tendency is to focus on the edges and the pores and
23       their shape as being the encompassing the concept of
24       Level 3 features.  You need higher magnification to
25       visualise them comfortably.  I will say more about Level
page  40
 1       3 features later in my presentation.
 2           I have used this slide in November last year and I
 3       think it is very important to state it again.  Quite
 4       often we have -- and when we say we, it's fingerprint
 5       practitioners and people involved in this area and
 6       sometimes myself as well -- we have a tendency to use
 7       the term "unique" without paying enough attention to
 8       what we are trying to convey.
 9           When I say all portion of friction ridge skin are
10       unique, it is a statement of the obvious.  Every entity
11       is unique and can be only identical to itself.
12       Everything is unique and the argument whether everything
13       is unique in nature, I'm quite happy to buy it without
14       any difficulty.
15           The issue about the use of fingerprint mark to be
16       compared against fingerprint -- sorry, I meant mark
17       against print -- is not whether it is unique, the issue
18       is whether we can distinguish individuals based on the
19       material left respectively as marks and the material
20       left respectively as print and we are not looking at the
21       finger itself which may very well be or is unique.  We
22       are looking at a production left on a surface under
23       specific circumstances for the mark, the circumstances
24       are not controlled.  It may be on various surfaces with
25       various type of material contaminating the finger, under
page  41
 1       pressure and distortion, and for the print most of the
 2       time it is under controlled conditions because we are
 3       normally taking these print with ink in the process of
 4       gathering known information about the individual, hence
 5       the control about the quality is most of the time higher
 6       than the quality we can expect on chance marks.  But we
 7       are comparing exacts(?) of the unique finger produced if
 8       we are comparing things which are degraded information
 9       compared to the information we can gather ideally by
10       looking at the finger itself.
11           The real issue is more on the distinguishability
12       than on the uniqueness.  It's how good examiners are in
13       their capacity to distinguish marks and print when they
14       come selectively from the same persons or from different
15       persons.  It's a problem of distinguishability and not a
16       problem of uniqueness.
17           To illustrate this, I will show you a few images
18       showing the degradation and the various shapes and forms
19       a print and then marks from the same individual can
20       take.  Here you have an inked impression from one
21       individual rolled nail-to-nail, as we say, with ink.
22       The next impression is the high resolution LiveScan
23       device.  It's a numerical device, optical device, that
24       is used to take impressions.  This impression is
25       obtained in a process we called flat: that is not rolled
page  42
 1       nail-to-nail but the finger is just deposited on the
 2       surface of the sensor.  The right image is the same
 3       finger but acquired this one on a lower resolution
 4       LiveScan device.  All these three images are coming from
 5       the same individual.  They are not identical.  They are
 6       all different but they reflect features which we can put
 7       in relation.
 8           Now, if I take that person and ask that person to
 9       produce marks, these are marks left by this very finger
10       originally on various substrate and detected with
11       various detection techniques.  How the mark will be
12       visible here is completely different from how the print
13       were.  So they are not identical.  But when we try to
14       say can we distinguish that impression, the mark, from
15       the print taken from that individual, we may come up to
16       the conclusion that we cannot distinguish the features
17       which are in agreement between the two or we can.  In
18       that case, we can't.
19           The extent of information available on the mark will
20       of course vary quite a lot depending on the quality of
21       the mark.  Here we have an under-representation of the
22       whole information.  Here we have a far extensive
23       representation of the information available on the
24       friction ridge skin.
25           Two other examples on plastic, on tape.  Sometimes
page  43
 1       the substrate itself makes things more difficult because
 2       of texture, background may make more difficult the
 3       visibility of some features.  If we look at the
 4       background of this tape, the surface itself may have
 5       sections which are unconnected one to the other, so when
 6       the finger comes into contact with that surface it may
 7       leave a mark which would be distorted or more difficult
 8       to read.
 9           I move now to a key chapter of my report which is in
10       relation to ACE-V and the identification process.  ACE-V
11       is a protocol which has been suggested to be used in
12       fingerprint identification by David Ashbaugh.  His
13       original source was a publication by a document examiner
14       called Roy Huber.  They have adapted through the
15       application proposed by Huber.  It was also an
16       application that was proposed in Canada for footwear
17       mark examination and just applied this standard protocol
18       to fingerprint examination.
19           When I say "protocol", I want to insist on the
20       concept that it gives you guidance as to how to proceed
21       from one step to the other.  It invites you to do first
22       an analysis before moving into the next step, which is
23       called the comparison and I will go into the detail of
24       both of these four steps.  So it is a protocol which
25       will tell you somehow the standard operating procedure
page  44
 1       that needs to be followed when we conduct a fingerprint
 2       examination.  It's a protocol, but the protocol does not
 3       tell you a lot about the decision-making, when a feature
 4       is sufficiently different to be considered as a
 5       discrepancy, or when do I have enough information in the
 6       comparison to satisfy the examiner that he is talking
 7       about the same source?  These two fundamental elements
 8       are not declared or disclosed in ACE-V.  ACE-V is a
 9       protocol.  It's a very good protocol.
10           The next two slides are coming from a flow chart
11       which I collaborated with my colleagues in preparation
12       of the book we published a few years ago at the CRC
13       Press.  The reference is on the top of the slide.
14           The analysis is focused on the mark only.  By
15       extension, we can do the analysis of the print should
16       the print be of so poor quality that it really deserves
17       a full analysis, but most of the time the material which
18       is difficult is the mark.
19           The mark is analysed just to understand on the face
20       of the mark information only what is the reliable
21       information we can derive from the mark.  It's an
22       acquisition of information which will help to understand
23       what are we talking about, is that really a friction
24       ridge skin impression, how is the substrate, do I see
25       superimposition, multiple touch, slippage, distortion?
page  45
 1       These factors are identified and should be identified
 2       during the analysis.  It sets the scene for the next
 3       step of a comparison and then the end product of the
 4       analysis is a decision.  So I will go through some
 5       practical examples which I felt, I hope, my Lord, will
 6       help understand this quite cryptic flow chart I
 7       presented a year ago.  So that's where I added a few
 8       slides to discuss a concept which is named "tolerances".
 9           Examiners will set tolerances after this
10       information-gathering step of the analysis stage.  It
11       can be viewed as somehow as the boundaries they will
12       allow in their comparison process and I will expose this
13       with an image.  This is a mark which you have already
14       seen before on this adhesive tape.  The mark itself,
15       when you put a finger in contact with an adhesive tape
16       you will all have experienced that it sticks to your
17       finger and when you remove it there is distortion which
18       will occur and distortion is visible on this image.  An
19       examiner looking at this mark will be careful in his
20       judgment as to the exact shape/flow of the whorl he's
21       expecting.  Because we are talking about an adhesive
22       tape, then the distortion may have affected the shape of
23       the core and the global angle of that core.  So an
24       examiner will allow, when he's searching on a ten-print
25       card, for distortion to have occurred; hence he will
page  46
 1       apply quite a large tolerance in looking at potential
 2       for potential correspondence.
 3           I will use this little illustration (indicated)
 4       saying, well, at this time when the person will embark
 5       into a comparison, his match window, so to speak, will
 6       be larger because he has identified distortion issues on
 7       the mark.  So not only is he looking at a whorl, but
 8       he's looking at a whorl which may have various shapes at
 9       the core because of distortion and hence you will not be
10       surprised that he's retaining this as a potential
11       candidate, although if we look at the straight angle of
12       the whorl it's not exactly the same as on the mark and
13       this is because of a distortion.  Now, the explanation
14       for the distortion should not come after I revealed the
15       print.  The explanation for the distortion should come
16       at the time of the examination of the mark, hence at the
17       analysis stage.  So this is the concept of tolerances
18       when it apply here in this example to the general flow.
19           By the same token we can extend the concept to
20       minutiae and I will discuss the concept of tolerances on
21       the features.  I will restrict the discussion to
22       minutiae because it's easier to show but there is no
23       loss of generality.  We can expand the discussion to
24       pores and edges of course without losing any generality.
25       Depending on the clarity of the mark, the robustness of
page  47
 1       the observation can vary and by robustness I mean
 2       something which is related to reliability.  If you look
 3       at a mark and you can see clearly a ridge ending, you
 4       will say that this is a robust feature.  It is a
 5       reliable feature because it can see it.  In a sense,
 6       that should the corresponding print be made available to
 7       me I expect that ridge ending to appear as a ridge
 8       ending on the print.  If by virtue of the quality of the
 9       image or distortion or slippage my judgment as to
10       whether I can really see a ridge ending being a ridge
11       ending is reduced, then I have higher tolerances I will
12       allow for something on the mark to appear a little
13       bit different than the ridge ending.  It can be a
14       bifurcation.  So my reliability is lower, the
15       reliability of the judgment I make on the type of
16       minutiae is lower.
17           Now, I will use two concepts for the reliability
18       which I hope it will not confuse the Inquiry.  I think
19       it's helpful to distinguish between presence of features
20       and type of features when we talk about minutiae.  When
21       you follow a ridge, you come up to an area where three
22       ridges reduce in the flow to two ridges so obviously
23       something happened in that movement of the ridge flow.
24       Hence we can postulate the presence of a minutiae
25       occurring at that stage, at that position.  I may not
page  48
 1       see whether that minutiae is of type ridge ending or of
 2       type bifurcation.  So what I suggest is to distinguish
 3       the two.  We distinguish between the mere occurrence of
 4       an event from the type this event will take being a
 5       ridge ending or a bifurcation.
 6           To make my further illustration I will use a system
 7       which we designed for training purposes.  It is called
 8       PiAnoS.  It has been designed with the purpose of doing
 9       collaborative exercises with students.  The difficulties
10       I experienced in the past when I was asking a group of
11       students to carry out comparison on a set of mark and
12       print was they were providing me as a feedback various
13       chartings of a various nature using non-standardised
14       tools to show what they want to show me and that led us
15       to develop an annotation system which will force
16       students to use the same type of annotation all
17       throughout the class and then I can review them and also
18       it is online, which is not the purpose here, but it
19       distinguish the annotations made during analysis from
20       the annotations made later during the comparison stage.
21       So here I'm going to use the analysis side of things.
22           The conventions we use are logos to designate
23       minutiae.  The triangle -- and this is just a convention
24       using PiAnoS -- the triangle is an event.  If I see
25       something happening but I am not sure whether it really
page  49
 1       will happen or not so there is some uncertainty in that
 2       judgment and it may well happen because of the quality
 3       of the mark, then the uncertainty is displayed by the
 4       broken lines.
 5           The more secure I am in my judgment as to the
 6       robustness of the event, there is something happening
 7       here, there is three ridge coming in and two going out,
 8       then the full lines will be used.
 9           This deals with the event.  Then of course we can
10       move even further and say, well, not only I can see an
11       event but I can guide as to whether it's a ridge ending
12       or a bifurcation.  So using the same convention, here
13       not only -- an event is observed and you will guide more
14       likely towards a bifurcation for the square or a ridge
15       ending for the round.  The next one is in all likelihood
16       a minutiae will be present and also in all likelihood
17       the type will be respectively a bifurcation, the square,
18       or a ridge ending here for the round.
19           So what we have here is an expression of the
20       tolerances.  The minutiae where in analysis phase would
21       be of low reliability means that we have much more
22       tolerances in the leeway we will accept should a print
23       be put forward as a potential source.  The more down we
24       are on that scale the lower the tolerances will be.  If
25       on the mark a feature is indicated as in all likelihood
page  50
 1       a ridge ending, you expect to observe it as a ridge
 2       ending on the corresponding print should it be made
 3       available.
 4           So there is an increase of reliability as we go down
 5       this hierarchy of event and type and also there is an
 6       increase in demonstrability.  In other words, it's
 7       easier to convey to a lay person the existence of
 8       minutiae which are down on that chart than to convey
 9       minutiae which are up on that chart, but a chart
10       captures something which is important is the reliability
11       on the feature higher as you go down compared to the
12       features on the top; in other words, the strength of the
13       relationships we may obtain from features declared as
14       reliable will be higher than the strength obtained from
15       features declared as less reliable event.
16           I've applied this to one mark and the little round,
17       this is my analysis of this mark, the round reflects
18       ridge ending and they are full so it means that in all
19       likelihood I expect to observe a ridge ending should a
20       print be made available to me.  The same for these two;
21       the same for this bifurcation here (indicated).
22           The dotted lines reflect that I expect a bifurcation
23       to be present here but I would allow for a ridge ending
24       to happen; in other words, I will not commit myself to a
25       bifurcation with certainty or in all likelihood but I
page  51
 1       would allow for a bifurcation to occur.  This is because
 2       of the lack of clarity of the shape of the ridges in
 3       these two ridges coming together.
 4           The same applied to these two formations; they are
 5       triangles.  It means that because of the flow of the
 6       ridges I would expect something to happen in between.
 7       It could be a short ridge but the quality of the image
 8       is so poor that I don't have a lot of reliability in
 9       these observations.
10           The same applies to this little ridge ending here.
11       The clarity of the mark in that region just forbid me to
12       have high confidence on the type of minutiae, although I
13       have quite high confidence on the event itself.  You can
14       do the same with ridges but this is not the purpose of
15       today's presentation.
16           When we say a feature is reliable I quite like the
17       concept that it should not be reliable to me but it
18       should be reliable among a set of examiners looking at
19       the same mark.  What you have here is all the
20       annotations by the students in the first year of masters
21       why they did examine independently, without having any
22       reference to the print, this mark and their annotations
23       reflect, if you see this one is quite a robust feature.
24       Every student said it's a ridge ending.  Whereas this
25       point here (indicated) has been seen only by three
page  52
 1       persons, me included, and the two others said, well,
 2       there is uncertainty as to the location and to the type
 3       of minutiae.
 4           The consensus is the validation of the concept of
 5       robustness of reliability.  In other words, don't
 6       believe me because I'm saying it is reliable.  The fact
 7       that when you offer this comparison to a range of
 8       examiners looking at it independently they will all
 9       assign high reliability to these features as opposed to
10       this one (indicated).
11           So these features will be with low tolerances, for
12       example, whereas this one will be with higher
13       tolerances.  When I will be embarking to a comparison
14       process, I would expect the potential corresponding
15       print to display the minutiae, ridge ending, ridge
16       ending, ridge ending in that direction and another one
17       as displayed here without allowing a lot of leeway, a
18       lot of tolerances.
19           When it comes to this feature here (indicated)
20       because it has been declared as having larger tolerances
21       during the analysis, then of course you will allow your
22       window for looking match to be broader a little bit and
23       you will allow variations to occur in this area, even on
24       a print which will indeed correspond to it.
25           That is the concept of reliability during the
page  53
 1       analysis and what I am suggesting here is that it is
 2       helpful to see this by degree and not only by black and
 3       white, I see something or I don't see something.
 4           That was a question which was asked --
 5   THE CHAIRMAN:  I wonder, we normally stop for 20 minutes and
 6       if it would be convenient --
 7   A.  I think it will be a very convenient time.
 8   THE CHAIRMAN:  We will do that now and sit again at 12.00.
 9   (11.42 am)
10                         (A short break)
11   (12.02 pm)
12   THE CHAIRMAN:  If you would like to resume,
13       Professor Champod.
14   THE WITNESS:  Thank you.
15           The point I was trying to make with this slide is
16       that from time to time it may happen that you need more
17       than one image of the same mark to obtain all the
18       information which is relevant.
19           On this image to the left, the section in green can
20       be seen and there is a possibility to use this bottom
21       section in your analysis.  The section in the white is
22       not useless but of course if you look at the image on
23       the right which has been taken under a different
24       illumination condition the image on the right is much
25       more clear for the ridges in the white section.  So the
page  54
 1       analysis can be formed from more than one image of the
 2       same mark using one illumination or one production for
 3       one part of the image and another production for the
 4       bottom part of the image here, respectively the white
 5       area, the ridges on the white area and the ridges in the
 6       green area.
