Regardless of whether the perceived problems for expert witnesses in the new criminal trial procedure appeared in practice or only in the imagination of the forensic toxicologists, strategies to overcome these problems had to be found. The textbook authors offered three strategies: two of these strategies corresponded to what I have previously called redundancy and intuitive comprehensiveness, the third one attempted to transfer the responsibility of method selection to the state.
To illustrate redundancy, consider the following quote from the textbook Die gerichtliche Chemie für Gerichtsaerzte und Juristen written by the Viennese chemist and physician Franz Schneider (1812–1897) and published in 1852:
[T]he substance has to be examined with all characteristic reagents and the chemist must explicitly point out on which reactions he bases his statements on the presence of a substance and in which properties the substance in question differs from all other substances which are related or similar in their effect. Only by performing these steps will his judgment be exhaustive and, at the same time, will it counter those doubts and objections which might be raised by the court or the attorneys.Footnote 5 (Schneider 1852: 6; emphasis in the original)
In this view, the best way to overcome critical questions and the aforementioned risks was not to choose a single method, but rather to use all accessible methods for identifying the substance in question. According to this logic, if one method was not enough to convince the court, the sheer mass of evidence produced by following this strategy surely would succeed. Of course, a stategy of redundancy also diminished the risk of producing false positives, yet this is not how Schneider justifies it here. Instead, he explicitly argues that it provides the best way to face “doubts and objections” from the court; thus it is justified as a defense strategy against lay persons and not (solely) as a way to do better analytic work.
This strategy of employing a great number of redundant methods did not work well in practice. The main reason for this was that there was a tendency in forensic toxicology to strive after and use methods to employ the smallest possible sample size. In poisoning cases, the sample was limited and the court might need to preserve leftovers to ask for second opinions in possible appeals (Carrier 2019: 49).
The Quarrel About Standard Methods
One possible way to get around the whole question of method selection was the implementation of standard methods set by the state. If the state specified standard methods, the courts could then examine expert testimonies on a purely formal level. This idea was most prominently supported by Carl R. Fresenius (1818–1897) even before the beginning of the reform to the trial procedure. In 1844, he published the article Ueber die Stellung des Chemikers bei gerichtlich-chemischen Untersuchungen in which he began by explaining that one of his tasks in this article would be to highlight,
how useful it would be if the state would prescribe well tried and tested methods for the detection of poisons as a standard so that every chemist commissioned with the task of an investigation of a poisoning must follow this standard.Footnote 6 (Fresenius 1844: 276)
Fresenius goes on to explain that every choice that the chemist makes can and will ultimately face harsh criticism in court. He provides an example of a chemist who is commissioned by the court to find arsenic:
If he [the chemist] chooses one of the older methods [to find arsenic] and he cannot detect arsenic, it will be said, “how can one apply such a method when we have much more accurate and better ones?; if one had chosen Marsh’s apparatus, one would have found arsenic.” If he chooses Marsh’s apparatus and he detects arsenic, the defendant’s attorney will without doubt say: “what can one think about these results, results, that is, which are obtained by a method that can easily be misleading […].” If, finally, the chemist wants to satisfy everybody, and he divides his substance in as many parts as there are methods, examining every part with a different method and cannot find arsenic, it will be said that there is nothing more natural than that result, since every beginner knows that every reaction has a minimum sensitivity […].Footnote 7 (Fresenius 1844: 283f.; emphasis in the original)
Fresenius highlights how the chemist’s choices are restricted and each strategy carries disadvantages. Since the matter came down to a choice, it could always be attacked as possibly not being the best one. Redundancy could not be the answer since then the amount of sample material would simply not be sufficient. The only viable path Fresenius saw for escaping this problem was to let the state make the choice. He drew support for this position from the pharmacopeia, in which directions for how pharmaceuticals should be prepared or checked were written down. Just as the pharmacist was only responsible for acting as prescribed by the pharmacopeia and not for checking each recipe, Fresenius argued, the forensic toxicologist should only be responsible for using the correct standard method and not for deciding on the method itself. The choice of good methods should not be the responsibility of the individual expert but rather of the state (Fresenius 1844: 285; cf. also Reinhardt 2011).
