The Chemical Interpretation of Alchemy

A Critical Review
William B. Jensen
Department of Chemistry, University of Cincinnati
Cincinnati, OH 53706

In the field of archeology, it is becoming increasingly only the appearance of gold, such as the recipes for
common to experimentally test various hypotheses as gold-like materials found in the well-known Leyden
to how ancient civilizations managed to accomplish Papyrus (4), the true alchemical literature is full of
feats that, in retrospect, seem almost miraculous (1). authors who thought they were performing a real auri-
Can one travel across the Pacific in a crude reed boat? faction, when at best they may have unknowingly
Could the Egyptians have built the pyramids using achieved only an aurifiction. This was possible be-
earthen ramps to elevate the building blocks? Were the cause, as Boyle later complained, alchemists often
statues on Eastern Island moved from the quarry where failed to properly test their products to determine
they were carved to their current locations using log whether they really had, aside from the requisite color,
rollers? Is it possible to accidentally discover glass by the other known properties of metallic gold (5).
building a fire on a sandy beach? What was the source
Since we know from the principles of modern
of tin used in the ancient manufacture of bronze, etc.? chemistry that a true aurifaction is impossible using
Rather than merely speculating on these questions, it only chemical means, this must imply that reports of
has now become common for archeologists to experi- successful aurifactions, if not imaginary or cases of
mentally test the likelihood of various proposed an- mistaken identity, must have relied on having, know-
swers. Though such tests cannot positively prove that a ingly or unknowingly, introduced gold into the system
proposed scenario is what actually happened, they can at some point in the process. The traditional skeptical
eliminate physically impossible suggestions from fur- rationale has been that this was done on purpose and
ther consideration.
that the alchemist in question was a charlatan intent on

To a lesser degree, historians have begun to apply duping a potential source of financial support – usually
a similar experimental approach to various questions in of the royal variety (6).
the history of chemistry and especially to the evalua-
A typical example of this sort of rationale was
tion of assorted recipes and processes found in the al- provided by the 19th-century Scottish chemist, Thomas
chemical literature (2). The purpose of this review is Thomson, in his well-known history of chemistry (7):
not only to call attention to the various experimental
studies and chemical speculations reported in the lit- Sometimes [the alchemists] made use of crucibles with
erature, but also to provide a critical evaluation and to a false bottom. At the real bottom they put a quantity of
suggest a few additional possibilities for future study. oxide of gold or silver. This was covered with a portion
of powdered crucible, glued together by a little gum-
Possible Aurifactions med water or a little wax. The materials being put into
this crucible and heat applied, the false bottom disap-
Joseph Needham, in his epic study of technology and pears, the oxide or gold or silver is reduced and, at the
science in ancient China, proposed that a distinction be end of the process, is found at the bottom of the cruci-
made between what he called aurifaction and aurific- ble and considered as the product of the operation.
tion (3). Aurifaction refers to serious attempts to truly
Sometimes they make a hole in a piece of charcoal
make gold via transmutation, whereas aurifiction refers and fill it with oxide of gold or silver and stop up the
to the making materials that mimic the appearance of mouth with a little wax; or they soak charcoal in solu-
gold but which are known to be imitations. Which of tions of these metals, or they stir the mixtures in the
these labels applies depends on the beliefs of the per- crucible with hollow rods containing oxide of gold or
son performing the experiments in question. Though silver within, and the bottom shut with wax. By these
there are ancient examples of recipes that were explic- means the gold or silver wanted is introduced during the
itly known by their authors to produce products having process and considered as a product of the operation.

1
 WILLIAM B. JENSEN

Sometimes they have a solution of silver in nitric If the resulting aqueous tetrachloroaurate anion is re-
acid, or of gold in aqua regia, or an amalgam of gold acted with either mild (i.e. alkali carbonates) or caustic
or silver, which being adroitly introduced, furnishes the (i.e. alkali hydroxides) alkali, it will produce a yellowish-
requisite quantity of metal. A common exhibition was brown precipitate of gold trihydroxide:
to dip nails into a liquid and take them out half con-
verted into gold. The nails consisted of one-half gold, AuCl4-(aq) + 3OH-(aq) ➝ Au(OH)3(s) + 4Cl-(aq) [6]
neatly soldered to the iron and covered with something
to conceal the color, which the liquid removed. Some- which, on careful drying, will yield, in turn, blackish-
times they had metals, one-half gold and the other sil- brown digold trioxide:
ver, soldered together and the gold side whitened with
mercury. The gold half was dipped into the transmuting heat + 2Au(OH)3(s) ➝ Au2O3(s) + 3H2O(g) [7]
liquid and then the metal heated; the mercury was dis-
sipated and the gold half of the metal appeared. though, as one modern writer has warned, the hydroxide