 7           It happened quite often when the quality of the mark
 8       is limited that there is a need to obtain output maybe
 9       with different densities in order to help in some
10       sections of the mark which might be visible in one
11       output and other sections of the mark which will be
12       visible on another output.
13           Of course, with digital technology, laboratories
14       nowadays tends to merge these images together, digitally
15       speaking, in order to get one output which will show the
16       entire range of features but before the digital time it
17       was customary to offer, especially with difficult marks,
18       to offer an examiner with various output images at
19       different densities in order to help him to make
20       decision in difficult areas.  So the analysis can be
21       obtained or made on more than one image, should it be
22       required.
23           Now I would like to discuss about documentation and
24       I have taken some option in my report.  The level of
25       analysis I have shown you of course takes time and it
page  55
 1       can be seen as taking too much time for operation of
 2       services.  I always favoured somehow of a pragmatic
 3       approach to this.  By pragmatic it needs to be
 4       encapsulated into procedures but I have never been a fan
 5       of the all or nothing decisions.
 6           So if you take the mark on the left, it seems to me
 7       because the image is legible by itself, there is not a
 8       lot of room for interpretation.  It seems to be a waste
 9       of time to document fully that mark with all the
10       minutiae and all the ridges.  But when the mark is of
11       limited quality like this one (indicated), and it is a
12       mark which I am going to use a few times later, where
13       there is potentially signs of distortion, difficulties
14       to relate ridges one to the other, invisibility of
15       features which are hard to get from the image, number of
16       ridge features which are limited, then I think that mark
17       deserves a fuller documentation.  When I say
18       "documentation", it's documentation made at the time of
19       examination, during the analysis.  So the extent of
20       documentation has in my view to be adapted to the
21       quality of the mark and there is a lot of marks going
22       through identification bureaux which would not require
23       the level of documentation you have on the right but on
24       the other hand there is a series of marks of which the
25       quality is so that it would be deserved to have a fully
page  56
 1       documented analysis of the mark before and documented
 2       before it is moved into the comparison.
 3           From the analysis, there is some decisions and these
 4       are decisions about how the mark will be processed in
 5       the next step.  The analysis have laid out the
 6       tolerances which we discussed but in addition it is
 7       customary to make a decision as to what will be done
 8       with that mark in whether it will go into the comparison
 9       stage or not.
10           The first possibility as a conclusion from the
11       analysis is to say that the mark is of comparison value.
12       By this the examiner is saying that the mark has a
13       potential to exclude but will not be sufficient in
14       itself to individualise.  So there is merit in doing a
15       comparison because there is opportunity for exclusions
16       but the mark is not enough for individualisation.
17           I have included in the ones of comparison value
18       there is a subset of marks which will be declared of
19       identification value.  I tend to use identification and
20       individualisation as being equivalent in all this
21       discussion.  So these are the marks which of course are
22       of comparison value but in addition they will offer,
23       should the corresponding print be made available, they
24       will offer the potential to declare an identification.
25           There is in some services -- and it depends of
page  57
 1       course on the organisations -- marks which will be
 2       declared that they can be searched on an AFIS system, on
 3       an automatic fingerprint identification system.  Quite
 4       often the requirement for marks to be searched on an
 5       AFIS system is in relation to a minimum number of
 6       minutiae because AFIS system use minutiae as the basis
 7       of the template against which fingerprints are compared.
 8       The usual practice is to suggest eight minutiae as a
 9       minimum for the mark to be searched on AFIS.  So it's
10       another subset of the marks within marks declared of
11       identification value.
12           I noticed in conversation and also in publication
13       that there is quite often confusion about the meaning of
14       these three-tier decisions: comparison value,
15       identification value and the mark which offered the
16       potential to be searched against an automatic
17       fingerprint identification system.
18           Now we can move to comparison.  The comparison, it
19       is only at that stage that the print becomes available
20       to the examiner and I will simplify the argument by
21       saying that there is, the attention has focused on the
22       one particular print, that the others which may have
23       been put forward have been excluded.  The decision flow
24       are as follows: first, from time to time it might be
25       required to consider the quality of the print itself
page  58
 1       because sometimes the ink impression or LiveScan
 2       impressions are of such poor quality that the best thing
 3       is to resort for the demand for additional comparison
 4       material.
 5           If the material is considered to be sufficient for
 6       comparison, then the flow of examination goes from
 7       general to particular and we use the level 1, Level 2
 8       and Level 3 as a logical progression from looking first
 9       at the ridge flow, then moving, if the ridge flow is in
10       agreement then we can move down to the examination of
11       the Level 2 Detail and if the Level 2 Detail are in
12       agreement then it flows naturally to move to compare the
13       Level 3 details which are on the next slide.
14           The tolerances have been set during analysis so they
15       set the boundaries above which a disagreement will be
16       considered to be enough to exclude.  If there is
17       dissimilarity at each of these levels which are
18       considered to be significant, in a sense that they are
19       beyond the tolerances set during the analysis, then the
20       exclusion decision will be made.
21           Even though the chart may give the impression that
22       it is trivial to decide upon the exclusion, it is not
23       trivial at all and I will come back to this theme later.
24           The square here is an attempt to describe a process
25       whereby the comparison starts by identifying on the mark
page  59
 1       a set of features which will be used as a target.  It's
 2       a mechanism whereby examiners can register a set of
 3       characteristics which are easy to memorise and easy to
 4       search on the potential prints.
 5           This target is, if obtained and observed on the
 6       potential print, then new addition from the mark in
 7       terms of minutiae, generally going from the target to
 8       cover the whole surface, will be searched on the print
 9       so each transaction, so to speak, is an attempt to
10       falsify the proposition that the mark has not come from
11       the source at hand, so there is an iterative process
12       from the target until covering all the minutiae on the
13       mark to be looked on the print.
14           This slide deals with the Level 3 features and
15       that's exactly the same process.  I hope I had been
16       careful to use the term "from the mark to the print" all
17       the time.  The comparison process always starts from the
18       information which is unknown and should not start from
19       the material which is known.  So the features to be
20       searched against are the features of the unknown source,
21       the mark, to be looked and found potentially in
22       correspondence on the print.  The process always goes
23       from the mark to the print.  The features identified on
24       the mark have been identified during the analysis stage.
25           Again, it may happen and it happened to me as well,
page  60
 1       that in some cases once you have seen some observation
 2       on the print it helps to reconcile some observation made
 3       during the analysis and it happened that features on the
 4       mark then help to understand what is going on -- excuse
 5       me, I will rephrase that -- features on the print
 6       observed following the comparison help to understand the
 7       nature of the feature on the mark so you go reversely
 8       from the print to the mark.
 9           I think it will be too restrictive to suggest that
10       things should only go from analysis, mark to the print;
11       in other words, that only the features that had been
12       identified during analysis should be used in the
13       comparison against the mark that if features are
14       observed against the print -- if features are observed
15       on the print they should not be used in the
16       decision-making process because they hadn't been
17       identified during the analysis.
18           I think we have to allow from some circular
19       pathways.  I allow this on the basis that we have a
20       documented analysis which is stored separately as the
21       documentation associated with the comparison.  So if
22       there is some additional features which became obvious
23       only during the comparison with the print, these
24       additional features I would not invite to discard them
25       at all, but they should be annotated in a way that makes
page  61
 1       explicit that they were not identified during the
 2       analysis stage and it is only when we had the benefit of
 3       the print that they became rationalised and evident in
 4       the comparison process.  So the yardstick or the
 5       mechanism for quality assurance is, again, down to
 6       documentation but, again, if there is proper
 7       documentation I don't see any harm at benefiting from
 8       observation of the print.
 9           Now of course there is the extreme scenario where
10       there is no reliable information on the mark which
11       materialise only as reliable observation when the print
12       was considered, that this scenario, if it is properly
13       documented, there is always a possibility to go back and
14       identify what was identified and found during the
15       analysis stage and then open the discussion as to the
16       opportunity or relevancy of using features that has been
17       only identified during comparison.
18           So I take this case from analysis to comparison.
19       For the analysis stage we will take this mark and try to
20       identify the features of interest and these are the
21       annotations which I suggested in analysis stage.  This
22       is documented and stored separately.
23           From these features I can then move on to
24       comparison.  I think I don't need to discuss the
25       conclusion of the analysis because it is obvious that at
page  62
 1       some point a decision is made as to how this mark will
 2       be used.
 3           This is to illustrate again the concept of
 4       tolerances.  These are the markings made by the students
 5       on the very same mark.  These are mine (indicated) and
 6       these are my students' (indicated) and we did that
 7       completely independently.  If we see this minutiae was
 8       declared to be a ridge ending in all likelihood, whereas
 9       this one (indicated) was declared by me to be
10       potentially a ridge ending but I will allow for some
11       variation around the type of minutiae.  We can see that
12       all students identified it as a ridge ending and there
13       is more variation as to the exact nature of the feature
14       and positioning of these features because the quality of
15       the impression is lower.
16           I move now to the comparison with a potential print.
17       Here is the minutiae charting on the comparison to be
18       put against the mark.  The comparison is a factual
19       acquisition of data.  I haven't at that stage
20       interpreted the meaning of the correspondence.  I have
21       put in evidence here.
22           The evaluation stage will be the place where things
23       will be assessed as to their contribution to the
24       proposition that this print has the same donor as the
25       person who left -- the finger who left the mark and that
page  63
 1       leads me to the evaluation stage in the process of the
 2       ACE-V.
 3           The evaluation stage is the fundamental inferential
 4       step.  Traditionally, in the fingerprint area there is
 5       only three conclusions which may be derived from the
 6       evaluation.  One conclusion is the conclusion of
 7       individualisation, which I use also the term
 8       identification for it.  The second conclusion is a
 9       conclusion of exclusion.  The third conclusion is the
10       term used, and it depends sometimes between the bureaux,
11       the term used is inconclusive, meaning that there is no
12       opportunity either to individualise nor to exclude.
13           The profession has been using two different types of
14       approaches to evaluate their correspondence and I will
15       refer to the empirical approach and to the holistic
16       approach.
17           Before referring to these two approaches I felt it
18       would help me at least to declare how I see the relative
19       weighing of information that the comparison phase have
20       led us to.  I see evaluation as a relative weighing
21       between features which some are found in agreement and
22       some are observed differences made during the
23       comparison.  The relative weights is assigned by
24       informed judgment, mainly deriving from training and
25       experience.  So it quite helpful to me to view this as a
page  64
 1       scale where during the evaluation stage the examiner
 2       will put the weight in favour of the identification on
 3       features which are declared to be found in
 4       correspondence and he would put the weights associated
 5       to the perceived differences on the other side of the
 6       pen.
 7           Of course, if there is an observed difference which
 8       cannot be reconciled at all it will have a huge weight
 9       and then make tilting the balance in favour of the
10       exclusion immediately.  That will be the case if for
11       example I am comparing a loop pattern on the mark which
12       is compared against a print which is an arch.  The
13       weight will be so heavy on the exclusion side that
14       immediately the scale will tilt.  But there is all the
15       intermediate positions where we have to weigh
16       correspondences against differences as far as their
17       weight are concerned and that's how I like to see the
18       evaluation stage.  You will identify later that either
19       the holistic approach or the empirical approach are not
20       very transparent on this weighing aspect of things.
21           So if I focus on something which I have found in
22       agreement in that comparison, you have here this ridge
23       ending which was the robust feature identified and then
24       two ridge ending facing each other in this shape called
25       an island and another bifurcation which was identified
page  65
 1       on the print but during analysis I expressed some
 2       uncertainty as to its exact type.
 3           The weight associated to this or to other features
 4       on the same mark, we have other features like this
 5       combined minutiae, will be informed by my expert
 6       judgment and in that case I quite like the concept of
 7       using some statistical information I may have available.
 8       So I have not only my judgment as an examiner, which is
 9       not doing fingerprint examination on a daily basis but
10       through practice there is, of course, a judgment as to
11       the rarity of the features that developed but also
12       systematic studies may help me to assign the weights and
13       I know that an island is a rarer type of minutiae
14       compared to two distinct ridge endings or a bridge which
15       is this type, the one we have here (indicated) is a more
16       rarer type of combined minutiae as opposed to two
17       separate bifurcations.
18           So if I take this into account with the small
19       tolerances I identified during analysis I would say this
20       provides some heavy weight on the identification side.
21           There is also some differences which can be noticed.
22       The print shows a bifurcation and a ridge ending which
23       are quite close to each other, whereas during my
24       analysis I have indicated a potential ridge ending and a
25       potential ridge ending facing each other indeed but with
page  66
 1       quite a large distance between the two.  So obviously
 2       there is a difference.
 3           Now, because during the analysis I indicated that I
 4       need to allow for some tolerances to be applied, then I
 5       don't think the weight -- it provides some weight but
 6       the weight is not extreme towards the exclusion.  Had
 7       the mark been much clearer, well, I expect that it will
 8       match and according to my annotation two ridge ending
 9       facing each other like these ones (indicated) with such
10       a distance, I think the weight associated to the
11       difference would be much larger.
12           So there is a general rule in evaluation that
13       relates to the uncertainties discussed during analysis
14       and the general rule is that the weight is reduced as a
15       function of the increase of the tolerances or the
16       converse, the diminution, the loss of reliability.
17           If I take this configuration here on one mark
18       (indicated) and I take a second and in fact the mark has
19       been produced by print but let us consider it as a mark
20       and I take a second one which is an extract of one
21       configuration that I showed already, there is much more
22       uncertainty on these features (indicated) than on these
23       features and that was reflect during analysis I
24       indicated clearly that there is uncertainty as to the
25       exact nature of these features.  In other words, my
page  67
 1       tolerances are very limited for the mark 1, they are
 2       small because there is a lot of robustness in the
 3       observations, whereas my tolerances were much higher for
 4       the mark 2 and this is the same source.
 5           So when it comes to the weight, had I found a
 6       correspondence on mark 1, the weight would be much
 7       higher than the weight associated with mark 2, just
 8       because of the larger tolerances that has been
 9       identified during analysis.  So the decrease in
10       reliability during analysis lead to decrease of the
11       weight that this feature will have in the
12       decision-making process and the weight is -- the analogy
13       I have used is this two-side balance with weights moving
14       towards identification on one hand and weights moving
15       towards exclusion on the other.  So that was to set the
16       general principles which I feel are helpful to describe
17       the relative weighing of the features in a comparison.
18           I am back to slides which I have used a year ago.
19       The basic rules for individualisation, we tend to
20       distinguish between what has been named over the years,
21       the empirical approach, as opposed to the holistic
22       approach.
23           The empirical approach, the source I believe one of
24       the earlier source of using a fixed set number of
25       minutiae as a threshold towards identification above
page  68
 1       which identification will be declared can be traced back
 2       to a Frenchman called Edmond Locard.  In 1914 he wrote a
 3       paper in which he articulated a three-point rule which I
 4       call a tripartite rule.  Essentially, based on his
 5       practice, the practice of other identification bureaux,
 6       based also on the earlier statistical work that was done
 7       at the time and also based on his discovery two years
 8       before that that there is information in the pores,
 9       shapes of pores and edges, Locard brings in a rule in
10       three sections: one, there is this idea that above 12,
11       more than 12 minutiae, if present, clear, without
12       significant differences, then the identity is
13       established.  That's this 12-point rule section of
14       Locard's work which have been adopted by a lot of
15       identification bureaux.
16           Now, the irony of this is that the ruling or the
17       proposal of Locard was quite balanced and he tackled the
18       other options and said, well, between 8 and 12 then he
19       would expect an examiner to account also, to take into
20       account in his judgment, the rarity of the features
21       observed during the comparison.  He will use -- the use
22       of pores, shape of the edges, angles of a bifurcation,
23       specificity of the general pattern, will all be features
24       which Locard will invite the examiner to account for to
25       handle the situation between 8 and 12.  He is suggesting
page  69
 1       of course that is verified by another experienced
 2       specialist.
