In another case, in January 1995, Dennis Adams, who live in the area of the crime, was charged with sexual assault. The prosecution called a DNA expert who confirmed a correspondence between Adams' DNA profile and the DNA profile obtained from the semen recovered from the victim. No other incriminating evidence was presented by the prosecutor. Note that in this case, the victim did not recognize Adams as her assailant. Moreover, Adams' girlfriend gave him an alibi for the night in question. These known DNA pieces of evidence are not directly relevant to the discussion regarding the inversion or prosecutor fallacy that occurred in this case. Christoph, could you please focus more on the testimony regarding the DNA findings? Yes. The expert testified that the random match probability was one in 20 million. By random match probability, expert generally mean the probability to observe a match if an unknown unrelated person was the source of a recovered stain. We leave aside the discussion that occurred regarding how this value was calculated. We will take a value of one in two million as was finally accepted by the defense. Where did the transpose conditional or prosecutor's fallacy comes in? Well, it's when the prosecutor stated that one in two million represented the probability that Adams was not the source of the DNA. The probability of obtaining a DNA match if Adams was not the source of semen was falsely equated to the probability that Adams was not the source of the DNA. Franco, how come we see this in equations? Of course, let's have a look. We can use almost the same slide as before, but with the number 1 in two million. Both probabilities cannot be equal, but when we say it out loud, it's very hard to detect the [inaudible]. There is a confusion between different concept. We ought to be careful about that. So, please, never say that the probability of observing the evidence given the hypothesis is equal to the probability of the hypothesis given the evidence. These problems with probabilities and transpose conditionals are not new, they occurred in one of the major miscarriage of justice where forensic science played a critical role. Let's go back to the turn of the 20th century when Alfred Dreyfus, a French captain, was accused of treason. He was falsely suspected of delivering military secret to the German Embassy. The main piece of evidence is a document called the bordereau, which we have a copy here, that has handwritten information and conveys military information. The Dreyfus case has large political and historical implication, but we will concentrate here only on the forensic aspects as exposed in 1899. It all started with this document that you have on the screen. Bertillon is considered as the founding father of forensic science in France, and was asked to first reconstitute the document that had been shredded, photograph it, and then compare it with the handwriting of Dreyfus. So, Bertillon was at the peak of his reputation and was the first to present very novel techniques in court that had never been seen before, and this is the list of Bertillon's publications that he could bring to support the trust that the court should give him. He was a very exceptional expert witness who came to discuss his view about how the bordereau was produced. According to Bertillon, Dreyfus manufactured the bordereau using a specific word, interet, displayed, found in a letter from his brother. He used that word as a template. The paper of the bordereau was thin and translucent, allowing the application of an underneath chain of characters made of the successive apposition of a word interet, and that's the chain of words interet that you can see here. It led to a text with positions of start and ends of the words that could be found in matching location. Okay. Well, let me show you how the bordereau was prepared according to Bertillon's theory. So, I will show you here a copy of the bordereau. This document has been retrieved in shredded form from the German Embassy. It has been reconstituted by Bertillon. Let put the record straight before we move on. We now know that Dreyfus never wrote the bordereau. In fact, further forensic examination showed that Esterhazy did, through using his natural writing, compose the entire bordereau. Now, what I like to do is to take you through what is known as the Bertillon system, how according to Bertillon, and we know that this is not the way it was conceived, according to him how that document was produced. You will see that Bertillon took liberties by taking information from the bordereau to elaborate a very complex system to explain everything you can see on it. That document, his first observations where that this document has some regularities, and I show you his photographs of both the recto side of the bordereau and the verso. So, part. When you align a grid on that border hole at defined length or widths, Bertillon highlighted that some words will start and end in exactly the same position regarding the grid. That gave him the hunch that these documents might have been fabricated using some regular mechanism in order to be produced. So, when Bertillon overlaid that document with regular lines, these lines shown here, he realized that the handwriting underneath will interact with these lines always in the same position. Typically, positions where the word will start will be found with some sort of irregularity as far as the lines are concerned and this is what Bertillon named coincidences. Bertillon then searched in the personal files of Dreyfus and discovered among the letters, in one letters from the brother of Dreyfus. He discovered that word, the word in French is attehe. That word was completely described by Bertillon by measures. According to him, this is the key word that Dreyfus has used to construct the system whereby it will produce the forgery. How? By constructing what is named Aga Bahi or underneath text made of concatenation of the word attehe, one after the other. If you look at this chain, it is constituted by the word attehe in dark, applied one after the other and there is a second chain and because Bertillon wanted to explain all the coincidences you observe, they had to develop a second chain. The second chain here in grey constituted itself by the word attehe as well. The theory proposed by Bertillon to produce the final document is that, Dreyfus will have, by overlay, the sheet of a border home on top of that underneath chain of words attehe which will give the rhythm to the writing. It is this way that Dreyfus will have written the secret information to be passed to the German Embassy. So, it is not the word attehe that has been reproduced, it is simply this underneath structure in terms of positions and angle of the letters that will dictate how the forged document would have been produced. Bertillon went further in presenting what would be the favorite final results and this is shown here on the verso of the Bordereau. On that image, Bertillon is showing how you can superimpose perfectly the underneath chain of words, attehe on top of the text of a border or itself, and it is how we discovered that there is regularities called these coincidences between some words in the document and the vertical lines that reflect the rhythm of the juxtaposition of the word attehe. Then Bertillon offered a statistical argument, following the observation of four words, having the same position on the grid overlaid on the document. According to Bertillon, "The probability of observing such four correspondences by mere coincidence", meaning, following a natural unconstrained handwriting was one in five to the power of four for four correspondences and that is a value of one in 625. For Bertillon, this probability proved that the document, the Bordereau was a forgery. We had to wait for a review carried out by Darboux, Appel, and Poincare to expose the deficiencies of the Bertillon's system. The system suffered from multiple errors, among them, the final inference proposed by Bertillon. The Committee highlighted two things regarding the inference. The first is that Bertillon committed what now we know as the transposed conditional or inversion fallacy. He confused the probability of the observations of the correspondence, if we accept them, with have a probability of a document being a natural handwriting and hence, not a forgery. Now, you know that the second does not follow logically from the first. The second point raised by the committees that Bertillon considered only a one side of the argument. In a very similar way that was outlined before in the Sally Clark case, Bertillon looked only at the probability of a correspondences if the Bordereau was a natural writing, but never considered the alternative. How likely is it to find the correspondences if the document was truly a forgery? Without this balanced view, it is impossible to weigh the evidence. So, nothing really new here, as soon as a statistical argument is offered to a Court of Justice to qualify forensic science, the risks of misinterpretation are present. Thank you for watching this video. You can follow us in the next video.