In the textbooks, this plea for officially-set standard methods appears, however it is generally seen as something unrealistic and undesirable. This skepticism also rested on the observation that new methods were found too quickly for the state to react. Every law that would set the catalogue of standard methods was bound to be outdated right after its implementation (Schneider 1852: 5f.). Fresenius had already foreseen this issue in 1844, yet concluded that it was not a decisive problem. Laws, he argued, are always dependent on circumstances and must be changed when the circumstances they rely on change. Of course, it would be the responsibility of the state to adapt the standard methods to the developments in analytic chemistry, but this did not mean, in Fresenius’s view, that the state was unable to set a standard that would be valid for a certain time (Fresenius 1844: 286). He seems to have been less concerned about whether the expert actually used the best method and instead argued that a set of prescribed standard methods would bolster the authority of an expert testimony in courts of law. The methods themselves would then stand above critical questions and would have to be accepted by every layperson involved. However, he did not succeed in convincing his colleagues and there were no legally specified standard methods. To be sure, this did not mean that there was no standardization, but rather that the textbook authors did not want to give up their own right and responsibility to establish and develop these standards in practice.Footnote 8 The Marsh test for arsenic (and its variants) most certainly was a standard method and it was practically used as such, however, it was not legally prescribed but rather set by the community.Footnote 9 To establish such community standards, the textbook authors employed a different strategy in their method selection which was guided by the value of intuitive comprehensibility.
Intuitive Comprehensiveness, Part I: Showing the Material
The practical usefulness of the next strategy, which adhered to the value which I call intuitive comprehensibility to approximate the meaning of the German word Anschaulichkeit, was much greater than both redundancy and the plea for standard methods, which is also why Schneider himself preferred it. In the paragraph directly before the quote about redundancy cited above, he wrote:
Whereas the analytical chemist will choose such detection methods which are either distinguished by their characteristic chemical behavior or allow for the most accurate quantitative analysis, the forensic chemist must aim to present his substances in such a way that they can easily be recognized and distinguished by everyone.Footnote 10 (Schneider 1852: 5; my emphasis)
This quote is interesting for two reasons: first, Schneider draws a distinction between analytic chemists and legal chemists that relies on a specific form of method selection, and second, this method selection was oriented towards the abilities of lay persons to recognize certain substances. In Schneider’s view, the ability to find and use these methods convincingly was the defining attribute of professional legal chemists. The discussion of the use of multiple methods cited above was the strategy he recommended only in cases in which such methods were non-existent or for some reason not useable (Schneider 1852: 6). But what did it mean in practice for substances to be in a form which could be easily recognized by everyone? Schneider goes on to give an example:
If, for instance, his [the chemist’s] task is to prove a poisoning by copper or determine the proportion of copper in food products and the like, there is no question that the judge and all laypeople will be much more convinced of the presence of this metal if iron is used to precipitate it in its elementary form—in which it can be easily recognized by everyone due to its copper red color—than if it is dissolved in excess ammonia, which produces the dark blue color by which every chemist detects the presence of copper in qualitative analyses.Footnote 11 (Schneider 1852: 5)
Since even laypeople were familiar with the appearance of copper, seeing a substance which looks like copper would accordingly be convincing enough to establish that the substance in question really was copper. In Schneider’s view, this visual similitude would be more convincing than seeing a solution turn blue and hearing a chemist explain that this change in color demonstrated the presence of copper. In the latter case, the chemist would then have to go on to explain that no other substance would turn the same solution blue and perhaps even demonstrate this process again (Schneider 1852: 6).