 3           The third section which has been lost in history is
 4       Locard mentioned that below 8 then the evidence will of
 5       course not be presented as an identification but may
 6       provide corroborative evidence in the form of a
 7       presumption of strength, proportional to the number of
 8       minutiae and their clarity.  This paper, to me,
 9       represents very well the basis for the empirical
10       approach.
11           If we just make a very -- there is no all countries
12       but a few countries around us, France is using 12, so
13       restricted their practice to the first step in the rule.
14       Italy is using 16 to 17 points.  Again, it's not 12 but
15       it is the first segment of Locard's rule.  The 16 to 17
16       is originating from a paper published in 1911 by a
17       French pathologist called Balthazard and it is because
18       of that paper that the Italian -- it's the only country
19       in Europe where I'm aware of there is some judicial
20       decision as to the number of points required for an
21       identification and it could be la Cour de Cessation, the
22       Cassazione, took the decision in the Italy that 16 to 17
23       is required for fingerprint evidence to be presented as
24       identification evidence in court.
25           The Netherlands, Holland, between 10 and 12 and
page  70
 1       there is different processes involved for 10 compared to
 2       the one for 12 or above.  Germany is 8 to 12.  So it's
 3       clearly the second section of Locard.  Spain 12, South
 4       America, by and large, around 12 and other countries
 5       have different numbers.
 6           This is the empirical approach.  I go back to the
 7       question which was asked at the introduction of this
 8       hearing.  There is no statistical argument to justify
 9       the use of a strict rule like 12 where each feature is
10       counted the same, 1, 2, 3, 4, 5 up to 12, with the same
11       weight.  Whereas we know that feature's weight may
12       depend on the type, their type and their position on the
13       fingerprint, hence if we look at the statistics involved
14       around point, and when I was talking about the empirical
15       approach it is with regard to minutiae of course, the
16       statistics of the distribution of minutiae and their
17       relative frequencies show that some features are much
18       rarer than others and I have used the example of a
19       bridge before.
20           In addition, it is a restrictive position to
21       consider only minutiae because part of the element which
22       are displayed on marks goes beyond minutiae, Level 3
23       features such as pore position, shape of the ridges, can
24       be used and taken into account in forming a view.  So
25       it's difficult to find a scientific justification for
page  71
 1       the fixed point rule.  It is a policy.  In some
 2       countries it is a policy to bring the evidence to court.
 3       Italy only based on a specific court ruling.  In most of
 4       the other countries, it is something that examiners have
 5       somehow imposed on themselves to do to provide evidence
 6       in court.
 7           The critical aspect of the fixed number of points
 8       rule is that when the quality of the mark is reduced
 9       like you have in this comparison, which I will name
10       comparison F, the variation you may observe between
11       examiners as to the number of minutiae found in
12       agreement between that mark and that print will become
13       quite large and the number of 12 or 16 was giving some
14       sort of an illusion of transparency because the court
15       may think that if it is 16 for one, it has to be 16 for
16       another examiner.
17           The comparison was part of a study made by Evett
18       & Williams in England and Wales and on that comparison a
19       little bit more than 120 examiners reported the number
20       of matching minutiae observed in comparison between the
21       mark and the print.  We can observe that on that
22       specific case one examiner has seen 14 and another
23       examiner has seen 56 points of comparison based on the
24       same material.  That was the trigger for the move in
25       England and Wales towards a non-numerical standard, was
page  72
 1       that if 16 had to mean something as a yardstick to
 2       quality, then we would expect that a series of experts
 3       will agree -- of course there is always variation -- but
 4       will agree to a larger degree than between 14 and 56.
 5       That collaborative study published by Evett & Williams
 6       was the trigger for reconsidering the 16-point standard
 7       in the UK.
 8           We did a very similar test with Swiss examiners and
 9       that very test was based on the same images as the Evett
10       & Williams.  The view shared by examiners at the time
11       was, "Well, maybe we don't have the same rules to count
12       minutiae in Switzerland and maybe we have much more
13       robustness in our decision when we say there is
14       12 points".  So we did a survey exactly along the same
15       lines as that was done in England and Wales and the
16       results were exactly along the same lines as well.  If
17       we take comparison F, the Swiss examiners the lower
18       number of minutiae matching between the mark and the
19       print was 11 and the larger was 33.  That collaborative
20       test set the starting point of change amongst Swiss
21       identification bureaux to move towards the adoption of a
22       non-numerical approach, the abandonment of the 12-point
23       rule which was governing Switzerland was triggered by
24       this collaborative exercise.  The resolution taken by
25       the heads of identification bureaux to suppress the
page  73
 1       12-point rule have been adopted in 2007; so almost ten
 2       years after the collaborative exercise.
 3           The 16-point standard in England and Wales, and I
 4       understood that Scotland was governed by the same
 5       standard as well, dates back to an interpretation of a
 6       paper by Bertillon, another Frenchman, in 1912.  It's
 7       also probably derived from the views of the statistical
 8       calculations that were published by this pathologist
 9       French pathologist, Victor Balthazard.
10           We know that since 1996, the examiners and ACPO
11       moved towards the abandonment of the 16-point standard.
12       This is a reproduction of the images that had been
13       published by Bertillon in 1912 showing allegedly
14       16 points of comparison matching between two prints from
15       different individuals.  For a small story, these images
16       probably were interpreted outside their context at the
17       time that the 16-point standard was adopted because the
18       texts by Bertillon is very clear to the intent.  He had
19       no intent whatsoever to mislead with these annotation
20       masking(sic) and drawings, he just wanted to highlight
21       the need during the fingerprint comparison not to focus
22       only on corresponding or closely corresponding features
23       but to consider systematically the absence of difference
24       as key factors in the decision process.
25           I had used these three slides with you, my Lord,
page  74
 1       when we met a year ago.  It is an extract of a decision
 2       from the Court of Appeal in this case against Buckley.
 3       This is a case where the judge reaffirmed the principle
 4       that there is possibility to admit evidence even though
 5       the 16-point standard was not reached.  It's quite
 6       interesting to see that the ruling is very much
 7       articulated around the same principle as the principle
 8       that Locard laid down in 1914.
 9           There is very good guidance which has been given as
10       to how the discretion will be exercised to accept or not
11       the evidence.  I don't feel the need to go through these
12       texts completely.  The one I would like to draw your
13       attention on is the last sentence of the (c) point:
14           "... the expert's opinion is not conclusive ... it
15       is for the jury to determine whether guilt is proved in
16       the light of all the evidence."
17           This to me has a very strong importance in the
18       matter that we are discussing today and I will relate
19       this later in a discussion about what do we mean by
20       "individualisation" and "100 per cent certainty" and
21       "conclusive".
22           The other approach adopted in numerous countries as
23       well is called the holistic approach and it is based on
24       the resolution we discussed this morning by the
25       International Association for Identification (or IAI).
page  75
 1       That resolution was reaffirmed some years later in 1995
 2       at a fingerprint conference in Israel.  The resolution
 3       suggests that:
 4           "... there is no scientific basis for requiring that
 5       a predetermined minimum number of friction ridge
 6       features must be present in two impressions in order to
 7       establish a positive identification."
 8           For the reasons I explained before, there is not a
 9       single weight that comes on the balance.  There is not
10       only one possible weight.  It will all depend on the
11       specificity of the features and the clarity of these
12       features.  Some features will have a lot of weight,
13       other features will have less weight when we are talking
14       about corresponding features and the same when we are
15       talking about features which are differences.  Some
16       discrepancies will be enough in themselves to rule out a
17       possibility of common source and other differences may
18       have some weight but not an enormous weight as to be
19       sufficient in itself to exclude.  So this resolution
20       just formalised somehow that the way I describe needs to
21       be made of the basis of the mark itself and we cannot
22       simply assign a single counting rule or a simple
23       weighting factor to each of these features up to some
24       sort of a magic number.
25           This is interim status of the Standardisation II
page  76
 1       Committee.  The organisation in their annual meeting in
 2       Tampa this year adopted a reformulation of this
 3       resolution which, in essence, gives the same message
 4       that there is no scientific basis to require a minimum
 5       amount of corresponding friction ridge detail.
 6           The holistic approach is a way of the quantitative
 7       versus the quality of the features that have been
 8       observed in correspondence, as opposed to a numerical
 9       standard where everything is focused on quantity which
10       is illustrated by this line in red.  If the quality of
11       the mark is reduced, then there is more of a need for a
12       quantity of matching features in agreement to conclude
13       to a common source but if the quality is increased, then
14       of course the quantity of features required to come to
15       the same conclusion is reduced.  This is the essence of
16       the holistic approach.
17           The two examiners I have illustrated here is that we
18       have to accept that we will not have all the same
19       abilities when it comes to make these informed judgments
20       and one examiner may have his bar positioned a little
21       bit differently than another.  Of course, in an ideal
22       world, we would expect that there is a consensus;
23       although we accept that differences exist, that there is
24       a mean consensus between examiners at the examination of
25       the said same mark.
page  77
 1           The evaluation phase ends with a conclusion which is
 2       constrained at the moment by another resolution of the
 3       IAI, which is the resolution from 1980.  The resolution
 4       indicates that there is an.
 5           "... opposition to any testimony or reporting of
 6       possible, probably or likely friction ridge
 7       identifications."
 8           In other words, the only conclusions which are
 9       regarded as authorised by the IAI are conclusion of
10       "identification", "exclusion" or "I cannot tell"
11       (inconclusive).
12           I expressed that this morning already.  In fact,
13       there is a continuum between exclusion and
14       identification in terms of strength of the relationship
15       between a mark and a print and I use this grey scale to
16       represent the continuum.  What the IAI resolution is
17       proposing is to restrict this continuum into three
18       classes: the identification, the exclusion and
19       everything in between under the vocab of inconclusive.
20       This is a policy decision, of course, to make.  I always
21       viewed it as a continuum and it's a policy decision to
22       restrict the conclusion that will be presented in court
23       to these three possibilities.  There is nothing from my
24       perspective that would prevent to explore the meaning of
25       an inconclusive should the judge want to explore this
page  78
 1       with the jury.
 2           The last stage of ACE-V is the verification step.
 3       The verification step is defined as the examination by
 4       another qualified examiner resulting to the same
 5       conclusion.  The objective of a verification stage, of
 6       course, is to have an independent review of the
 7       conclusion and also it has to follow the same ACE
 8       (analysis, comparison and evaluation) protocol that had
 9       been followed by the First Examiner.  In the fingerprint
10       area, the verification stage -- I cannot move from one
11       slide to the other any more ... ah, it's come back.
12       Thank you.
13           In most departments, it is used as somehow the
14       ultimate guard quality measure in place and it is very
15       important.  This stage needs to be fully documented and
16       again -- and I haven't seen this in a lot of bureaux I
17       visited -- there is a need for having procedures in
18       place to handle dissenting opinions following the
19       verification stage.
20           There had been discussion about blind verification
21       and, adopting the same pragmatic approach I discussed
22       when it comes to documentation of the analysis, it seems
23       to me that blind verification on all conclusions is not
24       cost-effective and I'm not sure it is needed.  But as
25       the same guidance I tried to give on documentation of
page  79
 1       the analysis stage, it needs to be adapted to the level
 2       of the complexity of the mark.
 3           I don't see any need if the mark is qualified as
 4       simple offering a range of features which would be
 5       undisputed among peers that has been put into
 6       correspondence with the print to the satisfaction of the
 7       First Examiner, I don't see the need for multiple blind
 8       verification as a quality assurance measure.  However,
 9       if the mark is of limited quality, if the mark will be
10       subject to -- not dispute, will be subject to various
11       interpretation in the analysis phase from one examiner
12       to the other, then this mark, because of the uncertainty
13       as to the reliability of the information gathered during
14       the analysis, will be a mark where the blind
15       verification will be of relevance to be applied as
16       opposed to a non-blind verification.
17           The distinction I am making here by blind and
18       non-blind verification, by "blind verification" it is
19       intended that the person will verify will have no
20       knowledge whatsoever on the previous conclusion of the
21       First Examiner and that previous conclusion could be
22       either identification or exclusion.  So it's not the
23       case where the verification is asked only in
24       identification output in the first place but the Second
25       Examiner has no knowledge about even the conclusion that
page  80
 1       has been derived by the first, whether it be direction
 2       or it being an identification.  Whereas in most
 3       identification bureaux I visited, sometimes the
 4       verification is done independently from the First
 5       Examiner but the checker is aware that his colleague has
 6       identified; by being "aware", we can say he's not blind.
 7       I've no problem considering this verification as being
 8       independent because we would expect that the Second
 9       Examiner will document the features he has used in an
10       independent manner from the First Examiner.  So, as far
11       as documentation is concerned, there is always the
12       possibility to come back to the documents established
13       by both persons and see and discuss, if needs be, the
14       features that have been used both in analysis and then
15       in comparison by these two individuals.
16           The fact that the Second Examiner knew about the
17       output from the evaluation of the First, it is not
18       something that worries me enormously.  However, when the
19       complexity of the mark is so reduced, so there is less
20       information to work with in the information-gathering
21       stage, then I think it is of importance to consider a
22       blind verification process.
23           I move now to the concept of individualisation and
24       to some degree to discuss about statistics which I
25       discussed thoroughly in my report.  I didn't feel it
page  81
 1       will be helpful to go into the details of the models and
 2       likelihood ratios and I tried to limit myself to the
 3       generic concepts.  The first argument I would like to
 4       share with you, my Lord, is the definition of
 5       individualisation and that relates to this concept of
 6       100 per cent certainty which you may have heard a lot.
 7           This extract is from a publication by two Canadians.
 8       They tried to express what is meant by
 9       "individualisation" and I have used "identification" to
10       be synonymous in my whole presentation.  What is meant
11       is that at some point they have gathered so many
12       features in agreement without significant differences
13       that the probability that these features will occur by
14       mere coincidence is considered to be so small that they
15       will discount it.
16           What is behind the word which I think relates very
17       much to my perception of the evaluation process is it is
18       a probability.  It is a probabilistic judgment.  At some
19       point the examiner is facing so much information in
20       agreement without any weight on the exclusion side that
21       is saying that the chance of all these features to
22       manifest themselves on the mark by pure coincidence, a
23       numeric consequence refer here, to the chance that
24       someone else will present the same features is so
25       reduced that he's happy to report that this probability
page  82
 1       is zero.
 2           The other element which I would like to draw
 3       attention on is when we say "I have individualised this
 4       person", most practitioners will refer to the earth
 5       population.  They have singulised this person in the
 6       light or in the context of the earth population.  In
 7       other words, no-one else on the planet can be the source
 8       of that mark but the person identified.  So I'm using
 9       the term "earth population paradigm".  The starting
10       population considered on a systematic basis in this area
11       of fingerprint identification is the earth as potential
12       donors.
13           The inference of identity of sources is essentially
14       probabilistic.  That's what Tuthill and George are
15       saying.  It's essentially inductive.  It's not a
16       deductive process.  It would be a deductive process if
17       we would agree that only these ten people can be
18       involved and we exclude each of them by observing his
19       fingerprints and we are left with only one who has the
20       matching correspondence with the mark; hence it will be
21       a deductive process to say "I have identified that
22       person".  But in the forensic scenarios examiners work
23       in, invoking the earth population it is inductive
24       because we cannot exclude by deduction every entity in
25       that population.  So the strength of their relationship
page  83
 1       is from the observation the effect to make some judgment
 2       as to the cause who left the mark and this moving from
 3       effect to cause is essentially inductive and not
 4       deductive.
