[MUSIC] To conclude this session on assessing the value of forensic DNA results, let us review together what we have studied. This week we have discussed cases where DNA was said to have caused miscarriage of justice. We have illustrated what needs to be considered when assessing the value of DNA found in a very small quantity. With the case of the murder of Meredith Kercher, also known as the Knox-Sollicito case, we had illustrated the particularities that one must consider with small quantities of DNA. So stutters, drop-in and drop-out alleles. And we have insisted on the need for specialized probabilistic software that accounts for them. Peter Gill, a well-known DNA scientist, has told you in a separate interview about the importance of not considering DNA as a magic bullet. One should not prosecute an individual based on DNA profiling alone. Thus, not considering the other evidence is a recipe for disaster. Peter has also underlined that evaluation given source-level propositions is not the same as given activity-level propositions. Education regarding to how assess scientific evidence is critical at all levels of criminal justice system. During this interview, he has mentioned the Scott, Easton, and cases that we have seen this week. And also, the Madrid bombings that we will study together in week five, but this time from a fingerprint perspective. We have also seen that the experts in the case gave categorical opinions saying that the DNA was Amanda Knox's or Rafael Solicito's. This is not in line with the principles of interpretation, and it says that the scientists should only talk about the probabilities associated with their results. We will have the opportunity to come back on the topic of categorical opinion later in this course when dealing with fingerprint evidence. In all the cases, we the have shown that with trace quantities, stringent control procedures are needed on the crime scene, at the hospital and in the laboratory. Indeed, pollution is an aspect one needs to take into account. You have seen with Professor Pierre Margot why we use the term pollution rather than contamination. And you have seen also the importance of avoiding pollution, whatever the forensic discipline. It is important that in a case one should not rest on only one trace. We have to acknowledge, even if it rare, that errors, pollution can happen. And indeed, professor Luparia in his interview underlined the role played by pollution in the Knox and Sollecito case. He clarified the influence such an aspect had on the judgement of the Italian Supreme Court. We need to consider the possibility of pollution, especially when DNA is the central, and only, sometimes, element supporting the allegation of a person's involvement in a crime. We have seen with Alex that lay persons do not deal with these probabilities very well. We have seen also that what dominates the value of the trace with material that is very rare, such as DNA, it is the probability of error and pollution that dominates. And finally, as we have seen in week two, source level propositions are not adequate at all when dealing with small quantities of material. As a take-home message, here are the essential points of this week. You know how we obtain a DNA profile and are able to explain what a DNA profile is. You know now that with small quantities of DNA, one needs to account for drop-in, drop-out, elevated stutters. Dedicated probabilistic software will allow the scientist to carry out that complex task. Third, with the small quantities, whatever the trace, one needs to take into consideration, pollution and also depending on the case, legitimate transfer. Fourth, you know that accounting for the probability of error is not intuitive. And you know also that the forensic scientist needs to assist the court in that task. Indeed, what dominates the value of the results is the probability of error. So if the probability of the DNA profile given defense proposition is 1 in a billion and the probability of error 1 in 10 thousand, then the probability of the results, considering that there may or may not have been an error, will be 1 in 10 thousand. So much more common than the 1 in a billion. This concludes this week's session. Do not hesitate to ask any question that you have on the forum. Thank you and goodbye. [MUSIC]