and oxide are poorly characterized and so similar in their

All of the reactions for oxide of gold, or digold properties that many so-called samples may in fact be
trioxide (Au2O3), reported by Thomson are thermody- mixtures of the two or of various hydroxyoxides (9).
namically favored. Like Ag2O and HgO, simply heat-
Based on two well-known 17th-century reports of
ing this oxide will cause it to decompose into the metal supposedly successful transmutations – one by the
and dioxygen gas (8): Flemish iatrochemist, Jan Baptist van Helmont (1580-
1644), and the other by the Dutch-German physician,
heat + 2Au2O3(s) ➝ 4Au(s) + 3O2(g) Johann Helvetius (1625-1709), I would like to suggest
ΔG° = -37.1 kcal/mol rx [1] a possible alternative to Thomson’s account in which a
different gold compound was, in all likelihood, un-
and it is easily reduced by carbon, though since we are knowingly used to perform a supposedly successful
taking about a transmutation rather than a fire assay, it transmutation (10). The route of introduction was not
is puzzling why an alchemist would put a piece of char- via hidden compartments or disguised gold samples,
coal in the crucible rather than in the furnace beneath it: but rather via samples of the so-called philosopher’s
stone. This was provided in both cases by a mysterious
heat + Au2O3(s) + 3C(s) ➝ 2Au(s) + 3CO(g) stranger, so neither van Helmont or Helvetius would



ΔG° = -116.98 kcal/mol rx [2] have necessarily known that it was a compound of
gold.

Not considered in this account is the possibility
Van Helmont described this substance as “of a
that the gold oxide could also be reduced by the base color, such as is in Saffron in its Powder, yet weighty,
metal used in the presumed transmutation: and shining like unto powdered Glass,” whilst Hel-
vetius described it as “resembling glass or pale sulfur.”
heat + Au2O3(s) + 3Hg(l) ➝ 2Au(s) + 3HgO(s) Both of these descriptions are consistent with the
ΔG° = -60.52 kcal/mol rx [3] known color of hydrogen tetrachloroaurate (figure 1),
the primary product obtained by evaporation of the
heat + 4Au2O3(s) + 9Pb(l) ➝ 8Au(s) + 3Pb3O4(s) resulting solution in reaction 5, as well as with the col-
ΔG° = -517 kcal/mol rx [4] ors of other closely related gold compounds (Table 1),
whereas the reference to an appearance like powdered

Though all of this chemistry would have been glass suggests that the material in question was most
known to 18th- and 19th-century chemists, it is more
questionable how much was known to the alchemists
of earlier centuries. Digold trioxide cannot be made by
directly reacting metallic gold with dioxygen gas and
the only route to such gold compounds would have
been via the reaction of gold with aqua regia, well
known to both alchemists and early modern chemists
alike. In modern terms, can be represented by the net
ionic equation:

Au(s) + 4H+(aq) + NO3-(aq) + 4Cl-(aq) ➝ Figure 1. A tube of crystalline hydrogen tetrachloroaurate

AuCl4-(aq) + NO(g) + 2H2O(l) [5] tetrahydrate. Was this the philosopher’s stone?

2
 THE CHEMICAL INTERPRETATION OF ALCHEMY

Table 1. Various Gold Compounds and their Colors. this proposal. The first is that not all alchemists agreed
that the philosopher’s stone had a yellow color. Thus
Compound Color the 14th-century alchemist, Nicolas Flamel (1340-1418),
claimed that it was red, rather than yellow. This prob-
Au2O3 brown lem was actually noted by Helvetius, but the mysteri-
ous stranger who showed him the light yellow “stone”
Au(OH)3 brown replied that “the color makes no difference, and that
the substance was sufficiently mature for all practical
H[AuCl4] light yellow purposes” (7). This refers to the alchemical belief that
the philosopher’s stone was prepared by a progressive
K[AuCl4] light yellow
maturing process, each stage of which was character-
ized by a specific color – red or ruby being the final or
Na[AuCl4] orange yellow
fully matured product, though this was often preceded
Au2Cl6 orange red by a yellow or citron stage.

Consistent with this color sequence is the fact that
yellow hydrogen tetrachloroaurate tetrahydrate can be
likely in the form of a crystalline powder. converted into red digold hexachloride (figure 2) by

In van Helmont’s account this powder was added careful heating. Differential thermal analysis shows
to heated mercury in order to affect the transmutation, that this occurs in several stages: below 130.7°C the
presumably leading to the generation of metallic gold tetrachloroaurate first looses its water of hydration:
via the reaction (8):
heat + H[AuCl4]•4H2O(s) ➝
heat + 2H[AuCl4](s) + 3Hg(l) ➝






H[AuCl4](s) + 4H2O(g) [10]


2Au(s) + 3HgCl2(s) + 2HCl(g) [8]
whereas between 130.7 and 186.8°C it further decomposes
whereas in Helvetius’ account it was added to molten into red digold hexachloride and hydrogen chloride:
lead, leading to the analogous reaction:




heat + 2H[AuCl4](s) ➝ Au2Cl6(s) + 2HCl(g) [11]
heat + 2H[AuCl4](s) + 3Pb(l) ➝





2Au(s) + 3PbCl2(s) + 2HCl(g) [9] and, finally, at yet higher temperatures, into metallic
gold and dichlorine gas (11):