 5           They do apply this and Fingerprint Examiners do
 6       apply probabilistic reasoning without knowing it,
 7       without declaring it or being aware that they are very
 8       quickly and aptly applying probabilistic zoning in their
 9       process.
10           What is happening when somebody is claiming
11       certainty has been aptly described by Dave Stoney in
12       small paper -- I'm giving the reference in 1991 -- and
13       he described this as somewhat of a leap of faith.  At
14       some point, the examiners have identified so many
15       corresponding features that he is becoming subjectively
16       certain that the chance of duplication is zero.  I am
17       illustrating this as a jump.  There is a jump at some
18       point from a state of very limited uncertainty at some
19       point there is a claim that I'm sure and sometimes it
20       has been used, the term has been used 100 per cent
21       certain or absolute certainty.
22           Again, this little jump from a probabilistic
23       position to certainty position escapes the science.  It
24       has been accepted by courts for many years but it
25       escapes a logical argumentation.  It is a leap of faith
page  84
 1       at some point.  It is a small leap but it's a leap of
 2       faith.  At some point, the chance are considered to be
 3       so reduced that, for ease of discussion, it will be
 4       declared that an identification has been achieved and
 5       the chance of finding someone else on earth with the
 6       same features will be declared to be zero hence the 100
 7       per cent certainty.
 8           Of course as a scientist I would never claim that
 9       the probability of an event is equal to zero unless have
10       been able to observe it empirically myself.  So there is
11       a leap of faith at some point in the way we produce the
12       conclusion and when we state with absolute certainty or
13       with 100 per cent certainty.
14   THE CHAIRMAN:  I think we are going to have to interrupt
15       your presentation for lunch.  We will sit again at 1.55.
16   (1.05 pm)
17                      (Luncheon Adjournment)
18   (1.57 pm)
19   THE CHAIRMAN:  I am sorry, we had to interrupt you again but
20       on you go, as soon as you are ready.
21   THE WITNESS:  Thank you, my Lord.
22           The purpose of this slide is to explain that the
23       identification process can be seen by analogy to a
24       reduction process.  So on the top of the triangle you
25       have all the individuals among which the person
page  85
 1       potentially of interest.  It will become of interest at
 2       the end, of course.
 3           By practice, Fingerprint Examiners will conceive
 4       this whole population as the world population, will take
 5       the world population as a starting point.  Now comes the
 6       scientific evidence here in the form of fingerprint
 7       evidence and this graph has its limits.  I will focus
 8       only in cases where there is supporting information in
 9       favour of the identification.  This graph is not
10       adequate to reflect on the exclusion side of the
11       balance.
12           The weight of evidence is shown here as this arrow
13       of distinct length.  The lengthier it is the stronger
14       the evidence can be viewed in favour of the identity of
15       sources.
16           The little blue dot, blue segment here (indicated)
17       reflects what I have called previously the leap of
18       faith.  The weight of evidence is huge in order to
19       reduce from the world population up to the single
20       individual in the bottom.  So you need a large amount of
21       weight of evidence to come to that position where the
22       examiner may step towards the identification decision.
23           This outline here is the usual framework in which
24       the Fingerprint Examiners will operate.  So the expected
25       weight of evidence to be produced in every fingerprint
page  86
 1       comparison cases, if we are talking towards identity, is
 2       of that magnitude, if I may use the length as an analogy
 3       to the weight of the evidence.
 4           Of course there is two elements in this.  One is the
 5       population and the second is the weight.  Had the
 6       circumstances been different, for example, that we
 7       don't -- and when I say "we", I would think about the
 8       court, won't use the world population as a starting
 9       point but would use potentially the population of a
10       city, the weight of the evidence which would be
11       necessary to come to the same situation as before is
12       reduced, but you come to the same conclusion.
13           So the argument I would like to make here is that we
14       need to distinguish two things in interpreting forensic
15       evidence.  There is weight and there is the putting the
16       weight into the perspective of the case.  The
17       perspective is given by this reduction.  Where do we
18       start?  If you start with the world population you need
19       that weight to come to that point (indicated).  If the
20       court had started with another more limited population
21       then you need a different weight to come to the same
22       ultimate decision in the end.  So we cannot ask for the
23       weight of the evidence to deal with the two elements.
24       That is the argument which was made by the Court of
25       Appeal in Doheny and Adams.  As I read it, the court
page  87
 1       ruling was to invite the scientist to concentrate on the
 2       weight of the evidence only and express the weight of
 3       the evidence without putting in perspective to any sized
 4       population, unless he is invited to do so by the judge.
 5           When we talk about weight of evidence, we try to
 6       remove it and treat it independently from the issue as
 7       to what is the contribution in the context of a case,
 8       putting this into a context of a case needs the
 9       fact-finder to establish the, what technically speaking
10       we call the prior odds to establish the starting point
11       as a function of the other element presented in the
12       case.
13           So the identification process can be seen as a
14       reduction process.  It's a reduction process obtained by
15       a factor.  The factor has the only effect to reduce the
16       number of potential sources down to a limited set.  It
17       is probabilistic and there comes a point where we will
18       claim -- and that's the jump -- that at the end of this
19       reduction process only remain in the group one
20       individual and that will be the individualisation.
21           I will talk about probabilistic models.  There is
22       more in my report.  I will focus mainly on minutiae and
23       I would like to discuss about the potential contribution
24       of these models into the global discussion about
25       fingerprint evidence.
page  88
 1           These models and the ones I am going to concentrate
 2       on invite the consideration of two specific questions.
 3       The first is an assessment of the probability of
 4       observing the degree of similarity and/or dissimilarity
 5       between the two impressions should they come from the
 6       same source and this is the first question that the
 7       model invites you to assess, statistically speaking.
 8           Again, we can view this as a balance.  To balance
 9       this, we need to consider the alternative probability
10       which is the probability of observing this degree of
11       similarity and/or dissimilarity if the two impressions
12       are originating from different sources.
13           Once both of these probabilities have been formed,
14       then we have an element that describes the weight of
15       evidence and technically speaking we refer to this
16       ratio, because we have a ratio between two
17       probabilities, we refer to this ratio in our forensic
18       jargon as the likelihood ratio.  If, by hypothesis, the
19       results of the examination are as likely if it is coming
20       from that source that it is coming from a different
21       source, in other words, that the probability associated
22       to question 1 is the same as the probability associated
23       to question 2, it simply means that the balance is even
24       and the forensic element does not help to progress the
25       case in one direction or the other.
page  89
 1           Conversely, if the findings explain themselves much
 2       better under the proposition of a same source, than
 3       under the proposition that a different person produced
 4       the mark, in other words, that the first probability is
 5       high and the second is small, then in ratio the finding
 6       will provide support for the proposition that the mark
 7       has been produced by that person.  We can make the same
 8       exercise reversely to support the exclusion.
 9           So the likelihood ratio, its magnitude translate the
10       reduction factor which I have shown before with length
11       of the arrow.  The likelihood ratio does not allow to
12       say either by absolute terms or by degree whether the
13       mark was left by that person.  It just allows to say
14       that regardless of the starting position it will make
15       you move into this triangle by a degree expressed by the
16       weight of evidence.
17           The models which I am going to discuss all aim at
18       assigning a likelihood ratio to features found in
19       correspondence between a mark and a print.  To answer
20       the first question, the model will have to encapsulate
21       knowledge about distortion because if we want to ask
22       ourself what is the chance of observing these features
23       if truly that person left the mark, we need to account
24       for the possibility that the mark that has been
25       deposited may be distorted, smudged, moved.  So behind
page  90
 1       the scene, the model will account for distortion and it
 2       has been informed with data distorted finger-marks that
 3       have been acquired under different distortion
 4       conditions.  That is what these images (indicated) are
 5       showing.  All these images are coming from the same
 6       finger but, depending on how the surface was touched, we
 7       expect to observe distortion in the relative or in the
 8       positioning, global positioning of the ridges, their
 9       shape and the positioning of the minutiae.
10           The model has learned how fingerprints distort due
11       to the application of a finger on a surface.  It is what
12       we call the within source variability, the model is
13       informed for the first question about the variability we
14       expect should that person truly have left the mark.  So
15       by multiple repetition by simulation of distorted
16       impression from that finger we can assess the chance of
17       truly observing what we have under our eyes, given that
18       it is the finger of interest which is at the source,
19       which is the first half of the likelihood ratio.
20           The second question is when we postulate that a
21       person of which we have kept the print for comparison is
22       not the source of the mark.  So in that case we ask
23       ourselves what is the chance of finding someone else who
24       by mere coincidence will display features which will not
25       be distinguished from the features observed in
page  91
 1       correspondence between the mark and the print.
 2           To answer the second question, we need a database of
 3       prints taken from different people and all these loops
 4       are from different people.  I have used eight here but
 5       obviously the model is informed by more than eight loops
 6       behind the scene.  It's from the statistical analysis of
 7       a distribution of minutiae on fingers coming from
 8       different sources then we can obtain a response,
 9       statistically speaking, to the second question.  So what
10       we do here is we invoke the concept of between sources
11       variability of marks and print.
12           The main actors in the development of models who
13       allow to compute likelihood ratios are to my knowledge
14       three.  The UK Forensic Science Service has developed a
15       model which I had the chance to be associated with in
16       its early stages and the model is, in my opinion, by far
17       the most extensively validated and is very close to
18       casework deployment.  The paper has been proposed to be
19       read at the UK Royal Statistical Society.
20           The laboratory in the Netherlands, the National
21       Forensic Laboratory (its acronym is NFI) has a research
22       group who explore the very same issue.  They are trying
23       to assess likelihood ratio associated with partial
24       fingerprint and our group has done extensive research in
25       the past and still today on the very same subject.
page  92
 1           All these three groups are aiming at the same
 2       objective: bringing a mechanism whereby we can assign a
 3       robust likelihood ratio to a comparison between two sets
 4       of features.
 5           The reason why I didn't feel it was fundamental to
 6       go into the technical details about the estimation of
 7       these numbers is because the input, the prerequisite for
 8       obtaining a meaningful statistical value out of any of
 9       these models, the prerequisite is that we should have
10       some sort of a consensus on the features which will fed
11       somehow the system.
12           These system are not automatically extracting
13       information from the mark or the print but they are
14       processing information which have been put forward as a
15       question to the system by a Fingerprint Examiner.  So
16       the prerequisite to obtain any meaningful number out of
17       a likelihood ratio calculator is that there is in
18       consensus and agreement as to the features which are
19       deemed to be relevant in the comparison at hand.
20           To illustrate this I will use this case which I have
21       used also in my report.  It is a mark which has been
22       detected on the blade of a knife with a chemical
23       technique called amido black and on the right you have
24       the potential corresponding print.  This is an
25       enlargement of a -- sorry, I've moved straight to the
page  93
 1       comparison stage and evaluation stage.  An examiner has
 2       indicated that he is able to annotate robustly all these
 3       features corresponding on the mark and on the print.
 4       The yellow ones are features that have been seen by the
 5       examiner only in the comparison stage.  They were not
 6       features which he or she indicated during analysis.  The
 7       reds are the ones which were annotated during analysis.
 8           When you compute the likelihood ratio associated
 9       with this comparison, putting in correspondence these
10       ten features with these ten features on the print, the
11       computer will give you back a likelihood ratio in the
12       order of 300,000.  This is very powerful evidence to
13       support the view that the mark has been left by the same
14       person as the person who produced the print.
15           Had a second examiner on the same case annotated
16       five minutiae in correspondence instead of ten, of
17       course the system computing the likelihood ratio on
18       these five minutiae will get a likelihood ratio which
19       will be much lower than the likelihood ratio you obtain
20       for ten minutiae and here the likelihood ratio is
21       estimated to be 5.  That provides some evidence but to a
22       degree, the size of the arrow of course is completely
23       different between the previous case and this case, the
24       same case but two examiners.  Hence the argument which I
25       suggest to the Inquiry is that the use of statistical
page  94
 1       modelling does not replace whatsoever the skill judgment
 2       of an examiner during analysis to agree on the sets of
 3       features which then may be used to obtain a statistical
 4       figure reflecting the strength of the evidence in the
 5       given association, if association is obtained.
 6           The system is not here to replace examiners.  The
 7       system may only offer help to guide their decision or to
 8       guide their informed judgment.  It brings a layer of
 9       systematic study to their decision-making process.  They
10       can have some back-up data to suggest how rare or
11       frequent a configuration of minutiae may be.
12           I would suggest that there is -- the use of
13       statistical models should not be seen as the holy grail
14       replacement of expert judgment in comparing
15       fingerprints.  It provides a layer of objectivity and
16       transparency but that has to work hand-in-hand within an
17       examination process which remains unchanged and is still
18       articulated along the steps of ACE-V, as we discussed.
19           So we are not talking about replacement.  We may
20       talk about calibration of informed judgments in the same
21       way I have used some data previously to express why I
22       felt that a bridge may contribute significantly into the
23       comparison process.
24           Another benefit which I have always suggested for
25       the use of these models is to tackle marks which fall
page  95
 1       within the inconclusive area and be able to guide
 2       judiciary as to the contribution of limited sets of
 3       features but still in agreement to how these features
 4       may help to support the proposition of common sources or
 5       otherwise.
 6           I just have two chapters before closing.  One is on
 7       Level 3 features and then I have some perspective to
 8       offer.
 9           The Level 3 features are referring to the intrinsic
10       small in terms of size details that either marks or
11       prints may display when left on surfaces.
12           To prepare the report which I submitted to the
13       Inquiry, I have benefited a lot from the work of one of
14       my PhD students.  His name is Alexande Anthonioz.  He is
15       currently doing his PhD project on the statistical
16       assessment of Level 3 features.  Whereas we have done a
17       lot of work in the community devoted to Level 2, there
18       is a paucity of data when it comes to Level 3.
19           One of the publications I referred to in my report
20       is a publication where we were interested to know if we
21       ask examiners, what do you mean by Level 3 features?  In
22       a survey we conducted a few years ago, it became obvious
23       that there is not a single definition about Level 3 and
24       examiners may sometimes put things which we would
25       classify as Level 2 as Level 3 and vice-versa.  Also
page  96
 1       what we noticed is that the perceived contribution of
 2       Level 3 features, in other words, where an examiner were
 3       asked about the expected strength of a given set of
 4       Level 3 features should a comparison give a
 5       correspondence, we noticed that the responses in terms
 6       of their judgment as to the strength varies a lot from
 7       one examiner to the other.
 8           The table you have on this slide shows you the four
 9       areas where there is commonality of agreement as to they
10       represent Level 3 features.  The ridge edges on the
11       first column, they are the specific forms and shapes
12       taken by the ridges as you follow them.
13           The pores, pores are the opening of the sweat canal
14       where a secretion is secreted -- sorry for repetition --
15       as eccrine glands produce the fluids.
16           The shape, as we have seen this morning, shape and
17       positioning is of interest.  Then there is the shape of
18       minutiae.  By shape of minutiae we mean the form a ridge
19       ending may take may be pointy, it may be bulky or also
20       the shapes that the adjacent ridges will form coming
21       close to the ridge ending here, we may have shapes in
22       forms of bottle necks or a ridge may approach -- the
23       right ridge may approach the ridge ending closely and
24       form a specific shape.  This is the shape of minutiae.
25       The angles of bifurcation falls also into that category.
page  97
 1       The last one is the width of ridges and furrows.
 2           To assess the strength of these features in the
 3       identification process one good starting point is to
 4       explore how these features are reproduced from one
 5       impression to the other.  The reproducibility is one of
 6       the key components in the robustness of the features and
 7       when we conducted experiments as to the reproducibility
 8       of these features from mark to mark, we noted that the
 9       only features which have a good level of
10       reproducibility -- by this term I mean that you can
11       expect to find these features in correspondence should a
12       corresponding print be made available -- the only
13       features which are well reproduced from mark to mark are
14       the relative pore positions and the specific shape of
15       minutiae.  The form of the ridge edges, the specific
16       forms of the pores are not well reproduced from one
17       impression to the other.