This corresponds well with what Burney called “plain matter of fact” (Burney 2012: 78–115). He argued that toxicology gained a special position in poisoning cases precisely because it could reproduce and show the poisonous material:
By enabling experts to present poison in its tangible material form, chemical demonstration held out the promise of disrupting the poisoner’s insidious designs. Through his reproduction of the equivalent of the bloodied dagger in his tubes and retorts, the toxicologist promised to translate this most ephemeral of crimes into a more conventional form of violence. (Burney 2012: 80).
The prime example for putting this idea in practice was the almost exclusive use of the Marsh test for arsenic in poisoning cases. This test was developed by the British chemist James Marsh (1794–1846) and relied on the production of the gas arsenous hydride in a reaction of arsenic, zinc, and sulfuric acid. The gas could then be lit, thus producing a metallic mirror of elemental arsenic on a porcelain or glass surface when held over the flame. If no arsenic is present in the analyzed substance, the gas produced during the test is not arsenous hydride, but hydrogen which does not result in such a mirror when burned (Watson 2006: 192–194; cf. also Burney 2012: 86–101).
The production of the elemental arsenic in the Marsh test is similar to Schneider’s example above. Of course, elemental arsenic does not have the distinctive color of copper, however, showing laypersons an actual metal when claiming that a metal was used as the poison could serve to underline the claim and make it more easily understood and therefore convincing. The Marsh test had the desired attributes of a good analytical test—its high sensitivity, for instance, being an important oneFootnote 12—but this usefulness of the Marsh test as a rhetorical tool seems to have contributed to making it the method of choice for identifying arsenic (Carrier 2019: 53). Arsenic was a popular choice among poisoners in the nineteenth century (Watson 2004: 33). However, it was not the only choice and for many of these alternative poisons, toxicologists lacked methods as demonstrative and conclusive as the Marsh test. In the following, I will argue that there was nonetheless a way to use intuitive comprehensiveness and avoid reliance on redundancy in such cases by employing a strategy of explicit comparison.
Intuitive Comprehensiveness, Part II: Explicit Comparison
Explicit comparison as a rhetorical device does not appear very prominently in the textbooks. To illustrate what I mean by this term, I will give an account of a practical example found in a poisoning case that took place in Springe near Hanover in 1870.
In September 1870, the scribe Fricke in Springe had fallen ill with symptoms of heavy vomiting and diarrhea. His wife, Dorothee Fricke who later was one of the defendants in this case, called the local physician who suspected that the scribe had cholera and tried to treat him for that. However, two days later, on October 2, 1870, Fricke had died. Since there was no sign of a crime at first, Fricke was buried without a postmortem. Only a few days after his death, the journeyman dyer Bernhard Pilz who had lived for nearly a year in Fricke’s house as a boarder wrote his parents that he would either marry the widow Fricke with their consent or never come home again. Pilz’s father refused vehemently, and in November, he came to Springe to bring his son home. However, as the bill of indictment reports, Dorothee Fricke told Pilz’s father that Pilz had killed her husband and that she would denounce him to the police if his father took him away. However, Pilz’s father did not comply and instead took his son with him to the local mayor and the public prosecutor where he himself reported this conversation and let his son make a statement.Footnote 13
This was a rather typical poisoning case in the nineteenth century. As Katherine Watson has shown most poisonings happened within the family.Footnote 14 Out of the 540 cases, she analyzed in 320 (so nearly sixty percent) the main target was part of the family of the poisoner. Wives were one of the biggest groups of poisoners only surpassed by (step-)mothers (Watson 2004: 47).Footnote 15
The poison used in Fricke’s case, however, was not typical. In his statement to the police, Pilz confessed that Fricke’s wife had poisoned her husband using potassium chromate (“chromsaures Kali”Footnote 16) which he had provided. He had obtained the potassium chromate from the dye house where he worked. He was quick to explain that he had never wanted to kill Fricke but he and the widow had started having an affair a month before the scribe’s death. Mrs. Fricke complained about her husband very often. The late Fricke allegedly mistreated his wife when he was drunk—and in fact seems to have been drunk rather often—and was said to be unfaithful to her. As Pilz explained in his confession, she had told him that she could not live with her husband anymore