Welcome to this first video that is dedicated on a case against Dallagher. This is a case where an earmarks was used as evidence and that was disputed. The aim of this video is as follows. We will discuss the elements that are necessary in order to identify a person from a trace. You will realize that identification is the domain of the Court. You will thus understand why identification was said to be outside of the scope of the ENFSI guideline on evaluative reporting. But Christoph, why and how was this earmark used? The case is a murder that occurred in Huddersfield, in England, in 1996. A 94 years old lady, Ms. Dorothy Wood was found dead in her house, suffocated with a pillow, in her bed. On a window, below the transom window used to gain access by the offender to the house, earmarks were detected using fingerprint powder as we have seen. Four overlapping marks, from a left ear and one mark from a right ear were detected. The mark is a two-dimensional impression of various anatomical features of the external ear. These features takes some fancy anatomical names such as, tragus, anti-trugus, helix, anti-helix as you can see on a typical mark. Because the mark is obtained from a [inaudible] position of a 3D ear organ on the surface, we can expect variation from one mark to the other. Thus, this is due to movement, pressure, maliability of cartilage, etc. It is for this reason, that when a suspect is arrested, more than one reference prints will be taken to reflect what we commonly called the within source variability. That is, how variable marks can be from the same person, marks after marks. When the investigation focused on Mark Dallagher, they took a similar set of control ear prints. But Mark Dallagher was known to the police. He didn't come to the police notice due to his ears. The mark was not searching a database. If I'm not mistaken, he was known for previous burglaries and the police told the elderly woman was killed in the aftermath of a burglary, when the burglar realized her presence. Yes, ear prints are typically a modus operandi on the scene. They can be found on burglary cases. By the way, the first documented use of ear prints in police investigation are from Switzerland. Hirschi published in 1917, in the Interpol Criminal Police Review, the first cases where international burglaries were identified in Bern, thanks to earmarks, they left on the scene. The identifications are shown here. So, why is it relevant to carry out such a comparison? It is because prints will be relatively stable when they are left by the same person and quite discriminative when they originate from different individuals. We call this, the between sources variability. Franco and I leave for example, different earmarks. It is because there is a contrast between the within source variability, which is relatively small, and the between sources variability which is larger, but earmarks can potentially be associated with their corresponding prints. So, in the case, the police carry out a super imposition between the mark obtained from the crime scene and the best prints from Dallagher. Similar to the overlay metals presented here, what conclusion we reach? In this case, two experts gave an opinion for the prosecution. The first expert, Cor Van Der Lugt, a Dutch police officers, specialized in ear print examination, concluded based on his experience and training, that, I quote, "The marks found on the window pane are the same - except for differences that can be accounted for -as the known left and right ear prints from Dallagher." He was and I quote again, "Absolutely convinced that they are from the same donor." The second expert, was a Professor of Forensic Medicine and Science at the University of Glasgow. He stated essentially, that he had the same opinion and that it was very likely that it was this defendant who made those prints." I am prepared to go so far as to say that there is a remote possibility that they may have been left by someone else, but it is remote. I am of a firm opinion that it is very likely to be the same person, but I cannot be 100 percent positive." In all this pattern evidence disciplines, how can an expert conclude to an identification? Can we say that ear prints are unique and will it suffice to conclude with certainty? Alex will bring some light on the issue. He will deconstruct the identification process. The way Cor Van Der Lugt testified is very typical of how experts in pattern evidence used to testify and unfortunately some of them still do. They will invoke uniqueness as one of the foundations of their opinion. Typically, they would say fingerprints are unique. Hence, I can conclude that the mark left on the crime scene originated from John Doe. In the Dallagher case, we hear, earprints are unique, hence I can individualize Dallagher." But the sentence, "All ears our unique." is no more than a statement of the obvious. Every entity is unique. What matters is how and to what extent objects can properly be distinguished. So, how is an expert to make that decision taking what is called a leap of faith, which is what a decision in the face of uncertainty is? Bayesian decision theory is very adequate here, but I shall explain what is at stake without any mathematical formula. The experts will start by making some observations on the mark and on the print. These observations will take the form of features that they will highlight as being in agreement, but also potential differences. Needless to say that if the differences cannot be reconciled on the grounds of distortion or within source variability, then the expert will conclude to an exclusion. We will not deal with this case here. We will assume that the expert has observed corresponding features between the mark and the print, without any discrepancies that could not be accounted for. The weight to be assigned to these findings is given by our established framework. That is, by a ratio of two conditional probabilities. Hence, there are two questions that have to be dealt with in the expert's mind. First, what is the probability of making these observations if, indeed, the mark has been left by the same person? In our case Mark Dallagher who also provided a print. Second, what is the probability of making these observations if the mark has been left by another, unknown person? In reply to the first question, absence of notable differences, the probability will be assigned as close to one. The weight of the findings will then be driven mainly by the response to the second question. The probability of an adventitious correspondence. Clearly, it would be inappropriate to claim that this probability is zero. Any scientist should account for a potential correspondence by chance alone. Let's say that these probabilities are assigned as one in a million, for