Unhappily I have been unable to locate any free
energy data for the tetrachloroaurates that would allow heat + Au2Cl6(s) ➝ 2Au(s) + 3Cl2(g) [12]

one to quantitatively verify that these reactions are
both thermodynamically favored. However, given the
Once formed through controlled thermolysis, the
notoriously unstable nature of most simple gold com- red digold hexachloride, like the yellow hydrogen tet-
pounds, the ease with which they are reduced to metal- rachloroaurate in reactions 8 and 9, can react with ei-
lic gold by a wide variety of reducing agents, and the ther metallic mercury or molten lead to produce metal-
thermodynamic stabilities of the various byproducts, it lic gold, and the proposed reactions can both be ther-
is highly probable that both have the required negative modynamically evaluated due to the availability of the
free energy values (6). It is also of note that both van necessary free energy data for digold hexachloride (6):
Helmont and Helvetius reported observing an efferves-
cence on adding the powder to the liquid base metal, heat + Au2Cl6(s) + 3Hg(l) ➝ 2Au(s) + 3HgCl2


most likely due to the escaping gaseous hydrogen chlo-


ΔG° = -110 kcal/mol rx [13]
ride byproduct postulated in these equations.

There are, however, several pertinent objections to heat + Au2Cl6(s) + 3Pb(l) ➝ 2Au(s) + 3PbCl2(s)
ΔG° = -201.92 kcal/mol rx [14]

The second objection is more serious. Reactions 8,
9, 13 and 14 would produce only a small amount of
gold due to the small amount of the philosopher’s stone
used in the so-called transmutation and this gold would
have to be separated from the bulk of unreacted base
Figure 2. An ampoule of digold hexachloride. metal via scorification. However, both van Helmont

3
 WILLIAM B. JENSEN

and Helvetius reported that the entire masses of the

mercury and lead, both of which were many times
greater than the mass of powder used, were instantly
transformed into gold and that this gold was properly
tested and found to be authentic (12). The alloys of
gold with both mercury and lead are white or gray,
rather than gold, in appearance and neither would pass
for gold in a conventional assay test.

Of course I am not the first person to propose that
the so-called philosopher’s stone might actually have
been an easily decomposable compound of gold. In-
deed, several alchemists hinted as much. Thus Eire-
naeus Philalethes (aka George Starkey, 1628-1665)
described the stone as “a species of gold” and Michal
Sendivogius (1566-1636) admitted that it was “nothing
other than gold digested to the highest degree.” In 1987 Figure 3. The appearance of tin disulfide when precipitated
John Schroeder suggested, based on the color sequence in the standard qual scheme.
often quoted for the preparation of the philosopher’s
stone (13): chemists as “mosaic gold” or aurum musivum. Though
European recipes for its preparation date back as far as
black ➝ white ➝ yellow ➝ red ➝ violet the 14th century, the authors fail to directly quote a
specific example but rather summarize the ingredients
that the stone was actually digold oxide (Au2O), which given in a 1678 book by Johann Kunkel involving the
has been described as having a grayish-violet color (9). heating of a mixture of tin amalgam, ammonium chlo-
Like the gold (III) compounds discussed above, there ride, and sulfur in a crucible.
is little doubt that this compound would easily form
Since heating a simple mixture of tin and sulfur
metallic gold upon heating or upon reduction by either produces only blackish tin sulfide (SnS), the pertinent
metallic mercury or lead. However, Schroeder further question is why the presence of ammonium chloride
implies that this oxide was made by heating gold in air, allows the oxidation of the tin to proceed all the way to
which, as already noted for the case of the trioxide, is the disulfide stage. After experimentally verifying that
not chemically possible. Indeed, more recent reviews the recipe works using metallic tin rather than tin
of gold chemistry claim that there is still no definitive amalgam, and that other ammonium halides can be
evidence for the existence of this oxide, which, even if substituted in place of the chloride, the authors pro-
it does exist, is likely to be extremely unstable. ceeded to postulate a rather elaborate reaction se-

In any case, though thermodynamically favored, quence to rationalize these results:
both reactions 8 and 9 and reactions 13 and 14 still
await experimental verification by someone lucky heat + Sn(s) + S(s) ➝ SnS(s) [15]
enough to be able to afford the present-day cost of the
necessary gold compounds needed to determine whether heat + (NH4)Cl(s) ➝ NH3(g) + HCl(g) [16]
the reactions in question are also kinetically labile.
2HCl(g) + SnS(s) ➝ H2S(g) + SnCl2(s) [17]
Aurifictions
H2S(g) + NH3(g) + xS(s) ➝ (NH4)Sx [18]
Moving on to tested examples of known aurifictions
rather than proposed tests of presumed aurifactions, SnS(s) + (NH4)Sx ➝ SnS2(s) [19]
attention should first be called to the studies of An-

thony Butler, Christopher Glidewell, Joseph Needham
This proposal is unsatisfactory for a number of
and coworkers in the 1980s (14, 15, 22, 24, 26). In a reasons, not the least of which is that steps 18 and 19
1985 paper these authors examined two alchemical remain unbalanced and employ an incorrect formula
recipes for the preparation of tin disulfide (SnS2) – one for ammonium polysulfide, which should be written as
European and one Chinese (14, 15). As older chemists (NH4)2Sx, since most polysulfide anions are known to
who have taken an undergraduate course in qualitative carry a 2- charge. Without thermodynamic evaluation,
analysis know, this compound has a golden yellow color it is also not obvious why the polysulfide anion, but not
(figure 3) (16), and was thus known to European al- elemental sulfur, is able to oxidize tin from the II to the