and that she would either run away or poison him and then marry Pilz.Footnote 17 She tried to convince Pilz to provide her with poison from the dye house, but he refused at first. However, a few days later, she found at home a small amount of potassium chromate which Pilz had brought home from work—although he never explained the reason for bringing this home—and which he claims to then have forgotten at home. She gave that to her husband who started to feel ill and when Pilz came home, she told him to bring her more of this substance. Pilz claimed that it was only at this point that he had learned that potassium chromate is actually poisonous and that he started to provide her willingly with more. However, he claimed to never have given Fricke the poison himself.Footnote 18 Mrs. Fricke did not confess to anything and claimed that she had had no affair with Pilz before her husband’s death and that she had never poisoned her husband. If her husband had died by poison, she claimed, Pilz must have been the one giving it to him.Footnote 19
Following Pilz’s confession, Fricke’s body was exhumed, and a postmortem was performed. The protocol of the postmortem, which was performed by the district medical officer and his assistant in the presence of two officers of the court and the prosecutor, reported that no signs of an illness explaining the death of Fricke were found.Footnote 20 However, even if such internal signs had been found, as many textbooks throughout the nineteenth century stress, there was no definitive medical proof of a poisoning which could then be found in symptoms or in a postmortem. Neither symptoms before the death nor appearances on the internal organs, such as inflammation, were specific enough to infer poisoning. All the textbooks stress that symptoms or findings in postmortems could contradict the assumption of poisoning by a certain substance but they could never result in a positive proof (cf. for example Brach 1850: 381f.; Caspar 1860: 408; cf. also Burney 2012: 79f.). The only way to get certain proof that a criminal act had taken place was to gather samples from the body during the postmortem, such as (parts of) internal organs or their contents, which would then be sent from a pharmacist for chemical analysis. In most cases, the court would accompany the samples with clear orders to search for a specific substance or a class of substances based on the results of the preliminary investigation; sometimes questions about the poisonous effects of these substances, that is, what quantity was needed to kill a person, or what the normal symptoms of this poison were also appeared in the orders. Whereas it was important for officers of the court to be present at postmortems, the pharmacist usually would perform the analysis by himself or with an assistant. There was no official witness to oversee his analysis. The textbooks argue that this is mostly a matter of convenience since chemical analyses could take days rather than hours (Brach 1850: 20).
This all was also true in the Fricke case. Samples of different organs and their contents were gathered during the postmortem and sent to the pharmacist Viktor Sertürner (1834–1887) in Hamelin who was commissioned to look for inorganic poisons that could be found in dye houses, especially potassium chromate. There were possible strategies pharmacists could follow when confronted with a completely unknown poison, however, these separation processes took time and material and were avoided when possible. Sertürner systematically excluded the presence of different metals in the sample and finally found chromium. After isolating it, he used two reactions to establish its presence. First, he used silver nitrate which resulted in a red precipitate, and second, he used lead to produce a yellow precipitate. He included both results with his written testimony.Footnote 21
This is exactly the kind of color reaction Schneider frowned upon. Without at least some chemical explanation, there was no reason for laypersons to believe that these red and yellow substances would prove the presence of chromium in the samples: precisely the situation that Schneider wanted to avoid (Schneider 1852: 5). It is possible to understand this behavior as the attempt to adhere to the value of redundancy with limited resources (as I did before, cf. Carrier 2019: 48), which is what Schneider would have recommended in cases where no good method was available to produce the substance in an easily recognizable form. However, it seems to me that redundancy is not sufficient to understand this case of expertise. Sertürner did not use all or even many of the possible methods to detect chromium and he did not use the same methods on different substances to show the different results of this test: he only used two methods, and moreover two very similar methods. In this case, redundancy (in the form of at least more than one method) was combined with another form of intuitive comprehensibility: explicit comparison.