the sake of our example. In reality, for earprints, we would be rather inclined to assign it in the order of one in a thousand for example. Now, moving from one in a million to an identification decision is a leap that requires two further elements. The first element is an assignment for the initial probability of the person of interest being the source of the finger mark. This assignment depends in part on the number of individuals that can be reasonably suspected. The second element is a weighting of the losses and utilities associated with decision consequences. Not necessarily defined its monetary outcomes, but from a larger societal perspective. Here, the decisions are, identification of Dallagher, exclusion or inconclusive. In the case here, the expert decided that an identification was the best decision to make, because he made an ID. This means that, in the expert's opinion when you combine the weight assigned to the findings, which is in the order of a million with the initial probability of Dallagher being the source of the mark, which on the basis of a reasonable assumption about the size of the suspect population, the best decision to make, the one that we as specialist would call maximizing the expected utility or minimizing expected loss is identification. Applying this process is not wrong, from a logical perspective. However, in coming to this conclusion, the expert had to do more than just assessing the observations made when comparing mark and the print. The expert has taken an implicit position on the number of potential suspects, plus probabilities for them being the source and also, on the expected societal costs and benefits associated with this decision. In our view, scientists doing so are taking too much on their shoulders. They're going above and beyond their area of competence. Some scientists would claim that by the forensic task given to them, they're allowed to take this step. However, it is clear that deferring this ultimate decision to the expert is putting him or her in a position that goes beyond the scope of a scientific examination. The expert is asked to make a judgment. For us, this should be left to the court. For these reasons, we advocate a ban on identification decisions and to concentrate testimony of the expert on the only two questions that can be justifiably linked with his area of expertise. I repeat them again. First, what is the probability of making these observations if the mark has been left by the person of interest? Second: what is the probability of making the observations if the mark had been left by another unknown individual? The ratio of these two assignments gives the likelihood ratio. This measure of probative value, properly communicated and understood, is the critical element, from a logical point of view, to help decision makers such as the court, proceed further and make decisions in the light of other circumstances of the case and given societal priorities. When we allow experts to make these decisions, this is an open door to conclusions that can be tainted or biased. For example, whether the suspect is a good suspect according to the information gathered during the investigation, having prior knowledge on other evidence against an individual may lead experts to consider only a restricted pool of suspects. Whether a case is a petty crime or a serious murder, this might affect the balance of the societal [inaudible] such as the loss of a missing association and the loss of making a wrong identification. Finally, whether another mark in close proximity has already been identified, hence, making the expert strengthen his belief that this mark is also from the same source. I know that you will debate these questions again in relation to finger mark evidence, but let me conclude by showing that at the international level, there is now a clear movement to warn against and abandon categorical opinions. I will present three recent sources to testify to that trend. First, in 2016, the US National Commission on forensic science indicated that forensic discipline conclusions are often testified as being held to a reasonable degree of scientific certainty, or to a reasonable degree of discipline certainty. These terms have no scientific meaning and may mislead fact finders about level of objectivity involved in the analysis, its scientific reliability and limitations, and the ability of these analysis to reach a conclusion. Second example: in September 2016, also in the US, the President's Council of Advisors on Science and Technology concluded in a quite critical report for forensic science, "Statements suggesting or implying greater certainty are not scientifically valid and should not be permitted. In particular, courts should never permit scientifically indefensible claims such as: 'zero', 'vanishingly small', 'essentially zero', 'negligible', 'minimal' or 'microscopic' error rates; '100 percent certainty', or proof 'to a reasonable degree of scientific certainty', identification 'to the exclusion of all other sources'; or a chance of error so remote as to a 'practical impossibility.'" Third example: in 2015, the US Army's Defense Forensic Science Center has announced a change in its practice of reporting on the results of a comparison between a finger mark and a print from unknown source. Their strongest claim is now reported as follows: "The latent print on Exhibit so-and-so and the record finger/palm prints bearing the name so-and-so have corresponding ridge detail. The likelihood or probability of observing this amount of correspondence when two impressions are made by different sources is considered extremely low." Thus, to conclude, we can now see why the ENFSI guideline makes it clear that categorical opinions are conclusions that go beyond the sole assessment of the findings. The guideline specifies, "The categorical conclusions of identification expressed by examiners in areas such as the comparative examination of fingerprints, handwriting, and marks go beyond the soul assessment of the forensic findings. These types of conclusions sit outside the scope of this document. However, in these cases, the strict evaluation of the strength of the forensic findings remains a balance between one, the degree of correspondence between features shared by the two specimens, and second, the probability that those features would be observed in another source, which amounts to an assignment of a likelihood ratio." Thank you, Alex. My first learning point here is that we should not allow experts to identify or individualize, as they say. But what happened then in this case? Well, in this case, submissions were made to the Court of Appeal highlighting some of the issues that we just discussed. Thank you for watching this video. You can follow us in the next video.