4
 THE CHEMICAL INTERPRETATION OF ALCHEMY

IV oxidation state, presumably via the half reactions: bled upon by alchemists, though he cites no original
recipes to confirm this hypothesis (18). The first class
Sn2+ ➝ Sn4+ + 2e- [20] involves roasting mixtures of various metallic oxides and
sulfides to produce the results summarized in Table 2.
2e- + Sx2- ➝ 2S2- + (x-2)S [21]
The second class involves reacting the oxides of
copper with various thiosalts, all of which are found as

In the case of China, the authors quote an English naturally occurring minerals that were accessible to
translation of a circa 300 AD recipe from a text by an early alchemists, to give the results in Table 3.
alchemist named Ko Hung. After first pointing out that
several of the ingredients were mistranslated and argu-
ing that a previously unknown term stands for ammo- Table 3. Copper Alloys Produced by Roasting
nium chloride, the authors proceeded to experimentally with Various Thiosalt Minerals
verify that it works. The modern equivalent involves
heating a mixture of tin filings, dried alum, and ammo- Oxide Sulfide Product Color
nium chloride in a crucible sealed inside an iron bomb
CuO AgSbS2 Cu6.2Ag2.2Sb Gold
for five days at 500°C. Here the alum acts as the oxi-
dizing agent in accord with the proposed equation:
CuO Ag8SbS6 Cu3Ag2 Bright Gold
heat +12Sn(s) + 4KAl(SO4)2(s) ➝
Cu2O AgSbS2 Cu1.5Ag2S Gold
3SnS2(s) + 9SnO2(s) + 2Al2O3(s) + 2K2(SO4)(s)




ΔG° ≈ -483 kcal/mol rx [22]

The third class involves reacting either copper

Though reduction of sulfur from VI to II- may at oxide or copper sulfide with various arsenates and
first seem thermodynamically improbable, this reaction stibinates, all of which could have been made by al-
is probably driven by the high thermodynamic stabili- chemists by reacting the naturally occurring oxides
ties of the various oxide byproducts (17). The role of and/or sulfides of arsenic or antimony with either liver
the ammonium chloride in this process is indirect, the of sulfur (calcium polysulfide) or caustic soda.
authors noting only that the process works without it,
but that, when present, it improves crystal formation of Table 4. Copper Alloys Made via Roasting with
the tin disulfide product.
Various Arsenates and Stibinates

More recently Opferkuch has called attention to
several classes of metallurgical reactions for the indi-
rect synthesis of various copper alloys having a gold- Oxide Sulfide Product Color
like appearance that may have been accidentally stum-
CuO Na3AsS4 Cu7AsS Golden Yellow

Table 2. Copper Alloys Produced by Roasting CuO Na3SbS4 Cu16SbS Golden Yellow
Oxide/Sulfide Mixtures
Ag3AsO4 CuS Cu1.3AgS Golden Yellow
Oxide Sulfide Product Color

In 1987 Glidewell briefly mentioned a recipe for
CuO PbS Cu65Pb Copper aurification found in yet another book by the 4th-
century Chinese alchemist Ko Hung (15). This was
Cu2O As4S4 Cu7.5AsS0.5 Silver-Gray affected by adding to molten lead a purple powder
made by heating a mixture of potassium nitrate, hema-
CuO Ag2S CuAg1.1 Gold tite (Fe2O3), sulfur, mercury sulfide, lead amalgam, and
malachite (basic copper carbonate). Experimental rep-
Cu2O Ag2S Cu1.6Ag Gold
lication revealed that the resulting powder was more
CuO red than purple in color, and that it did indeed impart a
Sb2S3 Cu10SbS0.3 Bright Copper
brilliant metallic gold sheen to the surface of the mol-
PbO CuS CuPb20.7 Gray ten lead provided that the surface was first freed from
all contaminants. This color was found to be only sur-
Cu2O SnS Cu17SnS0.3 Yellow-Copper face deep and was attributed to the formation of a mi-
cro, semi-transparent, surface film of yellow lead ox-

5
 WILLIAM B. JENSEN

ide, the metallic appearance being due to sheen of the

The authors found that significant dissolution oc-
underlying molten lead. This was further assumed to be curred only if they used concentrated acetic acid and
the result of the oxidation of the freshly exposed sur- halide ions were also present, leading them to postulate
face by the nitrate in the mixture: the net reaction:

Pb(l) + NO3- ➝ PbO(s) + NO2- [23] HgS(s) + 8NO3-(aq) + 8H+(aq) + 2Cl-(aq) ➝

SO42-(aq) + HgCl2(aq) + 8NO2(g) + 4H2O(l) [24]
the function of the remaining ingredients of the red
powder remaining, in Glidewell’s words, “obscure.”
despite their failure to mention observing any red NO2