Intuitive comprehensibility always relied on a form of comparison.Footnote 22 When Schneider recommended using forms of substances which are known to everyone, he assumed that everyone had actual experiences seeing the substance in question and thus would be able to recognize it. In the case of Schneider’s recommendation, comparison was implicit; there was no direct reference to the comparison itself. In the Fricke case, however, an explicit comparison was used. In a second letter, Sertürner provided the court with two other precipitates, again red and yellow showing the presence of chromium. This time, he used pure potassium chromium showing that potassium chromium actually behaved in the way he claimed in his first letter. Furthermore, he did not use just any potassium chromate for this demonstration, he used potassium chromate that he obtained from the dye house at which Pilz had worked.Footnote 23 A closer comparison between the alleged poison and his findings in the chemical analysis was hardly possible. Laypersons might have had no prior experience with the chemical behavior of chromium but this demonstration could provide such experience.
Since the protocol of the hearing itself is not to be found in the dossiers, I can only speculate about the effectiveness of this strategy. However, neither the judge nor the jury seemed to have any doubts about the poisoning itself. In the list of questions on which the jury had to deliberate, the question whether the crime itself had taken place was not included. The only questions the jury had to answer were essentially whether Mrs. Fricke had deliberately poisoned her husband, whether Pilz had also deliberately poisoned Mr. Fricke, and whether Pilz had helped Mrs. Fricke by knowingly providing her with poison so that she could poison her husband. The jury decided that Mrs. Fricke had acted alone, although helped by Pilz. She was sentenced to death, but this sentence was later changed to prison for life by a royal pardon. She was released after more than 20 years of prison in 1892 after another royal pardon.Footnote 24 Pilz was sentenced to prison for seven years and released again in 1878.Footnote 25
This example shows another form of intuitive comprehensibility than the one which can be seen in the example Schneider gave in his textbook. Not only methods that produced well-known substances but also this form of explicit comparison can be better understood if seen as a strategy of demonstrating one’s own expertise to laypersons. In the textbooks, however, this kind of demonstration played no role whatsoever. Comparative probes, where the same experiments are done with solutions of known substances, are mentioned to rule out false negatives or false positives but not in this form as a rhetorical device. Usually, the experiments with the original samples and the comparative probes were to be conducted at the same time and under the same conditions to rule out experimental errors or contamination (cf. for example Otto 1884: 12; Schürmayer 1854: 241). In this case, however, it seems that the experiments with the potassium chromate obtained from Pilz’s workplace were done at a later date. There is, after all, no mention of a comparative probe in the extremely detailed description of every single experimental step in the first written testimonyFootnote 26; but it was only in the second letter to the court a week after the first written testimony that these comparison probes were mentioned.Footnote 27 Additionally, there was no chemical reason for obtaining the potassium chromate from Pilz’s former workplace. A comparative probe to rule out mistakes could have used any potassium chromate. Understood as a rhetorical device, however, it suggests that not only had potassium chromate been found during the first analysis, but that the detected potassium chromate was furthermore identical to exactly the potassium chromate Pilz had access to.
Comparative probes and the use of explicit comparisons might look similar in practice (because they are, after all, both comparisons), however they are distinguishable by their use for different purposes. In his typology of comparisons, for example, Hartmut von Sass distinguished, among others, between “Vergleichen als Erkenntnisgewinn” and “Vergleiche als erklärende Abkürzungen” (von Sass 2011: 40–42) which might be useful in understanding the difference I see between comparison probes and explicit comparisons. In the context of analytical practice, the comparison is used for knowledge production. The audience of the comparison is the acting scientist himself who gains knowledge about the chemical composition of a probe. In the context of the expert testimony, the audience consists of the judges, juries, attorneys, etc. Here, the comparison acts as an “explanatory shortcut“ and as a means of providing visual aid where the material itself could not be shown. Its function is no longer knowledge production but convincing others of one’s own knowledge. Thus, the comparison becomes a rhetorical device and is no longer just best scientific practice.
