The Smithsonian Founder Developed Poison Tests

In 1822, James Smithson wrote an article describing a method he used for detecting arsenic. Curiously, he had written about this test just a few years earlier, and once he had written on a subject it was unusual for him to revisit it. But in this case, he explained, the “importance of the subject” justified it. The subject was the need to identify the agents in cases of suspected poisoning, and it was a problem that many chemists were struggling with. Poisons were an increasing danger in the early nineteenth century, not only in cases of deliberate poisoning but also from the growing problem of environmental pollution. One of the unintended effects of the Industrial Revolution had been to bring vast quantities of toxic materials into contact with humans, and one of the challenges of Smithson’s time was to find ways to deal with this. Two of the most problematic substances were arsenic and mercury, and in his article, Smithson described sensitive new methods to detect each of them.

The bulk of the article was devoted to the test for arsenic. In 1817, the noted physician and toxicologist Mathieu Orfila had written that “the preparations of arsenic are, of all the poisonous substances in the mineral kingdom, the most fatal.” And yet in England, where arsenic was widely used as a pesticide, it was essentially unregulated and readily available.

Arsenic, the twentieth most common element in the earth’s crust, is gray in its “elemental” form, with a metallic appearance. Surprisingly, in this form it is not poisonous. Arsenic is only poisonous when combined with other elements, and the form usually used as a poison is a combination of arsenic and oxygen commonly known as “white arsenic.” Known to the public in Smithson’s time as “ratsbane” and to chemists as “arsenious acid,” it has no distinctive taste or smell. When ground into a powder it looks very much like sugar or flour, so it could be added to foods and beverages without arousing suspicion. It is not very soluble in liquids, however, so as a poisoned food or drink cooled the arsenic often settled out of solution, giving it a gritty texture, one of the few indicators of its presence.

By the time this happened, the victim had usually consumed some of the arsenic, and it did not take much to be fatal. In Smithson’s time substances were generally measured in grains, a grain being the average weight of a single grain of wheat. Depending on a person’s physiology, a lethal dose usually consisted of no more than four or five grains, although death could result from as little as two.

The initial symptoms of arsenic poisoning were acute abdominal pain and rapid pulse, followed by convulsions, and then severe vomiting and diarrhea, often leading to death. Arsenic was the chief example of what early toxicologists called an “irritant poison,” and it was a particularly unpleasant way to die. Autopsy typically revealed severe damage to the digestive tract, and it was common to find both the stomach and intestines so inflamed that they assumed a deep red color, their inner surfaces covered with mucus and blood. Other organs could also be affected, depending on the individual, the amount of poison consumed, and how it was administered.

With such distinctive symptoms, it seems like the use of arsenic would have been simple to determine, but even when suspected its actual use was difficult to prove. One of the most egregious examples was the case of Robert Donnall, an English physician who fell into debt and who, in the fall of 1816, was charged with murdering his wealthy mother-in-law. English newspapers avidly followed the case and reported that she had fallen ill on two separate occasions, each time after drinking tea he had brought her, and that she died the second time after fourteen hours of vomiting. As the town’s medical expert, Donnall urged that she be buried quickly, but before that could happen the authorities received an anonymous letter accusing him of murder. It was reported that when he was shown the letter, “His hands trembled, and . . . it dropt from his hands upon the floor.” Under these suspicious circumstances, and despite the doctor’s objections, an autopsy was ordered, and the victim’s stomach was found to be inflamed in a way consistent with arsenic poisoning. Before it could be tested, Donnall—who was inexplicably in the room—“accidentally” dropped the organ into a partially filled chamber pot, conveniently compromising the evidence. Nevertheless, chemical tests found arsenic in the victim’s last meal, and the doctor was charged with murder.

His trial the next spring was attended by huge crowds. The evidence initially seemed overwhelming, but Donnall mounted a vigorous defense. For example, to explain the inflammation of the victim’s stomach, the defense brought in three expert witnesses to testify that it could have been caused by cholera morbus, a rare form of dysentery. Although highly unlikely, the prosecution’s examining physician had to admit that this was at least a possibility. They also challenged the chemical tests that had detected arsenic. The victim’s last meal had been “smothered rabbit”—rabbit stewed in onions. A defense witness described a test in which he had put sliced onions and meat, but no arsenic, in a pot which he let stand for several hours. Using the same two chemical tests as the examiner, he reported that they both incorrectly indicated the presence of arsenic. It was devastating testimony, and when the case went to the jury, the ability of chemical analysis to reliably detect arsenic seemed so suspect that it took them only twenty minutes to find the defendant not guilty.