In 1976 Schwartz and Kauffman published a two- fumes. The absence of chloride as an explicit ingredi-
part article entitled “Experiments in Alchemy” outlin- ent in the recipe was rationalized by the fact that a
ing a series of simple laboratory experiments for use in Chinese alchemist would have used naturally occurring
a history of chemistry course (19). Despite the title, nitre as his source of potassium nitrate and this is al-
virtually all of these experiments were based on the most always contaminated with potassium chloride. A
early technological literature rather than the true al- source of concentrated acetic acid was much harder to
chemical literature. The only aurifictions discussed explain. Though mentioning that it might have been
involved 19th-century alchemists and are therefore concentrated by freezing, as the ancient Chinese had
beyond the scope of this review. In any case, the done in the case of ethanol (24), they did not experi-
authors either merely summarized the claims of these mentally test this hypothesis. However, in 1980 Butler
modern alchemists or reported, sans details, that their and Needham, using modern models of ancient stills,
personal attempts at replication were a failure. further investigated whether the required concentration

From the relevant and only slightly relevant, we could have been obtained instead via distillation of
now move to the improbable. In 1974 R. J. Gillespie vinegar, but found that the ability these stills to frac-
and coworkers reported the synthesis of a mercury tionate acetic acid-water solutions was quite poor, so
hexafluoroarsenate having the compositional formula an ancient source of concentrated acetic acid still re-
Hg2.86(AsF6) that was found to contain infinitely-extend- mains a mystery (25).
ed polycationic mercury chains (20). This compound
Lastly, the authors’ experiments also suggested that
formed crystals having a metallic gold color and Gillespie, the role of the copper sulfate in the recipe was cata-
perhaps in jest, implied that a past encounter with this lytic. However, they failed to rationalize why the
material by ancient alchemists might account for some original recipe also reported that the final solution was
of their naive reports of having successfully trans- red, implying that it may not have been a solution at all
muted mercury into gold. Later the corresponding an- but rather a suspension of the red cinnabar in water (or
timony analog, Hg2.81(SbF6), was also synthesized and was this color due to dissolved NO2 fumes?).
was found to have a similar metallic gold appearance
In 1987 Butler, Glidewell and coworkers explored
(21). Why I have labeled this suggestion improbable is several other recipes from the same 6th-century Chi-
because these compounds were made by condensing nese book, this time dealing with the dissolution of
either AsF6 or SbF6 onto elemental mercury in liquid gold and silver in vinegar solutions of potassium ni-
SO2 at -196°C – reactants and reaction conditions trate (15, 26). In these cases the anonymous Chinese
hardly available to your average alchemist. author indicated that the dissolutions in question
should be carried out in lacquered bamboo tubes with
Associated Processes reaction times of between 30 and 100 days. For the
dissolution of silver, the authors postulated that the
Alchemical recipes other than those dealing with auri- organic matter in the bamboo first reduced some of the
fictions have also been the subject of modern experi- nitrate anion to nitrite and that this, via a complex reac-
mental studies. Thus in 1980 Butler et al published a tion mechanism involving the generation of a NO+
study of an anonymous 6th-century AD Chinese recipe intermediate, then oxidized the silver via the net over-
for the solubilization of cinnabar (HgS) (15, 22). This all reaction:
was an important process for Chinese alchemists be-
cause of the supposed medicinal properties of mercury NO2-(aq) + 2H+(aq) + Ag(s) ➝
compounds. Based on an earlier English translation by Ag+(aq) + NO(g) + H2O(l)
Needham and coworkers (23), this called for mixing ΔG° = +44.45 kcal/mol rx [25]
the cinnabar with a vinegar solution of copper sulfate
and potassium nitrate and letting the mixture sit a bur- However, as may be seen from my thermodynamic
ied porcelain jar for 30 days. evaluation, this reaction is not thermodynamically fa-

6
 THE CHEMICAL INTERPRETATION OF ALCHEMY

vored under standard conditions, though it might still

In this article Principe attempted to replicate Val-
be driven sufficiently to the right via a Le Chatelier entine’s recipe for the making of so-called antimony
shift due to the escaping NO gas, and, in keeping with glass, followed by an extraction of the glass to produce
this, the authors did indeed report that silver does dis- a “tincture of antimony” for medical use. Unlike some
solved in acetic acid solutions of potassium nitrate of the previous authors, Principle fully quotes the
carried out in glass, but only if some nitrite anion is original recipe, but fails to give the details of his own
also added.
experiments beyond stating their final results. The rec-

For the dissolution of gold, the corresponding rec- ipe begins by first calcining or oxidizing natural occur-
ipe also specified the addition of green vitriol or iron ring diantimony trisulfide or stibinite (Sb2S3) to a “white
sulfate heptahydrate to the mixture. However the ash” – a reaction which Principe represents as:
authors found that the recipe worked only if iodide
anion was added as well. Since potassium iodide is not heat + 2Sb2S3(s) + 10O2(g) ➝
found as a contaminant in naturally occurring nitre but 6SO2(g) + Sb2O3•Sb2O5(s) [27]
potassium iodate sometimes is, the authors further as-
sumed that the function of the iron sulfate in the recipe In fact, this is probably incorrect. Differential thermal
was to reduce the iodate to iodide. When coupled with analysis (DTA) of the oxidation of Sb2S3 shows that
reduction of dioxygen gas in the air, the resulting se- between 283-478°C one obtains, in keeping with Valen-
quence of intermediate steps sums to give the follow- tine’s description, only white diantimony trioxide and
ing, rather complex, overall reaction, leading to the sulfur dioxide as the primary products (29):
dissolution of gold as a diiodoaurate (I) complex:
heat + 2Sb2S3(s) + 9O2(g) ➝
2IO3-(aq) + 16H+(aq) + 15Fe2+(aq) + Au(s) + O2(g) ➝ 6SO2(g) + 2Sb2O3(s) [28]