18           It may happen in the process of production from mark
19       from one person over and over again that mark 1 will
20       find the same shape of the pore than mark 120, but
21       although these events exist they are rare.  Most of the
22       time you won't see reproducibility.  So the lack of
23       reproducibility gave a good indication that these
24       features are not reliable in the identification process
25       for the decision-making.  Their contribution, their
page  98
 1       weight in my balance, exists but they are quite limited
 2       in terms of weight.
 3           It means also the converse.  Observing differences
 4       on these features because of a lack of reproducibility
 5       the weight on the exclusion side will be limited as
 6       well.  In fact, we could use a sort of guiding principle
 7       that if there is a feature which we will not consider
 8       for exclusion purposes it would be safe not to consider
 9       it for inclusion purposes.
10           One question which was put forward to us by the
11       Inquiry was can we identify on Level 3 alone?  We have
12       taken the position, which is in the same line as the
13       position of SWGFAST, we would not recommend to rely
14       solely on third level details to individualise.
15           First, the exercise is a little bit of a theoretical
16       exercise because most marks which will display Level 3
17       features, they will also display level 1 and Level 2
18       features and certainly Level 2 features.  So it's quite
19       difficult to envisage a situation where the only
20       reliable information on the mark is composed of Level 3
21       features as I defined it.  Quite often it walks hand in
22       hand with level 1 and Level 2.
23           The paucity of research as to the selectivity and
24       discriminative power of these features, the fact that
25       the only research available supports or provides
page  99
 1       evidence that reproducibility is not well in place for
 2       many of these features, both together lead us to
 3       consider that although they may contribute with some
 4       weight in the decision-making process, we would not
 5       regard this as good practice to rely solely on Third
 6       Level Detail to individualise.
 7           I will end my presentation with some perspectives.
 8       I've taken the liberty in discussion during the meeting
 9       last week just to highlight two or three elements which
10       was a striking feature for me being involved in this
11       Inquiry through the role you kindly allowed.
12           When I was faced with the task of trying to
13       understand or identify areas of agreement and areas of
14       disagreement, we aren't taking any position as to the
15       merit of the positions expressed.  The absence of
16       contemporaneous documentation and especially of the
17       analysis made the task meaningless or impossible.  So if
18       there is a first perspective I would offer is that in
19       case of disputed conclusions there is no way we can
20       unfold the chain of evidence without having proper
21       documentation of the analysis phase, otherwise
22       everything is based on post hoc justification of
23       conclusions that have been already taken and people have
24       already signed up to them and there is no way to unfold
25       the evidence to understand before making any evaluation,
page 100
 1       before making any comparison, what was the positioning
 2       as far as to the reliable features identified during
 3       analysis and how things developed from there.
 4           It seems to me that the report following the
 5       Mayfield bombing, the mark which was mis-attributed to
 6       Brandon Mayfield -- sorry, it was the Madrid bombing, I
 7       apologise.  The report was made possible because of the
 8       evidence available to the Inquiry team as to what was
 9       seen and searched through the AFIS before Brandon
10       Mayfield was suggested as a potential source and when
11       people went back to this information it was possible to
12       understand and identify what was misread, if I may say
13       so, during the analysis stage of that mark.  In the case
14       the Inquiry is focused on, we don't have this
15       information available.
16           That leads me to the next slide which I believe the
17       two perspectives I am offering can be made regardless of
18       the decision of this Inquiry in relation to the disputed
19       marks.
20           There is a need to maintain and having adherence to
21       standard operating procedures.  These procedures should
22       ensure a proper and separate undertaking and
23       documentation of the ACE-V process and it should also
24       include mechanism to handle conflicting results and
25       errors.
page 101
 1           I have been asked in the past to be a technical
 2       auditor for accreditation bodies under ISO17025 and I
 3       visited laboratories in Sweden and in Finland for their
 4       procedures in relation to fingerprint identification.
 5       These standards can be written in different ways but if
 6       the procedures are very clear about the separation
 7       between analysis and comparison, the level of
 8       documentation which are expected in both phases and how,
 9       if any conflicting opinion arise during that process,
10       either in verification stages or through other QA
11       mechanisms the laboratory has in place, if there is a
12       clear understanding of a process that have to be
13       followed by the laboratory to handle these events, then
14       I think we will make a lot of progress.
15           The last one is I reinforce something I have said
16       multiple times.  The marks of very high quality do not
17       require the same level of in-depth analysis -- and by
18       in-depth analysis I certainly mean the documentation
19       associated with it -- the same extensive and potentially
20       blind verification that a complex mark would require.
21           So there is a need to design a process whereby we
22       can distinguish the way simple marks are handled as
23       opposed to complex cases.
24           Every mark has to be classified somehow.  So if we
25       view this as some sort of a triage at the end of an
page 102
 1       assessment before a formal analysis comes into play, the
 2       view that has been taken by some Swiss identification
 3       bureaux is to use the 12-point rule as a benchmark to
 4       make this distinction between complex cases and simple
 5       cases.
 6           The above 12 clear identifiable features in analysis
 7       phase the mark will be declared to be simple and simple
 8       doesn't mean simplistic.  If below 12 or with signs of
 9       disturbance and difficulties or what practitioners term
10       often red flags in analysis stage the mark will be
11       considered as complex and will go through a different
12       route.
13           The level of training of the examiners who will be
14       allowed to verify a complex case is different in
15       Switzerland compared to the level of training an
16       examiner is required in order to be allowed to verify a
17       complex case.
18           The 12-point rule has been kept not as a decision
19       threshold after the evaluation but it has been kept as a
20       quality assurance mechanism with potential it's drawback
21       as well but offering a way to cope with the number of
22       marks submitted to a service and allow distinct pathways
23       handling easy cases in an economical manner without
24       running the risk of a misattribution and running cases
25       which are complex in a way which ensure that people
page 103
 1       appropriately trained to verify these cases are
 2       involved.
 3           Sir, that concludes my presentation.
 4   THE CHAIRMAN:  Thank you very much indeed.
 5           Mr Moynihan if you would like to continue.
 6              Examination by MR MOYNIHAN (continued)
 7   Q.  I'm very grateful to you, Professor.
 8           Needless to say, because the slides you prepared
 9       reflect the conversation we had at the end of last week,
10       I don't have too many questions because you have
11       incorporated much of the conversation in the
12       presentation.  There are, however, some points I would
13       like to discuss with you.  I can maybe begin by using
14       the ACE-V methodology just as a starting point.
15           First of all, I noted in an article you wrote in
16       particular with Mr Chamberlain -- but it may not have
17       been precisely that one -- that you do observe ACE-V as
18       an acronym is a fairly recent addition but as far as you
19       are concerned the ACE-V methodology is itself not new
20       and what you have written is that it would be wrong to
21       think of a pre-ACE-V fingerprint world and a post ACE-V
22       fingerprint world.  Is that fair?
23   A.  Yes.  ACE-V is an acronym that has been suggested would
24       translate good practice and good practice has been
25       around before ACE-V.  A good practice would say
page 104
 1       concentrate on the unknown, study the unknown first and
 2       then move to the comparison from general to particular,
 3       assess your findings only when you have finished with
 4       all of the comparison exercise and ask a second opinion,
 5       if needs be.  That, in essence, are the steps of ACE-V
 6       under this acronym, which are, to me, just reflect good
 7       forensic practice that had been around before the
 8       acronym itself.
 9   Q.  Also what I have observed that you have written is that
10       though ACE-V is, as you describe it, a protocol for
11       working, this acronym does not in fact tell us of the
12       standards that individual practitioners actually apply
13       in arriving at a decision of identity.  Again, is that
14       fair?
15   A.  That's fair, yes.
16   Q.  In that case, what I want to do it just step back from
17       the protocol and look at the practices which result in a
18       conclusion.
19           First of all, you have told us about events and
20       event seems to be the generic term which is used by
21       fingerprint practitioners.  Looking to paragraph 59,
22       ultimately, of your own paper for the Inquiry, which is
23       ED0003, paragraph 59, page 18, you had been looking at
24       statistical data and you said:
25           "These data are important to Fingerprint Examiners
page 105
 1       when they want to weigh different types of minutiae.
 2       Results from these surveys help to calibrate expert's
 3       judgment on the relative frequency of a given minutiae.
 4       The above data show also that a simple additive rule
 5       with equal weight for each minutiae -- up to 12 points
 6       for example as a numerical standard -- is an inadequate
 7       model from a statistical perspective."
 8           I want to look at that in a more simplistic way.  An
 9       event, as I understand it, a fingerprint practitioner
10       could ultimately decide an event could simply be a
11       coincidental feature on the underlying substrate and,
12       therefore, in fact be irrelevant to identity of the
13       mark.  That's possible?
14   A.  I just may need to clarify what we mean by event.  In
15       the presentation I did I have restricted the concept of
16       event to stops of ridges or stops in a bifurcation which
17       end into one ridge.  Then we may have uncertainty as to
18       the nature of the event being it a ridge ending or a
19       bifurcation but when I say I can see an event, in the
20       presentation of this morning I would qualify it as being
21       an interruption or stop for a bifurcation or the ridge
22       and not other features from the substrate which might be
23       confused with an event.
24   Q.  I am grateful because, as you say, the invitation to
25       clarify the terminology is exactly what I am coming to.
page 106
 1           By an event I am meaning something broader than what
 2       you were discussing yourself in the presentation.  By an
 3       event, what I mean is that there is an appearance in the
 4       image from the mark, so simply the broadest collective
 5       term, there is an appearance.  A fingerprint
 6       practitioner will have to, first of all, take a judgment
 7       whether that appearance is extrinsic to the mark, part
 8       of the background substrate or intrinsic.  That's the
 9       first?
10   A.  Yes, and that would be, in the little scale I showed, it
11       would be the passage from in all likelihood there is an
12       event or you think it might be something but it might be
13       something else, meaning it has nothing to do with the
14       mark of interest.
15   Q.  I am grateful because I wasn't actually myself going to
16       discuss that but what I wanted to do was to concentrate
17       on the intrinsic features and that would then bring me
18       to the same definition of event that you have been
19       using.
20           An intrinsic feature of the mark which is ambiguous,
21       it could be either a ridge ending or a bifurcation and
22       that is what you have been discussing just now today?
23   A.  Yes.
24   Q.  We have heard some evidence to suggest that in some
25       situations a practitioner might say what matters is that
page 107
 1       there is some event, either a ridge ending or a
 2       bifurcation, what matters is there is some event
 3       coincident between the mark and the print but the
 4       precise nature of that event is of secondary importance.
 5           Taking what you said in paragraph 59, would that
 6       form of reasoning be unduly simplistic because the
 7       precise nature of the event as either a ridge ending or
 8       a bifurcation, the distinction could be statistically
 9       very important, namely that a particular feature, a
10       ridge ending or a bifurcation, might be either rare or
11       common?
12   A.  It may depend on -- when examiner will say it's not
13       important it depends important for what?  If it is
14       something they will consider to be important to qualify
15       this as a significant difference or important to exclude
16       when the mark is of low quality of course the ability to
17       distinguish between a ridge ending and bifurcation is so
18       weak than a difference between a mark where you move
19       towards a ridge ending and a print towards bifurcation,
20       that difference might not be important in a sense.  That
21       difference alone will not dictate a conclusion of
22       exclusion.  So in a sense if that is what was meant by
23       important I would tend to agree.
24           If by important we mean once we have identified
25       corresponding features of clarity, so in terms of the
page 108
 1       reliability of the observation are in place and on one
 2       side we have, in a case we have a set of bifurcation and
 3       in another case we have a set of ridge ending, the
 4       strength of the information is different if we are
 5       talking about ridge ending or if we are talking about
 6       bifurcation.  So if by importance we mean the importance
 7       in contribution in my scale side of the balance towards
 8       identification, if the mark allows an assessment, an
 9       analysis, as to the nature of features that will have an
10       impact on the weight that will be assigned if
11       correspondence is found during the comparison stage.
12       That is done in analysis.  Did I say analysis first?
13   Q.  So if I understand it correctly, what you are indicating
14       is a correct characterisation of the event as a
15       bifurcation on the one hand or a ridge ending on the
16       other could have a material bearing on the true weight
17       to be applied to that event?
18   A.  Yes.
19   Q.  The second aspect of this that interests me, I think it
20       is covered by your answer but in a slightly different
21       angle, if an examiner is not sufficiently careful in
22       defining his event, he could in fact eliminate a
23       difference simply because he is not properly attuned to
24       the fact that truly it's a bifurcation in the mark, for
25       example, and a ridge ending in the print.
page 109
 1   A.  Yes, I think that's correct.
 2   Q.  I want, in following that through, to discuss with you
 3       then this question of tolerances.  If I understand it
 4       correctly, and this is the paradox for me in listening
 5       to what you have said, where the clarity of the mark is
 6       clear the tolerances that the examiner will set will be
 7       narrow and, therefore, before declaring an identity he
 8       will look for a very, very close correspondence between
 9       mark and print.  Is that correct?
10   A.  That's correct.
11   Q.  The paradox for me is that where the clarity is poor,
12       the examiner at the analysis stage may set a wide
13       tolerance, yes?
14   A.  Yes.
15   Q.  So that when he comes then to a comparison, he is again
16       applying wide tolerances to mark and print?
17   A.  Yes, absolutely.
18   Q.  That must run the risk that the chance of an
19       advantageous match between mark and print is increased
20       precisely because, the phrase you used is the window for
21       a match is wider?
22   A.  Yes, that's absolutely true.
23   Q.  How then ought a practitioner to guard against that
24       problem in a sense that, if I can put it at its most
25       extreme, the chances of finding a match in a complex
page 110
 1       mark are greater simply because the margin or window set
 2       for a match is wide to compensate for the lack of
 3       clarity, how do we restore that to a balance?
 4   A.  I think that's where the analysis is essential.  There
 5       is a translation in the documentation during analysis
 6       about the size of the windows: which are the features
 7       which are robust and which are the features with higher
 8       tolerances?
 9           Of course if all the features in analysis are
10       essentially of less reliability, so with large
11       tolerances, then you would need more of that information
12       in order to come to the same degree of weight compared
13       to a case where the tolerances will be smaller.  So,
14       yes, indeed there is a very -- the only way to guard is
15       to apply a strict distinct analysis from the comparison
16       in order that the features are declared to have large
17       tolerances upfront and not then a reduced tolerances
18       once you have found them in agreement on the print.
19   Q.  So it really very much depends on training practitioners
20       to be alert to the fact at the analysis stage they have
21       set themselves fairly broad tolerances they must be
22       conscious of when they come to the evaluation and they
23       weigh up the coincident features?
24   A.  Yes, and to me the verification stage should take
25       advantage of the documentation prepared during the
page 111
 1       examination of these two individual or three, if
 2       necessary, and the verification of complex marks should,
 3       in my view, not only be concerned with the agreement on
 4       the conclusion, but also having a technical review on
 5       the features used by these examiners both in analysis as
 6       in comparison.
 7           So it may highlight that in fact there are so many
 8       differences between the three people involved on the
 9       level of tolerances and reliability of the features that
10       will trigger a discussion as to the robustness of the
11       conclusion.
12   Q.  So, in fact, if you find across a range of
13       practitioners, where wide margins are being allowed,
14       inconsistencies in relation to the detail then you might
15       begin to wonder about the robustness of the overall
16       conclusion?