The problem was that the available tests for arsenic were all indirect. The ideal test would have extracted pure arsenic directly from the evidence, but such a test did not yet exist. The chemical processes used to do this required a large, solid sample and were rendered unreliable by the presence of organic compounds—and, of course, some kind of organic material was almost always present in the evidence of poisoning cases. As a consequence, the liquid tests actually used in these cases sought not to isolate pure arsenic, but rather to produce a distinctive sign of its presence. They worked by bringing the evidence into contact with a chemical solution that produced a precipitate of a specific and characteristic color when arsenic was present.

However, it seemed that every test developed for this purpose had some problem. For example, one of the most common tests used a solution of copper sulfate to test food suspected of being poisoned. If there was arsenic in it, a lively grassgreen precipitate known as “Sheel’s green” (copper arsenate) would form. However, it turned out that if there was onion juice in the food and if that juice had come into contact with copper, such as from a cooking pot, this too would produce a green precipitate when tested. This is why the “smothered rabbit” in the Donnall case had tested falsely for arsenic. Every test the chemists proposed seemed to have a similar weakness.

Textbooks of the time contain elaborate protocols for dealing with these uncertainties, the first of which was to use multiple tests, each of which had to give a positive indication of the presence of arsenic. They also recommended treating the sample ahead of time to avoid known problems. For instance, if the sample contained tea or coffee, the tannin would affect the results and needed to be removed with gelatins. If the sample contained oil, it needed to be boiled and the oils removed by capillary action from “wick threads.” If it contained resins, these needed to be removed with turpentine—but not alcohol, which would dissolve the arsenic. Other substances, such as milk, would change the color of the tests’ precipitates or make them difficult to see. In these cases some authors recommended that the sample be treated with chlorine, to “decolorize” it, but this could create new problems by affecting the colors of the tests’ precipitates.

In 1821, with the English public increasingly fearful of arsenic poisoning and the reliability of the tests for it widely questioned, the Society for the Encouragement of Arts, Manufactures and Commerce offered a gold medal to any person “who shall discover to the Society a test for arsenic in solution, superior to any hitherto known.” The announcement brought this issue to the attention of the scientific community, and although he did not apply for the prize, the following year Smithson published this article describing how to apply his arsenic test to poisoning cases. Smithson emphasized his test’s adaptability, extreme sensitivity, and the fact that it could detect arsenic in many different forms: as an acid, in combination with oxygen, and in its metal-like elemental form.

The test’s first step was to combine the sample with “nitrate of potash” (potassium nitrate, KNO3) and heat it with a flame until it dried and caught fire. This burned away the organic material and caused any arsenic to combine with the potassium. Now any excess potassium needed to be removed, and this was done by carefully adding vinegar until the solution was neutralized. This liquid was then dried and dissolved in water, which put the “arsenate of potash” (K3AsO4) into solution. Then, just as he had in his 1819 article, Smithson described how adding a small amount of “nitrate of silver” (silver nitrate, AgNO3) to the liquid would cause a “brick-red precipitate” to appear—but only if the original sample had contained arsenic. The distinctively colored precipitate that settled in the bottom of the liquid was “arsenate of silver” (Ag3AsO4), and even a tiny amount of arsenic would make a lot of it. He reported that a grain of arsenic acid would make 4.29 grains of the precipitate, that a grain of arsenic oxide would make 4.97 grains of it, and that a grain of pure arsenic would make 6.56 grains. This, along with the color, made the appearance of the precipitate hard to miss, and it was an important part of the test’s sensitivity.

Smithson’s test was based on the reaction of silver nitrate with compounds containing arsenic, but it wasn’t the first arsenic test to make use of this chemical. One of the tests used in the Donnell case had also used silver nitrate, but in that test any arsenic in the sample would combine with it to produce “arsenite of silver” (Ag3AsO3), which is a yellow precipitate. What they discovered in the Donnell case was that even when no arsenic is present, the naturally occurring alkaline phosphates found in animal tissues will also combine with silver nitrate to produce a yellow precipitate—and thus give a false positive for the poison.