AuI2-(aq) + 15Fe3+(aq) + 8H2O(l) [26]
In contrast, the mixed-valence oxide in equation 27,
This could not be thermodynamically evaluated due to now written as Sb2O4, is actually yellow rather than
the absence of free energy data for the resulting gold white in color, as is the pentaoxide, Sb2O5 – both of
complex. which are obtained upon heating the trioxide to higher

As may be seen from the above examples, the temperatures (30).
presence or absence of certain impurities in the ingre-
The next step is to fuse the white ash so as to pro-
dients used in many alchemical recipes plays a key role duce a golden yellow glass. However, initially Principe
in determining whether the recipe will or will not work was unable to get this step to work. Success was ob-
when attempts are made to replicate it in the modern tained only after following up a hint found in the writ-
laboratory. This same point was the focus of an article ings of the 19th-century chemist, Jean Baptiste Dumas
by Lawrence Principe, also published in 1987 (27), in (1800-1884), that the addition of a little SiO2, leeched
which he experimentally evaluated several recipes taken from the wall of the crucible, was necessary for suc-
from Basil Valentine’s famous 1604 book, The Trium- cessful glass formation. Principe also noted that the
phal Chariot of Antimony (28). Hungarian stibnite specified by Valentine is generally

Though Valentine was supposedly a 15th-century intermixed with a small percentage of silica, though in
Benedictine Monk and alchemist, historians of chemis- his own experiments he added silica directly to the
try have long suspected that no such alchemist ever molten Sb2O3 in order to successfully form the glass.
existed and that the works attributed to him were all
These results are in keeping with the modern lit-
written in the 16th and 17th century by a variety of erature on glass chemistry. Thus Rawson states that the
anonymous authors under this common nom de plume. glass-forming ability of the Group V trioxides follows
Indeed, many suspect that the book under considera- the order (31):
tion was actually the work of its first publisher – a
German salt maker named Johann Thölde. In addition, P2O3 > As2O3 > Sb2O3 > Bi2O3
it is somewhat misleading to characterize this book as
alchemical (see note 1). Though it contains some of the Indeed, Bi2O3 will not form a glass on its own and
usual alchemical rhetoric in the introductory sections, Sb2O3 only with difficulty and extremely rapid cooling.
most of it is concerned with the supposed medicinal However the binary system SiO2-Bi2O3 does form a
benefits of antimony and its compounds and it is there- glass and presumably the same is true of a SiO2-Sb2O3
fore more accurately described as iatrochemical, rather system. In general, even today, the question of whether
than alchemical, in its emphasis. a given mixture of oxides will or will not form a glass

7
 WILLIAM B. JENSEN

cannot be predicted theoretically and must instead be

tested empirically.

The next step in Valentine’s recipe was to leech
the finely ground glass with distilled vinegar, followed
by evaporation of the vinegar and dissolution of the
remaining reside in ethanol to produced a red “tincture
of antimony.” Once again Principe could not get the
recipe to work, the powered glass remaining inert to
freshly distilled wine vinegar. Only when he guessed
that the vinegar had been “sharpened” prior to distilla-
tion via the addition Salmiac (a mixture of ammonium
chloride and carbonate) did the extraction steps work.
However, when the final alcoholic tincture was ana-
lyzed, it was found to contain, not antimony, as Valen- Figure 4. The structure of the red Fe3[O(AcO)6(H2O)3]+
tine had assumed, but rather iron! This was rationalized complex ion found in basic iron acetate.
either by the fact that the Hungarian stibinite used in
the first step was found to be associated with minute process that DTA shows will occur between 595-
amounts of hematite or by the fact that Valentine had 610°C. Which of these rationales actually applies is a
used an iron rod to stir his molten glass, though Prin- subject for future investigation.
cipe failed to make a decision one way or another on
As our final example, we cite a 1997 paper by
this point. Rodygin (34) on the modeling of an alchemical recipe

Why the extraction worked after addition of for the production of acetone attributed to the 15th-
Salmiac is because these ammonium salts give off century English alchemist, George Ripley (1415-1490),
ammonia gas on being heated to dryness and this or rather on a rationalization of this recipe given by
would pass over into the receiver where it would react Jean-Baptiste Dumas in his 1839 volume on the history
with the vinegar distillate to produce ammonium acetate: of chemistry (35). This paper is unsatisfactory for a
number of reasons. Thus neither Rodygin nor Dumas
NH3(g) + H(C2H3O2)(aq) ➝ directly quote the original recipe and neither references



