How to Sublime Mercury: Reading Like a Philosopher in Medieval Europe

When it comes to “how-to” books, alchemy poses particular problems. Medieval alchemical treatises claimed to offer detailed advice on a host of spectacular products and processes, ranging from the Philosophers’ Stone, a transmuting agent capable of turning base metals into gold and silver, to medicinal elixirs that offered cures for otherwise intractable diseases, as well as the prospect of renewed youth and an extended life span. Yet, at least to modern eyes, no amount of “know-how” could teach anyone how to make these things—they are impossible practices, which no alchemist can ever have successfully carried out, nor described in accurate, replicable instructions. What, then, is the function of a “manual” of alchemy?

First, many of the early readers who studied alchemical writings did consider these ends to be obtainable, while recognizing that their sources might not always include complete or accurate information. Even seemingly straightforward processes can be difficult to reproduce in practice. For instance, the Liber de aluminibus et salibus (Book of Alums and Salts), a Latin treatise translated from an Arabic exemplar and pseudonymously attributed to the Persian polymath al-Rāzī, includes a simple procedure for subliming mercury with vitriol (a metal sulphate) and “common” or “general” salt.¹ According to the text, this mixture will sublime to the top of the vessel in the form of “white and shining” crystals, a description that sounds a lot like corrosive sublimate (mercuric chloride, in modern parlance).

As it happens, corrosive sublimate was a staple ingredient of fourteenth- and fifteenth-century alchemy, although its manufacture normally requires a nitrous salt, such as saltpeter, rather than common salt. Working with Professor Lawrence Principe at the Johns Hopkins University, I have attempted to solve this puzzle by following al-Rāzī’s laconic instructions in a modern laboratory setting.

The Liber directs us to take as much quicksilver as we want, then grind it with equal quantities of vitriol and salt until the quicksilver “dies” in them—that is, until globules of mercury are no longer visible in the mixture. But does “quantity” refer to weight or volume? Although the recipe does not say, the answer must be weight, since measurement by volume results in an excess of quicksilver, which stubbornly refuses to die:

Once we have killed our quicksilver (the Liber tells us), we should leave the mixture in a bread oven overnight. In the morning, we moisten it with salt solution before returning it to the oven. We repeat the cycle of moistening and overnight heating until the mixture has tarnished to a reddish colour (an effect caused by the rusting of iron in the vitriol). Later recipes refine this process, suggesting different ways of achieving the same result. One Middle English procedure recommends grinding the mixture several times a day for two to three days, with “esy dryynge and moystyg with a lytel vynegre.”² Another advises scooping the mixture onto an old potsherd and stirring it over hot embers until no moisture remains.³ We left ours on a low heat overnight.

The mixture is now ready to be sublimed. According to pseudo-Rāzī, it should be placed in an aludel, sealed, and heated (a process that takes longer in winter than in summer). We must use a soft rather than a hasty fire. Otherwise, as one fifteenth-century writer warns, the quicksilver will lose its active power and sublime in its “crude” form—a hazard that can easily be replicated in a modern laboratory. If performed correctly, the mixture sublimes into a substance as “white as milk.”⁴

This is the point when our reconstruction foundered. As already noted, corrosive sublimate requires a nitrous salt—common salt, as specified in both the Liber and the Middle English variants, should not work. In fact, our practice based on common salt conspicuously failed to generate any white crystals.

On the other hand, replacing the NaCl with sal niter immediately produced the desired effect:

Our initial failure illustrates the difficulty of precisely identifying ingredients from alchemical texts. Making corrosive sublimate seems to rely on an ability to distinguish between the properties of different salts, but can we acquire that knowledge entirely from books? Perhaps lack of technical expertise explains the “transmutation” of sal niter into common salt as a transcription error made by a scribe unversed in chemical procedures. Alternatively, our saline setbacks may result from our own lack of experience with this type of procedure.

We are certainly not the first experimenters to face this dilemma. Medieval and early modern recipe collections contain hundreds of procedures for subliming mercury, involving diverse salts and methods of preparation, and no doubt resulting in a variety of chemical products (even pseudo-Rāzī follows up his common salt recipe with a variant using sal ammoniac⁵). But readers also knew that not all recipes were reliable. As the English alchemist Thomas Norton warned the readers of his Ordinal of Alchemy (1477):

Avoide youre bokis writen of receytis,
For al such receptis be ful of deceytis.⁶

Norton was suspicious of recipes precisely because they offered fixed, literal readings of ingredients—but such readings could be wrong. The writers of alchemical treatises typically chose to present themselves as philosophers rather than artisans, whose knowledge stemmed from a profound understanding of nature gleaned not only from practice, but also from studying the writings of past authorities. These “philosophers” emphasized the privileged nature of alchemical knowledge and the need to preserve its secrets from impious or otherwise undeserving readers: an injunction manifested in their use of cover names, allegories, and other methods of obfuscation. In the hands of a self-identified alchemical philosopher, a term like “mercury” could take on almost as many identities as there are chemical substances.

This strategy placed a burden on readers, who sought to disentangle practical instruction from a web of possible meanings. Take, for instance, the opening procedure in one of the most influential works of Latin alchemy: the Testamentum. This lengthy treatise, probably written in the 1330s by a follower of the Catalan philosopher Ramon Llull (ca. 1232–ca. 1315), uses Lullian style diagrams to express alchemical doctrines and practices.⁷ Here the writer plots his ingredients onto an alphabetical wheel:

The wheel begins with “A” (Deus), signifying God. The practice starts with B (defined in both text and diagram as quicksilver), C (saltpeter), and D (vitriol azoqueus), a combination that strongly hints at the making of corrosive sublimate. Has the author of the text simply decked a simple procedure in the garb of Lullian philosophy? Or are the terms merely cover names for other substances, as suggested by the fact that vitriol is qualified as “azoqueus”?

A naive reader may have tried to follow the instructions as they stand. A reader more familiar with the conventions of alchemical writing might regard several of the terms with suspicion. Is ordinary mercury intended, for instance, or does B denote some other “mercury”? And if a procedure fails to work, who is to blame—the authority for setting down a faulty procedure or the reader for supplying a faulty interpretation? Under such circumstances, what the novice alchemist really needs is not a manual of chemical practice, but advice on how to read such practices in a philosophical way. It is in this regard that works like the Testamentum truly come into their own.

This is the final of four pieces in our pre-conference blog series that we are posting in tandem with The Recipes Project: Food, Magic, Art, Science, and Medicine.

Jennifer M. Rampling is Assistant Professor of History, History of Science Program, Princeton University, and former Editor of the journal Ambix.