17   A.  To me that should trigger some quality assurance
18       mechanisms to see what is going on if these examiners
19       have been presented with the same information and there
20       is complete disagreement on the features that have been
21       used and their weighing in the mechanisms to come to
22       their conclusion.  Here of course we are talking about a
23       complex mark and if I may just add it's different if we
24       have a mark which is of such extent that one examiner
25       may use the upper part to come to the conclusion and the
page 112
 1       other examiner may use the bottom part to come to the
 2       conclusion, then just for sake of efficiency of course
 3       they haven't used the same sections to come to their
 4       conclusion but that's because of the richness of the
 5       mark as a starting point but the case I think we were
 6       referring to is a case where the mark is of such limited
 7       quality that it has to be assessed completely.
 8   Q.  The final significance that has occurred to me listening
 9       to you talking about tolerances is reinforcing the need
10       for blind verification.  If one is looking at a complex
11       mark with examiners setting wide tolerances, wide
12       windows, there therefore must be a correspondingly
13       greater risk that they would be influenced in their
14       judgment by knowing the conclusion formed by others.
15           Is that fair comment?
16   A.  Yes, I think so.
17   Q.  If I then move on to a slightly different angle in
18       relation to tolerances, I understand from what you said
19       so far in that the tolerances are themselves at present
20       a subjective judgment by practitioners?
21   A.  Yes.
22   Q.  You are yourself aware of the American Academy of
23       Science's report earlier on this year and of what it
24       recommended, it said very little specific in relation to
25       fingerprints but I did observe that it recommends that
page 113
 1       research be done in relation to the potential distortion
 2       of movement and the likes.
 3           Is that with a view to it setting parameters for
 4       tolerances?
 5   A.  The parameters for tolerances indeed derive from
 6       knowledge about how a mark will appear when left by a
 7       person and it is the knowledge about the various
 8       instances a given mark from a same source may appear,
 9       that allows you to set the tolerances.  So the National
10       Academy of Science report may have advised to have more
11       research on distortion and tolerances.
12   Q.  Do you think that that is a subject that requires more
13       study?
14   A.  I would say that all subject in relation to the
15       interpretation of fingerprints would require some study.
16       If you look at the number of structured research in this
17       area, there is quite a paucity of research so distortion
18       is certainly a subject that would need further research.
19   Q.  Perhaps to try to understand how we've arrived at a
20       situation where, as you say, there's a relative paucity
21       of research, if I understand it correctly, the
22       literature might suggest that because such a high
23       standard was set historically, the 16-point standard, in
24       the United Kingdom, it was understood that if you found
25       16 points in sequence and agreement there was a very
page 114
 1       high probability of a match and, therefore, the use of
 2       such an extraordinarily high standard had, in fact,
 3       removed the need for detailed research, that there was,
 4       in a sense, a rule of thumb that was found to be
 5       reliable, removing or reducing the need for research.
 6       Is that fair?
 7   A.  I'm not sure it is the 16-point standard that led to the
 8       paucity of research.  As soon as you develop this
 9       scientific endeavour allowing only conclusion of
10       certainty, either using a point standard or otherwise,
11       then the consequence of this is that research dealing
12       with uncertainty is not at the fore of the priorities
13       because in a world which is dominated by clear-cut
14       answers talking about reliability or visibility of
15       features and features which are more reliable than
16       others it means, like I did this morning, it means that
17       there is grey levels in what we see and assess.
18       Tolerances is also a concept which is probabilistic in
19       nature and then the end result of how we make the
20       decision of individualisation is also probabilistic.
21           So as soon as we decide that it will be a profession
22       ruled by certainty then the impact is that the research
23       arm is cut immediately because it has to be
24       probabilistic.  To me that's the reason why we have so
25       few statistical research in this area.  I'm not sure it
page 115
 1       is directly related to the 16-point standard.
 2   Q.  If I move then from the question of tolerances to
 3       evaluation under a non-numeric system, I looked with
 4       Mr Chamberlain, with whom you wrote a chapter for
 5       Professor Fraser's book, at something where you and
 6       Mr Chamberlain attempted to define the manner in which
 7       the individual practitioner would declare a match, the
 8       threshold they would set when they were looking for
 9       sufficiency of minutiae in agreement, so the question is
10       we always ask how do you determine sufficiency?  What
11       was written at page 69 of that article was this:
12           "In a nutshell, the individualisation will be
13       reached when the examiner observes a level of agreement
14       across the three levels of legible features that exceeds
15       the highest level of correspondence he observed through
16       his/her training and experience in comparisons involving
17       non-matching entities."
18           Read literally, what I myself was thinking from that
19       is that that would mean that any one particular
20       practitioner, if in the witness box, I would say, "What
21       is your personal standard", and he or she would be able
22       to give me one because they would be able to say, "The
23       highest number of points I have observed in sequence and
24       agreement in non-matching entities is X", whatever X may
25       be, "and therefore I personally look for X plus one
page 116
 1       before I would declare an identity".
 2           In fact, there has been a consistency in approach of
 3       all the practitioners asked this question, on whichever
 4       side of the debate they are in relation to our
 5       fingerprints, and they all say they have no personal
 6       threshold.  It seems that the judgment is one that is
 7       dictated by the particular print in front of them.
 8           Can you explain why it is that they have no
 9       consistent personal threshold applied to all matches?
10   A.  A key component of the paragraph you mentioned is that I
11       think we said something across all levels of features.
12       The information is not restricted to and only to
13       minutiae.  So when we say across all levels of features
14       it encapsulates the concept that depending on the
15       quality you may require less Level 2 features to come to
16       the same level of confidence or sufficiency that if you
17       don't have the quality in terms of clarity of image you
18       would need more.  So because of this qualitative and
19       quantitative assessment it's very difficult to define
20       the fixed(?) number required.
21           That being said, the argument which we suggested in
22       that paper is an argument I believe was suggested to me
23       both by David Ashbaugh in discussions we had together
24       and it is also through the same line of reasoning that
25       firearm examiners will testify to the level of agreement
page 117
 1       in terms of striation between a questioned bullet and a
 2       reference bullet.
 3           There is an accountability when the examiners say
 4       this is the highest or I never have seen such level of
 5       agreement between two, a mark and a print coming from a
 6       different source.  By accountability I would expect that
 7       examiner to keep a record of the best non-known
 8       non-matches he has experienced in his career and being
 9       in a position to say, "Here is the case I am talking
10       about today where I had that amount of feature which I
11       can present and show and this is, in terms of agreement,
12       much higher to the best known matches I have a chance to
13       see in my career", and then he could even -- of course
14       it never happened that way but I just projecting myself
15       in a state where we are not at the moment but we can
16       even think that this collection of close non-matching
17       can be put forward and say these are the best
18       relationship we managed to have.
19           Now, by relationship I would see not only counting
20       points, all the points are important, but across all the
21       spectrum of features which may be used in the
22       identification process.  So in the case at hand during
23       analysis the examiner had said, "I cannot see any Level
24       3 features.  There is a lot of uncertainty as to the
25       clarity of the ridges and the shapes, et cetera, at the
page 118
 1       end of the day everything boils down to the minutiae",
 2       then I would expect that report for you that will be
 3       structured according to this will be cases where, in
 4       similar situation, in terms of quality, what would be
 5       the maximum observed level of agreement obtained on the
 6       known non-matches.  That's the essence of the sentence
 7       we proposed with Paul Chamberlain to project ourselves
 8       in a state where it's not only an opinion, it's an
 9       opinion which is based on systematic acquisition of
10       cases of known origin in that case, difference of origin
11       which can be proposed and put forward.
12   Q.  Just to be clear about this, in an ideal world one would
13       like to have some objective standard for evidence but,
14       if I understand you correctly, you accept, subject to
15       the statistical cross-check, you accept that
16       identification of fingerprints is, in fact, an
17       inherently subjective exercise?
18   A.  If by subjective exercise we mean that it is -- it
19       implies skilled knowledge and informed judgment by
20       examiners which have been trained to do this work I
21       think, yes, it is subjective exercise.
22           If by subjective exercise we mean arbitrary exercise
23       I think, no, it's not it's not arbitrary even though it
24       is, there is an element of subjective judgment while
25       looking at a mark there also I would expect some
page 119
 1       transparency by the expert to be able to explain why a
 2       feature is more robust and why other features are less
 3       robust.  So it's not subjective in the sense of being
 4       completely arbitrary.  It is within a framework of
 5       training, competency and protocol that this subjective
 6       activity is undertaken.
 7   Q.  Two follow-on points really from that.  First of all,
 8       given the subjective element to it, is that in fact
 9       where the necessity comes in relation to verification,
10       namely it's the fact of the consistent conclusion from a
11       number of different practitioners that the reliability
12       of the conclusion emerges and is reinforced?
13   A.  Yes, I think the consensus is the key to the process.
14   Q.  Just while I go on in this same theme, so far as all
15       your work is concerned, are you yourself satisfied that
16       on the whole the conventional fingerprint evidence does
17       indeed produce reliable results?
18   A.  Again, this is my personal belief.  Although we don't
19       know what is the black number associated with this area
20       of expertise, it is my strong belief that the comparison
21       between the mark and the print can provide, provided the
22       quality is present, extremely powerful evidence to guide
23       towards identity of sources.  So I think the black
24       number, which we don't know, I believe that number is
25       very small.
page 120
 1   Q.  Indeed, would it follow that your own belief would be
 2       that if procedures are properly followed the rate of
 3       error ought to be quite small?
 4   A.  Yes.  I mean, the rate of error is extremely small.
 5       Absolutely.
 6   Q.  Accordingly, while one can see in your writings that you
 7       say that fingerprint practitioners should not claim
 8       100 per cent certainty, you are not as a result
 9       inferring at all that the conclusions that they arrive
10       at are unreliable?
11   A.  I hope that I managed to show the little subtlety
12       between claiming 100 certainty and claiming extremely
13       strong evidence.  There is a subtle difference.  If by
14       claiming individualisation the expert is providing
15       extremely strong evidence towards identity of sources as
16       opposed to different sources, then I would agree.  I
17       have no disagreement with that judgment.
18   Q.  So in fact what it really comes to is what you are
19       advocating is that the expert should more accurately
20       represent the statistical significance of their
21       conclusion rather than claiming something that seems to
22       be, in fact, unjustifiable 100 per cent certainty?
23   A.  I think it's probably something for the judiciary to
24       consider what are the best ways to present such a
25       difference in court without overstating its power.  The
page 121
 1       difficulty I have with the 100 per cent certainty is
 2       that it gives or it may give, contrary to what the Court
 3       of Appeal said, a jury the impression that the evidence
 4       alone is sufficient to balance the case in one direction
 5       regardless of the other evidence at hand.  Whereas I
 6       would rather see forensic science contribute
 7       significantly in the judicial process but not having
 8       cases being where the forensic evidence had such a
 9       weight that no other evidence available may move the
10       case in the other direction.  The 100 per cent certainty
11       gives me this flavour of factual finding which I always
12       find may mislead the jury; whereas a concept of
13       providing extremely strong evidence is very powerful,
14       but allows for contradictory evidence to potentially
15       point in another direction.  So that's why I have
16       difficulties with absolutes presented that way in court.
17           But again I would invite, especially because of the
18       long tradition of presenting fingerprint evidence in
19       court that way, I think it should invite discussion with
20       judiciary about the proper ways and mechanisms to
21       present this to a jury.
22   Q.  Again, to use a word I have used earlier on, the paradox
23       here is you are not suggesting that fingerprint evidence
24       is prone to error.  Rather, what you are suggesting is
25       that its true potential weight should be properly
page 122
 1       presented to the jury to weigh up against all other
 2       sources of evidence in a case.
 3   A.  Yes, I think that's fair.
 4   Q.  If I can then use an article, another article that I
 5       used earlier on this week, an article written by an
 6       American author Mnookin, who seems to be somewhat of a
 7       critic of fingerprint evidence, poses the question if
 8       100 per cent certainty is unobtainable what actually
 9       replaces it?  Because, in particular with the IAI
10       forbidding degrees of probability one would rather fear
11       that one would leave a Fingerprint Expert without what
12       Mr Pugh described yesterday as an explanatory model, one
13       in other words leaves him without a toolkit to explain
14       to the jury how persuasive the identification might be.
15           Do you see that as a risk?
16   A.  I think -- I don't know exactly what the position of the
17       Standardisation 2 Committee will be on the 1980
18       resolution.  But I have taken publicly the position that
19       claiming 100 per cent certainty or individualisation in
20       the face of the earth population was a claim which was
21       difficult to substantiate through a scientific process.
22       Hence, I would rather to see the evidence expressed
23       differently by conveying the strength of support the
24       comparative features provide in favour of a view of a
25       common source.
page 123
 1           This level of support does not alleviate the
 2       contribution of the element.  Stepping back from the
 3       100 per cent certainty does not leave me in the middle
 4       of nowhere because the evidence, provided there is
 5       associated elements which are strong, will still guide
 6       very strongly in favour of the right source hence being
 7       perfectly useful, I believe, in court proceedings
 8       exactly in the same way DNA evidence is used nowadays.
 9           Of course I can see the argument from
10       Professor Mnookin that if you look at today's status of
11       the IAI 1980 resolution and if all of a sudden courts
12       potentially in the United States tend to refuse
13       testimony as to 100 per cent certainty I can see the
14       argument saying where you lie now with this situation?
15       But as far as the strength of the evidence is concerned
16       I don't think it is changing anything.  A strong case
17       will remain a strong case.  The difference is I'm
18       expressing it differently than using the term
19       100 per cent certainty.
20   Q.  Again, to use a phrase that Mr Pugh used yesterday, if
21       Fingerprint Examiners today don't have a scale by which
22       to measure their conclusion, because, as you say, they
23       have not been accustomed to thinking in terms of
24       probability, how can they demonstrate to a jury the
25       strength of the particular judgment in a particular
page 124
 1       case?
 2   A.  I think the difficulties we may envisage here are no
 3       different from the difficulties an expert has to justify
 4       an individualisation conclusion today.  The fact that we
 5       are talking about a different manner of expressing the
 6       weight of the evidence does not change anything and,
 7       indeed, explaining how an individualisation was reached
 8       or explaining how I would conclude to extremely strong
 9       support in favour of the view that the mark is coming
10       from that person it has to rely on the same mechanisms.
11       So I do not see any major -- although there is training
12       issue but I do not see any major scientific changes
13       while we are talking about the scale, where there is
14       more than one point on the scale or if we are talking
15       about a scale where there is only one point on the
16       scale, if we have to argue about the scale that's the
17       same difficulty anyway.
18   Q.  If I follow it through -- sir, I am just noticing the
19       time -- if I follow this through you have pointed to
20       some statistical work on the incidence of minutiae so
21       that will assist to enable Fingerprint Examiners to
22       calibrate their conclusion.  That is one approach, yes?
23   A.  The numbers associated with each different types of
24       minutiae are good to calibrate their judgment not on the
25       overall case but with bits and pieces in the comparison.
page 125
 1           The likelihood ratio models are much better to
 2       assess the contribution of the whole configuration
 3       because they will take account for distortion and they
 4       will take account for all of the features at once so
 5       they are much better guidance as to the contribution
 6       compared to the single values associated with each types
 7       of minutiae because we cannot really combine them
 8       easily.  So it's good to calibrate our judgment, how
 9       rare is that type of minutiae compared to the other but
10       to get the overall story they are not ideal and I would
11       rather use statistical models with likelihood ratios.
12   MR MOYNIHAN:  Sir, I have actually noticed the time.  I'm
13       sorry, I've overstepped it.
14   THE CHAIRMAN:  No, what I thought, we obviously want to
15       complete this witness' evidence today, so we will maybe
16       take a break when you reach your conclusion.
17   MR MOYNIHAN:  I shouldn't be too much longer, Professor.
18           What I, in fact, wanted to do was ask you one other
19       point in relation to the likelihood ratio and then to
20       look at two points in relation to documentation, so I
21       shouldn't have too long.