NH4+(aq) + C2H3O2-(aq) [29] its original source (36). Given that the terminology used
by Ripley was apparently highly allegorical (and there-
Unlike weakly dissociated acetic acid, this salt is com- fore far more typically alchemical than the example
pletely dissociated in aqueous solution and would gen- used by Principe), there is also the problem that no
erate a sufficiently high concentration of free acetate explanation is provided as to how Dumas translated the
ion to extract the iron in the glass as red basic iron ace- recipe into a set of concrete modern ingredients or
tate. This contains a heterocatenated complex ion (fig- whether his apparent French paraphrases of Ripley’s
ure 4) having the composition Fe3[O(AcO)6(H2O)3]+ allegorical names are reasonable.
(where AcO stands for acetate) that is known to be
According to these authors, the recipe begins by
soluble in ethanol. heating something called mecure des philosophers to

Since, when Sb2O3 is introduced into either the first produce lion vert and then lion rouge:
phosphate or borate glasses used for bead tests in
blowpipe analysis (32), the resulting beads are either heat + mecure des philosophers ➝ lion vert ➝ lion rouge
white or colorless, it is unlikely that a pure SiO2-Sb2O3
glass would be yellow. Thus the question of the origin which they interpret as:
of the yellow color in Valentine’s product remains to be
answered. The most obvious choice is that the yellow 2Pb(s) + O2(g) ➝ 2PbO(s) [30]
color is due to the iron (III) impurities in the glass.
However, iron (III) normally colors oxide glasses ei- As required, PbO is red. However the nature of lion
ther green or brown or, at best, a yellowish brown. vert remains obscure, since the only other known
(33). Though the latter color, at low concentrations, polymorph of PbO is yellow rather than green in color.
might be charitably interpreted as a golden yellow,
The lion rouge is then reacted with l’esprit aigre
there is yet another possible choice, since, if heated to des raisins to produce a matiere gommeuse:
a sufficiently high temperature during the fusion proc-
ess, some of the white Sb2O4 may be further oxidized lion rouge + l’esprit aigre des raisins ➝
to give the yellow mixed valence oxide, Sb2O4 – a matiere gommeuse

8
 THE CHEMICAL INTERPRETATION OF ALCHEMY

which the authors interpret as the reaction between lead and processes found in the true alchemical literature
oxide and vinegar to produce lead acetate trihydrate: are plagued by three recurring problems:

PbO(s) + 2H(C2H3O2) + 2H2O(l) ➝ 1.

Proper identification of the necessary ingredients

Pb(C2H3O2)2•3H2O(s) [31]
and reaction conditions.

Finally, the matiere gummeuse is strongly heated 2.
The hidden role of impurities.
in a retort to give dragon noir and three liquid fractions
described as l’eau ardent, flegme insipide, and an oil 3.
The pervasive, but imprecise, use of color as
called sang humaine:
the sole means of characterizing reactants
and products.
heat + matiere gommeuse ➝ dragon noir + l’eau ardent



+ flegme insipide + sang humaine
The first of these problems is much more serious
when dealing with European versus Chinese alchemi-
With the exception of sang humaine, this reaction can be cal recipes. As Needham has emphasized, the latter are
represented as (37): usually straight forward, whereas the former are often
obscured by the pervasive use of allegory, as illustrated
heat + 2Pb(C2H3O2)2•3H2O(s) ➝ by the Ripley recipe for acetone. Nevertheless, the
2Pb(s) + (CH3)2CO + CO2(g) + 3H2O(g) + O2(g) [32] Chinese recipe for mosaic gold did contain a previ-
ously unknown term for ammonium chloride and sev-
in which l’eau ardent is interpreted as acetone, flegme eral of the recipes for the dissolution of minerals and
insipide as water, and dragon noir as finely powdered metals in vinegar-nitre solutions failed to explicitly
pyrophoric lead, as the recipe calls for setting this on mention one or the other of these two ingredients –
fire to reform lion vert and then repeating the entire their presence being inferred instead from the nature of
process. Interestingly, the use of this reaction to pro- the surrounding recipes. In yet other cases, such as the
duce pyrophoric lead was rediscovered by Böttger in addition of Salmiac to the distilled vinegar used to pre-
the 19th century and is still used to prepare this mate- pare tincture of antimony, the replicator simply had to
rial for demonstrations in freshman chemistry (38). As guess that something was missing. As for reaction con-
a side bar, the term l’eau ardent was traditionally used ditions, most modern replicators, lacking the patience
in alchemy to denote ethanol (24), but since it trans- to let a process proceed for 30-100 days, have been
lates as “burning water,” it is not unreasonable that it forced to use much shorter reaction times. Other prob-
might also have been applied to acetone. lems have been glossed over, such as, for example, the