The fact that Smithson’s test produced a distinctive brick-red precipitate, together with his method of removing any organic material in the sample, was widely credited as having addressed the problem of using silver nitrate. His test was quickly adopted as one of the standard methods of detecting arsenic, and a positive result was a strong indication that the analysis was correct. Finding a reliable method of extracting elemental arsenic from the sample would continue to be a goal, but until such a method was found Smithson’s test served the purpose. As the distinguished chemist Richard Phillips noted in a widely cited article, “from repeated trials, I consider the confirmatory evidence afforded by this experiment as amounting almost to demonstration.” There were other expressions of support as well. At the end of a long discussion of arsenic in one of his textbooks, Robert Hare concluded, “Mr. Smithson’s plan . . . seems to me an excellent one.” In his widely reprinted Elements of Medical Jurisprudence (1825), Theodric Beck noted with approval that Smithson’s plan was “well calculated to detect the most minute portions of poison.” Smithson’s article was also reprinted in the United States where, as in England, arsenic poisoning was a significant concern.

Smithson’s test was quickly adopted, and for more than fifteen years it was one of the most reliable and widely used methods for detecting arsenic. But in 1836 the chemist James Marsh announced a new test that was finally able to meet the criteria of the Society for the Encouragement of Arts competition and earn its coveted gold medal. The “Marsh Test,” as it came to be known, was based on a new principle and used gaseous hydrogen to extract pure arsenic from a solution and deposit it on a glass plate. This was what everyone had been waiting for, and the test almost immediately became the standard method of arsenic detection. It remained in widespread use until the 1970s, when it was finally replaced by the new technologies of chromatography and spectrophotometry. As for Smithson’s test, like most of the other “liquid” arsenic tests, its use declined rapidly once the Marsh Test became available, and by midcentury it had been largely forgotten.

But Smithson’s article also described another test, this one for mercury, that would remain in widespread use for nearly a century. Mercury, like arsenic, is a poisonous metal whose toxicity varies according to its chemical form. Pure mercury is a liquid at room temperature, and in that form, it is so heavy that it is not readily absorbed—although it is still a health hazard. However, mercury acquires different properties when combined with other materials. It is toxic in the form of “corrosive sublimate” (HgCl2), and somewhat less toxic in the form of “calomel” (HgCl). Indeed, in Smithson’s time pure mercury and calomel were better known as medicines than as health hazards. At that time, mercury was one of the few treatments that seemed to provide relief from syphilis. “Blue mass” pills, which contained either calomel or liquid mercury mixed with chalk along with taste and coloring agents, were used by physicians to treat it and a wide variety of other medical conditions.

This is not to say that doctors in Smithson’s time were unaware of the hazards of mercury exposure. In 1811, Andrew Mathias, a member of London’s Royal College of Surgeons and surgeon extraordinary to Queen Charlotte, wrote An Inquiry into the History and Nature of the Disease Produced in the Human Constitution by the Use of Mercury, in which he accurately described the effects of mercury poisoning, calling it “the disease of the remedy.” Like his contemporaries, however, he saw these symptoms mostly as the undesirable side effects of an effective medicine. His book was not intended to warn about the dangers of using mercury, but rather to guide physicians in the best ways of using it while limiting its undesirable effects.

In 1818, just four years before Smithson wrote his article, at least three thousand Parisian workers were reported to be sick from mercury poisoning. Whole families of the city’s residents exhibited symptoms of poisoning. In 1821 a Health and Safety Council report called attention to “the skin diseases that are so common in Paris, especially in the indigent classes,” and blamed it on their washing clothes in mercury-polluted water. In the 1820s, government-encouraged changes to the workshops began to improve conditions for the workers, but because most of these changes were to enhance ventilation, the improvement was at the expense of nearby residents, whose exposure to mercury was now dramatically increased.

Smithson’s awareness of the problem explains his decision to include the mercury test in his 1822 article, which was otherwise about detecting arsenic. Although he described it only briefly, his test became the most practical and sensitive method for detecting mercury yet developed.

Read more in The Science of James Smithson: Discoveries from the Smithsonian Founder, which is available from Smithsonian Books. Visit Smithsonian Books’ website to learn more about its publications and a full list of titles. 

Excerpt from The Science of James Smithson © 2020 by Steven Turner and the Smithsonian Institution