22           So far as the likelihood ratio is concerned, I
23       understand what you said earlier was that you primarily
24       see the likelihood ratio as complementing the
25       traditional fingerprint evidence given by fingerprint
page 126
 1       practitioners and not something that would displace it?
 2   A.  Yes.
 3   Q.  That has two dimensions to it.  Firstly, the likelihood
 4       ratio is itself entirely dependent on the accuracy of
 5       observation of the fingerprint practitioner?
 6   A.  Absolutely.
 7   Q.  The second point that I wanted to ask you about, there
 8       has been some limited discussion of an alternative use
 9       of probabilities, which is to deal with very low
10       incidence of minutiae, for example, small portions of
11       fingerprint where traditionally a fingerprint
12       practitioner would not see enough reliable
13       characteristics to express an opinion.
14           Do you see the probability ratios or likelihood
15       ratios having a function there or not?
16   A.  Yes, I do.  I think the likelihood ratio may help to
17       guide as to the strength of this comparison which now
18       will be declared to be inconclusive.
19   Q.  How do you respond then to the proposition that if one
20       is dealing with a very small portion of a print there
21       may be a coincidental high incidence of match of
22       features there where if one went just slightly off
23       screen to the part of the print that is not deposited
24       one would find a significant number of differences?
25   A.  I'm sorry that was a little bit too quick for me.  Can
page 127
 1       you repeat the question, please?
 2   Q.  I will start again.  How do you deal with the
 3       proposition -- let us take a mark that if one had, say,
 4       50 per cent of the mark one would see in the inner part,
 5       near the core, a number of points of identity come
 6       slightly out from the core to the rest of the 50
 7       per cent one would see some killer points of difference.
 8       Then by chance what is deposited at the crime scene is a
 9       very small portion, namely the part that is nearest the
10       core with the similarities and the suspect,
11       unfortunately for him, has just not left the part with
12       the differences, the likelihood ratio might correspond,
13       it might give potentially on this view a misleading
14       likelihood of a match between the suspect print and the
15       mark; whereas a larger portion would, in fact, show an
16       inconsistency.
17           How do you respond to that potential source of, in
18       fact, misleading evidence?
19   A.  Well, the likelihood ration itself capture that there is
20       a possibility for coincidental match, if I may say so.
21       So the number itself captured that there is a
22       possibility, quite small, potentially, depending on the
23       magnitude of the ratio, that someone else will display
24       the features in the same positions.
25           Now, when you say by chance the defendant in your
page 128
 1       case had left only a surface which shows the matching
 2       features and not the area which shows the discrepancies,
 3       this probability in itself is very small.  So, yes,
 4       indeed, any matching element expressed with a strength
 5       of evidence expressed probabilistically imply almost by
 6       definition that we may face the coincident -- a random
 7       coincidence but with magnitude over ratio we'll express
 8       how often that may occur and if it is a rare occurrence
 9       then the likelihood ratio would be much higher than it
10       is in everyday occurrence that you can see this level of
11       coincidence by pure chance.  So you'd capture this.  In
12       essence, the likelihood ratio just gives you the
13       strength of evidence pointing in one direction without
14       telling it that that is the truth.  It guides you by
15       degree in favour of one proposition versus another.
16   Q.  If I come then to documentation, I have just a number of
17       points to ask you about documentation.  First of all,
18       principally what you are saying is that you favour the
19       need for documentation of reasoning both at the
20       assessment and verification stage for complex marks.
21       You have mentioned in Switzerland that a standard of 12
22       is used and also that consideration is given to rotation
23       and distortion.
24           Would you have in mind suggesting as a
25       recommendation a fixed definition of complexity or is
page 129
 1       that something that itself requires more careful study?
 2   A.  At the moment, the concept of what is a complex mark is
 3       not sorted in the specialised literature.  The Swiss
 4       attitude to this has been -- was not driven by a
 5       fundamental research project but was driven by the
 6       willingness also to move towards a change keeping some
 7       old practice in place which was driven by 12 and we felt
 8       that was a good transition scheme so that is why 12 had
 9       been retained as this selection or triage measure.  But
10       certainly there is room for research to define what is a
11       complex mark as opposed to a simple case.
12   Q.  I've asked you earlier on today, we have heard some
13       evidence of some statistical work having been done by
14       Pincante and Jane(?) to suggest that the 12 points that
15       one might trace back to Locard might by coincidence in
16       fact on modern statistics be a reliable indicator as to
17       the identity.
18           Do you have a comment on the reliability of the
19       statistics that would point in favour of 12 as being the
20       key number here?
21   A.  I do not recall the publication completely.  If my
22       memory serves me well, they showed through a statistical
23       model that the chance of finding on another person a set
24       of features when you have 12 minutiae in correspondence,
25       that probability is a very, very small number, something
page 130
 1       like 10 to the power of 16, 17 or 18 minus, hence the
 2       argument that 12 is safe.
 3           The view I have on this is that it is not because a
 4       model is able to calculate teeny-teeny numbers but these
 5       numbers have a realistic meaning and I like the approach
 6       what was adopted with DNA evidence in this country where
 7       at some point it has been suggested that even though you
 8       have a full DNA profile the match probability that will
 9       be quoted associated with that will be in the order of 1
10       in a billion.  Even though the mathematics of computing
11       the (inaudible) frequencies together by multiplication
12       leads you to 10 to a power of minus 15 or 16, the
13       argument was that if you want to be reliable in these
14       teeny-teeny-teeny numbers, the constraint on the
15       experiments you need to set up in terms of peer-wise
16       comparison is enormous.  With DNA, the argument was made
17       that at the current stage of knowledge we cannot
18       substantiate match probabilities below the order of
19       magnitude of 1 in a billion and I quite like the
20       approach and adopting exactly the same approach for
21       fingerprint evidence.
22           So even though by multiplying factors together like
23       the paper you referred to, you can obtain extremely
24       small figures which might lead to the impression that
25       there is no need to discuss any more when we talk about
page 131
 1       ten to minus 20.  These numbers has to be -- we have to
 2       ask ourselves how robust they are and I don't think
 3       these numbers, when they are so small, are robust so I
 4       would be tempted to fall back into a position of order
 5       of magnitude.
 6           Now, the paper of course demonstrates that the more
 7       minutiae, the more information you have, the larger the
 8       likelihood ratio will be which, in a sense, is
 9       absolutely correct and logical but I don't see the paper
10       as demonstrating that 12 points is safe for
11       individualisation.
12   Q.  If I try to understand that in a lay way, are you
13       indicating there is a problem about reliability of
14       mathematical modelling when it comes down to relatively
15       low samples of data?
16   A.  Yes.
17   Q.  If I then carry on and ask you two other questions in
18       relation to documentation, first of all, I observed in
19       one of your slides -- you may want to bring up again
20       slide 32 from your presentation today -- it just happens
21       to be one of a number -- I don't have too long to go.
22       This is the penultimate question.
23           I observe that what you have done is not just marked
24       the feature by the codes you have indicated by
25       triangles, circles and squares you have also taken the
page 132
 1       trouble to trace in what you understand to be the
 2       underlying ridge flow.  Is that tracing of ridge flow
 3       something you personally do as a matter of routine?
 4   A.  When documentation is required due to the quality like
 5       this one, I think tracing the ridges and sometimes even
 6       the valleys is extremely useful to understand the
 7       configuration.
 8           A mark left by friction ridge skin is a series of
 9       ridges and the ridges can be traced, so to me that's, in
10       fact, in the case like this one I start with ridge
11       tracing and overlaying then the minutiae afterwards.  So
12       the minutiae are just positioned as a logical
13       consequence of ending or opening of ridges so, yes, I
14       believe it is critical.
15   Q.  Is that what you teach?
16   A.  Yes.
17   Q.  In a sense, I think you have probably just answered my
18       conclusion question to this.
19           Do you find that it is in fact the tracing of the
20       ridges as much as anything that you attach weight to
21       rather than simply, as we have seen in some of the
22       tracings we have here, where practitioners just put a
23       dot on a precise point in an image?
24   A.  Yes, to me the information from the mark is expressed by
25       ridges in sequence and not expressed by points, like
page 133
 1       dots on the surface.
 2   Q.  The final question -- and I do promise it is the final
 3       question -- is a concept I have used with a number
 4       of witnesses called demonstrability, by which I mean
 5       plainly it requires skill, training and expertise to be
 6       able to observe features in a mark and to interpret them
 7       but when once an expert has arrived at a conclusion, has
 8       observed features and has interpreted them, should that
 9       expert be able to demonstrate the existence of the
10       feature to a lay person such as me rather than the
11       expert being able to say that my eye can see things that
12       no lay person can see?
13   A.  As long as we both have eyes with decent aptitude I
14       think that if I guide you to my observation you should
15       be able to see what I see.  The difference between a lay
16       person and someone with some knowledge in this area is
17       how they will interpret the findings but observing the
18       features themselves I think that, provided the
19       technology is available, that the demonstration should
20       be possible to a lay person.
21   MR MOYNIHAN:  Thank you, Professor.  I must say thank you to
22       everyone else for their patience this afternoon.  That
23       concludes my questions.
24   THE CHAIRMAN:  What I suggest is if we start again at 3.40
25       and then if we cannot finish by 4.30 then I think we may
page 134
 1       have to make new travel arrangements but I hope that it
 2       will be possible to finish if we can but I don't want
 3       anybody to feel kept out.  So we will sit again at 3.40.
 4   (3.34 pm)
 5                         (A short break)
 6   (3.40 pm)
 7   THE CHAIRMAN:  We will begin at the back for a change.
 8           Ms Jones, you have an application, have you?
 9   MS JONES:  Yes, sir, I have questions about the differences
10       between complex and simple marks really and how they
11       relate to documentation in particular.
12   THE CHAIRMAN:  Yes.
13                    Cross-examined by MS JONES
14   Q.  Professor Champod, you spoke about what you thought were
15       the different requirements for documentation depending
16       upon whether something was a simple or a complex mark
17       and I think you recognised that it is not clear how to
18       differentiate the two.  You mentioned the 12 points
19       example from I think was it Switzerland.
20           Can I ask you first are there any other examples or
21       guidelines so far as you're aware that are of assistance
22       in any other jurisdiction?
23   A.  No, I'm not aware of any other guideline of that nature.
24   Q.  Can you tell me -- I don't know if you know -- does
25       Switzerland work on the basis of the non-numeric system
page 135
 1       or the numeric system?
 2   A.  No, Switzerland since 2007 has adopted a holistic
 3       approach without any numerical standard.
 4   Q.  I wondered if you had a view as to how a numeric
 5       approach to one aspect of the process sat with a
 6       non-numeric approach to the holistic process.  I don't
 7       know if you have a view on that.
 8   A.  Yes, the pragmatic position taken by the Swiss
 9       identification bureaux is to use the 12-point as a
10       quality measure which can be then audited.  In other
11       words, during the audit of a bureau you may revisit
12       decisions that have been made according to this 12-point
13       rule to see if truly what has been classified as complex
14       needed to be treated as complex and what has been
15       classified as simple needed to be treated as simple.
16           This number is used as a quality assurance measure
17       as a yardstick in the processes, not in the
18       decision-making when it comes to evaluate the
19       corresponding features.
20   Q.  I appreciated that point but one matter that I wondered
21       about was that during the course of the Inquiry there
22       has been some discussion about the problems or issues
23       with the numeric system, whereby there appears to be at
24       least a suggestion that a numeric system may encourage
25       examiners to tease out points and I wondered whether
page 136
 1       that was something that had been analysed in the Swiss
 2       situation?
 3   A.  The 12-point rule I referred to applies to the analysis
 4       stage without any reference to the print.  The concept
 5       of teasing out points has been always referred to in
 6       relation to the addition the print may provide in terms
 7       of information when you suddenly can increase the number
 8       of minutiae seen on the mark because of the benefit of
 9       the print.
10           Here the 12-point rule is used in analysis only
11       without any reference to the known print as a
12       decision-making process and has to be documented
13       accordingly in the case file.
14   Q.  I wondered, though, whether it could be applied to the
15       12-point rule on the basis that examiners may be
16       encouraged or perhaps somehow persuaded to find
17       12 points where if, say, they count 11 and if they found
18       one point or event, then it means a more simple and
19       straightforward process will be followed.
20           I wonder whether that was something that there had
21       been any analysis of?
22   A.  Not that I'm aware of.  I think that's part of the
23       quality system to ensure there are mechanisms to avoid
24       the situation where, in other words, to make your life
25       easier that you push things up to 12.  I think the fact
page 137
 1       that life is complex should not force us to shy away
 2       from complexity and the management system of the
 3       laboratory should find ways to reconcile this.
 4   Q.  I wonder then whether that perhaps brings us on to
 5       another point I wanted to ask you about, whether there
 6       had been any research or analysis of the resource
 7       implications of the differing requirements for
 8       documentation because, I think, all of that perhaps
 9       could be said to link up but it may be a very
10       hard-pressed and overworked examiner may be more
11       concerned to deal with simple marks rather than complex
12       marks because of the workload and I wonder whether there
13       had been any studies or research into the actual
14       resource implications for notes.
15   A.  I'm not aware of any research that has been published on
16       that subject.
17   Q.  In your own investigations, I think you said that you
18       had been to a number of different bureaux.  Was that
19       something you were interested in looking at when you
20       were looking at the processes of different bureaux?
21   A.  Excuse me, could you repeat the question?
22   Q.  Sure.  I think you said particularly -- perhaps if I
23       link this into another matter that I was going to ask
24       you about.  You said in relation to blind verification
25       that that wouldn't be appropriate in all circumstances
page 138
 1       because of the cost implications relating to that and I
 2       wondered whether that was based on some kind of
 3       empirical evidence or was that really more of a
 4       commonsense approach to the issue?
 5   A.  No, that was based on a commonsense approach.
 6   Q.  Thank you.  I wonder then if you are able -- it may be
 7       you are not able -- to elaborate on a more practical
 8       basis on what type of documentation it is that you think
 9       may be appropriate either in simple or complex cases
10       what level of documentation or recording is it that you
11       have in mind.
12   A.  If we take the simple case, the key element to
13       documentation is having a legible image of a marked
14       image of the mark of interest which of course stays in
15       the case file and constitutes the primary documentation
16       of the case.  Some laboratories have developed pro forma
17       sheets for analysis which invite the examiner to assess
18       the various questions which I raised during my
19       presentations as far as the substrate is concerned,
20       identification of red flags, identification of issues
21       with distortion and that sheet can be quite limited with
22       check boxes to identify if there is issues recognised.
23       If none of the red flags are flagged up I think a good
24       indication about the amount of information available,
25       and again this is the screening process, would be to
page 139
 1       indicate the minutiae which have been observed on the
 2       mark.
 3           We may debate, if we have a 50 -- and I'm using 50
 4       as an example -- if we have a palm mark with 50 minutiae
 5       is there any need on the legible image to indicate point
 6       by point the minutiae of interest?  I think having
 7       written information at the time of examination by the
 8       examiner just saying, "I'm assessing this mark as having
 9       roughly 50 minutiae worth of pursuing for comparison and
10       hence this is a simple case", if I may put it that way,
11       I would have no problem with that personally.
12           The complex side of the house, the PiAnoS I showed
13       you with the annotation of ridge flow and minutiae with
14       degrees of reliability during this annotation process,
15       we designed it with a view of applying this to complex
16       cases and I think it's quite a good compromise if the
17       tool can be made, can be user-friendly.  It's quite
18       quick to annotate convincingly marks of limited quality
19       that way.