The absence of any description of this strategy in the textbooks does not mean that this strategy had never been used before. I have found at least one French case that can be interpreted in the same way. As mentioned above, France had introduced the new criminal trial procedure in 1808, so the French experts had to adopt to a similar situation nearly half a century before the German ones and in this case they seem to have come up with a very similar strategy.
In 1827, Louis Demollière and his wife Thérese Desplaces Demollière were accused of having killed their neighbors with arsenic. The neighbors, a retired gardener named Heurion and his wife, had decided to sell their land before their death under the condition that they would keep the right of residence on their old estate and receive a life annuity in the case of the death of one of them. The Demollière’s agreed and bought the land. In the bill of indictment, it seems to be clear that the Demollière’s did not keep their end of the bargain after Mrs. Heurion had died. Mr. Heurion was said to be malnourished and not satisfied with the treatment by the couple. He asked his nephew for help who inquired with a notary about the possibilities to annul the original contract. Before the contract could be annulled, however, Heurion died after suffering from a very violent and short illness. The symptoms of this illness were the usual vomiting and diarrhea which did not contradict the suspicions of poisoning in the village of Pontachat. Since Heurion got his food from the Demollière’s, who also stood to financially gain from his death, they were the prime suspects in this case.Footnote 28
An official investigation was started and a postmortem was performed on the victim’s body. Again, there was no medical evidence for poisoning, which is why samples were sent to Jean Gallon, the pharmacist in Étampes who was commissioned with the chemical analysis. Whereas in the German case, the pharmacist did his analysis alone, this early nineteenth-century French pharmacist was in fact accompanied by a medical doctor and two officers of the court as official witnesses. White grains suspected to be arsenic were found in the contents of the stomach and first analyzed by the pharmacist. Since the Marsh test was not developed until 1836, Gallot had to use other methods. The details of these methods are not important as he used a total of five different solutions on the white grains. What is important in this context, however, is that he also used the same methods on a solution of pure arsenic in distilled water, showing effects that were similar in this solution.Footnote 29
In this case, the interpretation of this strategy as adherence to redundancy is not far-fetched as the use of five different methods represented quite an effort. Also, the arsenic solution, could serve as a usual comparison probe since the experiments seem to have taken place at the same time. This is all true, but it seems that explicit comparison and thereby adherence to intuitive comprehensibility also played a role here. In the protocol, which was written by the medical doctor and not by the pharmacist himself, the physician wrote that the pharmacist prepared this solution “[p]our donner plus de certitude à nous”.Footnote 30 This suggests an intentional use of the arsenic solution as a rhetorical device—or at least that the physician had understood it as such—and not (only) as a means of doing better analytical work. More importantly, there is no similar comment on the high number of methods used to show arsenic in the protocol, which suggests that this fact in itself might not have been interpreted by the physician as providing more certainty.Footnote 31
After having found arsenic in the body and after having found a pharmacy where Mr. Demollière had bought arsenic two years prior, the case for the court was clear. Mrs. Demollière was sentenced to death since the majority of the jury was convinced that she was the one who had administered the poison. Mr. Demollière was acquitted.Footnote 32
Both the German and the French case show how the value of intuitive comprehensibility could play its role in method selection even if no methods in the sense of Schneider’s recommendation were available. The purpose of methods such as the Marsh test and of the use of an explicit comparison was ultimately the same: the results of the experiments became understandable for lay-persons. The verdict by the expert as to whether or not a poisonous substance had been found in the samples relied on demonstrating experimental results everyone could perceive. To follow this strategy in courts of law meant to make results as obvious as possible; credibility was not only a question of trust in a specific expert but also, and perhaps even more importantly, in specific methods and their evident results.