20   MS JONES:  Thank you very much.
21   THE CHAIRMAN:  Miss Grahame?
22   MISS GRAHAME:  No, thank you.
23   THE CHAIRMAN:  Mr Holmes?
24   MR HOLMES:  No, thank you, sir.
25   THE CHAIRMAN:  Mr Smith?
page 140
 1   MR SMITH:  Thank you, sir, there are just two matters I
 2       would like to ask about.  One relates to the question of
 3       probabilities.  I hope to be fairly brief on that --
 4       largely because I don't understand it myself -- and the
 5       second question relates to the issue of distortion.  I
 6       hope I can deal with that fairly briefly too.
 7   THE CHAIRMAN:  Two pretty broad subjects but --
 8   MR SMITH:  They are indeed.
 9   THE CHAIRMAN:  -- if you can deal with them promptly and
10       shortly, as I am sure you will, please do.
11                    Cross-examined by MR SMITH
12   Q.  Professor, I wonder if you can try and help me with
13       something.  You described earlier in your evidence the
14       issue of I think it was the Baysian Theory on questions
15       of probabilities and how that had been largely
16       criticised, I think, in at least one court case.  I
17       wonder if we can just stand back and give me some help
18       with the way that probability sometimes operates.
19           I think in simple terms -- and I do want to keep it
20       very simple -- is the Baysian theory a theory that tries
21       to introduce some degree of almost subjectivity into the
22       question of probabilities?  Is that a fair way of
23       summarising.
24   A.  Bayes Theorum is a standard theorem applied to
25       probability theory, probability as being either
page 141
 1       subjective or objective.  So it's not restricted to a
 2       Baysian or a subjective view of probabilities.
 3           That being said, the school of thought in terms of
 4       the philosophy of statistics called the Baysian view is
 5       indeed related to the concept of subjective
 6       probabilities.  The probability is expressed as a degree
 7       of belief which has been proposed in the literature in
 8       the fifties to complement what we know from either
 9       modelling or the frequentist approach to statistics
10       which is looking at occurrence of events in the long
11       run.  There are some events in life which occur only
12       once also or some events which cannot be put into the
13       context of repeated experiments.  Hence, in the fifties
14       the development of a school of thought using a measure
15       of uncertainty called subjective probabilities which
16       measured the degree of belief and it has been associated
17       with the Baysian Theory of probability theory.
18           When I have used the term Bayes Theorum, when I have
19       used likelihood ratio, it is in relation to the classic
20       definition of Bayes Theorum as in the theory of
21       probability without exposing the subjective probability
22       definition for all probabilities I discussed.
23   Q.  Just really by way of example to try and talk through a
24       real example and maybe you can confirm to me or correct
25       me if I get it wrong, if I was to ask a statistician,
page 142
 1       "What's the probability of me living for the next year?"
 2       then the statistician could then look at life tables and
 3       say, "Out of the entire population, X per cent of them
 4       will die within the next year" and that gives you your
 5       probability.
 6           Then if I say, "Well, actually I am aged 90" then
 7       that increases the probability of me dying over the next
 8       year.  So you have further information about the
 9       position and I think you can work out a formula pitching
10       these two factors against one another and come up with a
11       probability of my death in the next year based on that
12       information.
13           Can I apply this to the facts, as I understand them,
14       of the Shirley McKie case.  If you take it from me, when
15       I refer to the "crime", I'm not talking of the crime of
16       murder, I'm talking about the crime of allegedly leaving
17       a fingerprint at a particular place where the individual
18       should not be.
19           In trying to assess the probability, just without
20       looking at the fingerprint, the probability of that
21       fingerprint being left by Shirley McKie, one of the
22       factors that would be relevant is there was no
23       independent evidence of, as it were, another crime.  The
24       crime was leaving the fingerprint.  Do you follow the
25       distinction?
page 143
 1   A.  I think so.
 2   Q.  One of the factors is a lack of independent evidence
 3       that Shirley McKie was in the locus.  Do you follow?
 4       There's not someone who says, "I saw her there, I saw
 5       her near that area."  So there's nothing of that kind.
 6       Another factor would be there's no or there is in
 7       evidence that there was a limited number of people who
 8       were actually in the area.
 9           Would you agree, just generally speaking, all of
10       these factors when you're just asking the question of
11       probability markedly reduces the probability just of a
12       random fingerprint being left by someone else on another
13       occasion, would you agree that tends to point to a
14       reduction of probability if one looks it as simply a
15       probability exercise?
16   A.  If I may just try to put my own analogy to your
17       suggestion, if by the specific circumstances you
18       described you're telling me that instead of starting
19       from the earth population we are starting from a limited
20       set of individual, then yes, that's absolutely right and
21       indeed this is a probabilistic judgment.
22   Q.  Because just to say looking at probability in simplistic
23       terms you would have to proceed on an assumption almost
24       that any person in the world could have been in that
25       house at that particular time which clearly is a
page 144
 1       ridiculous starting point, isn't it, in real terms?  It
 2       may be statistically correct but in reality it's not?
 3   A.  Well, it's a matter for the court to assess what is the
 4       reasonable size of the initial population given the
 5       circumstances at hand; so I cannot say it's ridiculous.
 6       It depends on the position of the court.
 7   Q.  Now can I turn on to the other matter that I wanted to
 8       ask you about.  It is the question of distortion and
 9       movement.  I think you touched on this in your
10       presentation earlier, which I may say for myself I found
11       very helpful to identify the method of analysis under
12       ACE-V.
13           I am not sure how closely you have been following
14       the evidence before this Inquiry, but one of the regular
15       themes that has been coming up when apparent differences
16       between an inked mark and an unknown print have been
17       pointed out is the response by those who say it is a
18       match is to say, "Well, this is due to distortion or
19       movement of some kind: full stop".  No greater
20       explanation as to whether it was twisting, smudging,
21       double tap, overlay, anything of that kind.
22           Do you have any comment about just saying, well,
23       there is a difference but it must be due to movement or
24       distortion?  Can you comment on that in general terms,
25       if you can, as to a realistic explanation for
page 145
 1       differences between an inked mark and an unknown print?
 2   A.  Well, when an expert invokes physical movement of a
 3       finger to explain a difference, perceived difference,
 4       between a mark and a print, I would rather see some
 5       empirical evidence that what is suggested is a
 6       possibility.  The mere fact of invoking an explanation
 7       to me is not sufficient to make that explanation right
 8       and when I say "right" with a probability of 1: it
 9       happened that way.
10           I think as a matter of transparency I would advise
11       that if specific movement is invoked and there is
12       evidence that indeed through in controlled experiments
13       we can reproduce that that indeed is a possible
14       explanation, which might be quite rare but it's
15       something we've managed to demonstrate.  Otherwise I
16       have difficulties because we can always invoke
17       explanations.
18   Q.  Can I ask you, professor, just on a slightly different
19       point but related if you perhaps haven't worked this
20       out, I'm a lawyer and I'm imagining being in a court
21       situation.  A Fingerprint Expert comes in under the
22       non-numeric standard and I have to cross-examine that
23       expert and I say to him, "Why are you sure that this is
24       a match between the two?"  He will no doubt explain,
25       "Well, with my experience, I can see sufficient points
page 146
 1       of similarities, the detail is clear enough and my
 2       understanding and experience dictate that these two
 3       marks, the latent print and unknown mark, have common
 4       authorship".
 5           Now I would be right, I think, in saying that the
 6       exercise of that analysis is still based upon minutiae,
 7       ridge endings, et cetera, et cetera.  I am right so far
 8       about that, aren't I?
 9   A.  It may not be solely restricted to minutiae in terms of
10       the features that the examiner has used.
11   Q.  So it may depend on, for example, the quality of Third
12       Level Detail, for example.  Is that what you mean?
13   A.  If they are legible on the mark, yes.
14   Q.  Nonetheless, if I am re-examining and saying, "Well,
15       let's go one stage deeper than that rather than just
16       your experience.  What is it about these two
17       fingerprints that lead you to the conclusion that they
18       are of common authorship?" and from your evidence I
19       think you say he should be able to demonstrate that, he
20       should be able to say, "Well, look, there are a number
21       of ridge endings, a number of bifurcations, that to me
22       indicate this comes from the same author but it's a
23       futile exercise to count them up .I'm not going to say
24       there's 10, 11, 16, 18.  That doesn't matter" but he
25       must still be able to demonstrate them under the
page 147
 1       non-numeric standard.
 2           Is that your evidence?
 3   A.  Yes.
 4   Q.  I take it when you make reference to the fact that the
 5       type of minutiae may be important because I suppose, and
 6       I am not sure if there are any statistics on this, but
 7       one particular feature with another may be much more
 8       improbable than another set of features.  So, for
 9       example, if you have an upward opening bifurcation
10       immediately adjacent to a downward opening bifurcation
11       that is probably a very unusual feature.
12           Without asking you if there are studies on it, is
13       that something you would agree broadly might be unusual?
14   A.  Absolutely, yes.
15   Q.  Whereas if you have two upward opening bifurcations with
16       one straight ridge in between, that's relatively common
17       and it wouldn't give any great cause for excitement but
18       you have a difference in the positioning between two
19       features just because one is very different to the other
20       and that would be fair, wouldn't it?
21   A.  Yes.  Different types of minutiae may bear -- depending
22       where they are on the ridge pattern, they would provide
23       different weights indeed.
24   Q.  But I suppose to make that an important difference
25       between the two sets of data, an up and a down
page 148
 1       bifurcation as compared to two upward bifurcations, one
 2       would really need to have some kind of research, or at
 3       least experience, to say an upward and a downward
 4       adjacent to each other are unusual.  Again, is that not
 5       a valid observation?
 6   A.  The way you can approach the problem is either having
 7       access to some empirical data -- counted how many times
 8       I have that sort of configuration to two bifurcation
 9       downwards, for example -- and that will calibrate your
10       judgment as to the rarity of that event.  That's one
11       way.
12           Another way -- and that's the way most examiners
13       have been trained to do -- would be to say that through
14       the years of experience and the number of fingerprints
15       they have looked at in their career, they built up some
16       sort of a mental database about the rarity of these
17       features.
18           I would not say that one is better than the other
19       but certainly the systematic acquisition of data is more
20       transparent than the invoking experience.
21   Q.  Give me just one moment, please.  (Pause)
22           Finally, I wonder if you could help me with this.
23       I'm interested in the question of error rates in science
24       and, indeed, in fingerprinting.  As far as science is
25       concerned, generally speaking again, taking your
page 149
 1       particular area, what is the perceived wisdom about
 2       error rates?  Is science supposed to be error-free, if I
 3       can put it that way?
 4   A.  If we talk about a scientific approach to the search for
 5       identity of sources in forensic science, to put it that
 6       way, there is of course, and by definition, an error
 7       rate in that operation and it will be foolish to claim
 8       that because it's scientific, it's error-free.
 9   Q.  Can I compare that to where fingerprints fit in, in your
10       view, with regards to science as pure science or whether
11       it's largely a question of opinion.  Where do you fit in
12       into the spectrum?
13   A.  You mean with regard to fingerprint evidence?
14   Q.  Yes.
15   A.  I think we cannot make such a clear-cut distinction
16       between pure science versus art.  I consider fingerprint
17       examination as a scientific process.  The protocol is
18       typical of a scientific process.  The mechanisms that
19       are put in place for quality assurance are the same kind
20       of mechanisms you put in place in scientific research.
21       Indeed, at some point in the process there is informed
22       judgment that comes into play, but I would not value
23       less this judgment than the judgment of a scientist to
24       make decisions when he is doing some chemistry
25       experiments.  That's part of the intelligent approach to
page 150
 1       any experiment to build up on the scientist's experience
 2       to guide in the development of his approach.
 3           I won't see the world be divided between pure
 4       science approach to fingerprints versus blackheart(?)
 5       approach to fingerprints.  If you ask me to position
 6       myself on the spectrum as I think fingerprint comparison
 7       is part of forensic science and the term science is not
 8       usurped .so yes, it is a scientific process.
 9   MR SMITH:  Thank you very much, professor.
10   MR MOYNIHAN:  I think to everyone's relief I have no further
11       questions.  Thank you.
12   THE CHAIRMAN:  I have only two matters which I am sure it is
13       my fault for not getting it right.  On slide 34 there
14       was a reference to length and terms of ridge unit and I
15       think I read orientation and length in terms of ridge
16       unit.  I wasn't quite clear what that means.
17   A.  My Lord, indeed, I haven't defined it at all.  The
18       concept of ridge unit is something that was introduced
19       by Dave Ashbaugh as trying to describe the smallest
20       element along the ridge we can define.  So if you view
21       ridges as a sequence of units, each unit having a pore
22       on the top, you could view a long ridge as being a
23       series of ridge units attached one after the other.  Of
24       course, that tells the distance -- if these ridge units
25       aren't connected by two ridge endings, this distance is
page 151
 1       fixed because you expect -- the blueprint is fixed on
 2       the dermis.  So if you have eight pores from two ridge
 3       endings, you expect that distance to remain permanent
 4       and do not change.
 5           The length of the ridges are something that needs to
 6       be accounted for in the comparison process, hence my
 7       reply to previous questions when it comes to following
 8       the ridges.  Following the ridges allows us to assess
 9       the length of the ridges involved and to check that the
10       number of ridge units are compatible between the mark
11       and the print.
12   THE CHAIRMAN:  Thank you.  The other point I wanted to ask
13       you about we have heard reference before to a bridge but
14       what is your definition of a bridge?
15   A.  The bridge I presented in my presentation is when you
16       have two bifurcations connected by their branches.  So
17       the form it takes, it is either a Z if you follow it or
18       an S if it is the other way round.  This connection by
19       two bifurcations is often described as a bridge.
20   THE CHAIRMAN:  It is just to be sure we are talking about
21       the same thing.  Than you very much for a very
22       comprehensive presentation.  I apologise for you having
23       to do it twice but it was important that what I heard
24       should be heard in public and I think that has been
25       covered now.  Thank you for the really large amount of
page 152
 1       assistance that you have given to the Inquiry.  We can
 2       release you, I hope.
 3                      (The witness withdrew)
 4           So far as tomorrow is concerned, I understand, Ms
 5       Jones, you will begin tomorrow morning.
 6   MS JONES:  I think either myself or Mr McPherson were going
 7       to go first depending on what time we started at.
 8   THE CHAIRMAN:  At 10.00.
 9   MS JONES:  I think both of us are going to be very short.
10   THE CHAIRMAN:  10.00 or as soon thereafter as you are
11       available we will begin.  Then I gather that you can
12       agree between yourselves as to what order you wish to go
13       in.
14           That, I think, concludes the evidence.
15   MR MOYNIHAN:  Sir, that concludes the evidence.  What I
16       propose to do is to circulate tomorrow morning a table
17       which is being prepared which will list all of the
18       witnesses.  It will specify those who have provided a
19       statement only and also those who have provided both
20       oral testimony and a statement.  That will be made
21       available because I have attempted over time to sweep up
22       but I may have made some omissions.  Tomorrow I hope to
23       be able to issue to all concerned a comprehensive list
24       of the witnesses, the evidence from whom is available to
25       you in arriving at your conclusion.
page 153
 1           Yes, that is all the evidence, bar I suppose to be
 2       absolutely clear I have not yet had a definitive
 3       statement from Mr Russell in connection with the e-mail
 4       that we received some weeks ago about Mr Swann.  We have
 5       put a summary of what we understand Mr Swann's evidence
 6       to be and, absent any contradiction from Mr Russell, I
 7       am not myself proposing to recall Mr Swann, but I can't
 8       yet be definitive about that.
 9   THE CHAIRMAN:  We will see what transpires then.  The only
10       other thing I want to say is that I would not feel it
11       any discourtesy if anyone, having made their submission,
12       wants to leave and was not anxious to listen to other
13       submissions but I would be very pleased of course to see
14       you if you decide otherwise.
15           Tomorrow morning at 10.00.
16   (4.15 pm)
17         (Adjourned until 10.00 am the following morning)
18
19
20
21
22
23
24
25