As noted by Dumas, unlike modern chemists, proper temperature setting for a heating mantel if one
Ripley apparently had little interest in the acetone is trying to duplicate the heating effects of a dung bath.
product. Rather it was the red liquid byproduct, or sang
The often critical role of impurities for the success
humaine, that fascinated him and which, because of its of a given recipe was amply illustrated by several ex-
color, he identified with the long sought elixir d’sages. amples in this review. Sometimes the impurity was
It was the purpose of Rodygin’s note to discover the present in the starting materials, as in the case of KCl
nature of this liquid fraction. However, his description in the recipe for the solubilization of cinnabar or
of his experimental results is so terse that it is unclear K(IO3) in the case of the recipe for the dissolution of
whether he actually prepared this fraction using reac- gold, whereas in other instances the necessary impurity
tion 32 and then analyzed it using “gas-liquid chroma- was extracted from the chemical apparatus used in the
tography, IR and UV spectroscopy, and both 1H and process, such as silica from the crucible wall in the
13C NMR,” or whether he analyzed a mixture modeled case of the recipe for antimony glass, or iron (III) from
on what he thought it contained based on a search of an iron stirring rod in the case of the recipe for tincture
the chemical literature. In any case, he concluded that of antimony, or the organic reducing agents from the
it was a solution of lead bis(acetylacetonate) in a mix- wall of the bamboo tubes used for the Chinese recipes
ture of acetic acid and acetylacetone and that the red for the dissolution of metals.
color might be due to polycondensation products of the
Less often noted is the third problem. That color
latter compound. played a central role in alchemical thought has been
known since the pioneering work of Hopkins at the
Conclusions beginning of the 20th century (39). Indeed, it was the
red colors of the so-called tincture of antimony in Val-
This review reveals that attempts to replicate recipes entine’s recipe and of the sang humaine in the Ripley

9
 WILLIAM B. JENSEN

recipe that caused their original authors to focus on metallurgical or pharmaceutical literature ever explicitly identified
these two incidental products rather than the products itself with alchemy (the case of iatrochemistry is less clear cut).
of interest to the modern chemist (i.e. antimony glass Also this argument from period usage would logically preclude
and acetone). Of all the properties that can be used to use of an Arabic term to describe alchemical practices in ancient
characterize a material, color – unless quantified via China, India, and Egypt. I personally prefer to use the less pro-
modern spectroscopy – is the least dependable. The vocative term “protochemistry” as a blanket descriptor for pre-
visual color of a given substance is known to vary with modern chemical practice and to retain the traditional historical
such circumstances as the presence of impurities, its use of the term alchemy for the above two practices so as not to
degree of hydration, and its degree of subdivision. Due produce an unnecessary discontinuity with earlier scholarship.
to oxidation or exposure to chemical fumes, the surface
3.
J. Needham, L. Gwie-Djen, Science and Civiliza-
may have a color different from the bulk material, etc. tion in China, Vol. 5(II), Cambridge University Press: Cam-
And of course, not only do individuals vary widely in bridge, 1974.
their ability to visually identify colors, different cul-
4.
W. B. Jensen, Ed., The Leyden and Stockholm Pa-
tures do as well. Are the colors recognized by the an- pyri: Greco-Egyptian Chemical Documents from the Early
cient Chinese identical to those recognized by medie- 4th Century AD, Oesper Collections: University of Cincin-
val Europeans? At what point did these two cultures nati, Cincinnati, OH, 2008. Available online.
distinguish between orange and red or between ruby
5.
L. M. Principe, The Aspiring Adept: Robert Boyle
red and violet, etc.? I am unaware that the implications and his Alchemical Quest, Princeton University Press:
of these questions for the study of the alchemical lit- Princeton, NJ, 1998, Appendix 1. This is an attempt, albeit
erature have ever been explicitly confronted. highly biased, to reconstruct Boyle’s unpublished dialog on
“The Transmutation and Melioration of Metals.”
Applications to the History of Modern Chemistry
6.
For typical examples see H. C. Bolton, The Follies
of Science at the Court of Rudolph II (1576-1612), Pharma-
In closing I cannot resist briefly noting that the ex- ceutical Review: Milwaukee, WI, 1904. For a more sophisti-
perimental approach to historical questions in the his- cated interpretation, see P. H. Smith, The Business of Al-
tory of alchemy is being increasingly applied to ques- chemy: Science and Culture in the Holy Roman Empire,
tions in the history of modern chemistry as well, most Princeton University Press: Princeton, NJ, 1994.
notably by the late Melvyn Usselman. These include
7.
T. Thomson, The History of Chemistry, Vol. 1,
replication of some of William Hyde Wollaston’s micro Colburn & Bentely: London, 1830, p. 30. This is apparently a
electrochemical apparatus (40); of Wollaston’s and paraphrase of an account written by the French chemist
Thomas Thomson’s analyses of various carbonates and Éttienne Geoffroy in 1722. Thomson’s punctuation is highly
oxalates, often quoted as the first important experimen- eccentric by modern standards, so I have modernized it but
tal confirmations of Dalton’s then recently formulated not the wording.
law of multiple proportions (41); and a reconstruction
8.
All thermodynamic data are either from W. M.
of Liebig’s original apparatus for organic combustion Latimer, Oxidation Potentials, 2nd ed., Prentice-Hall,
analysis and replication of many of his analyses (42). Englewood Cliffs, NJ, 1952; or M. K. Karapet’yants, M. L.
Karapet’yants, Thermodynamic Constants of Inorganic and
References and Notes
Organic Compounds, Humphrey: Ann Arbor, MI, 1970.


9.
The best review of the basic inorganic chemistry

1.
D. Ingersoll, J. E. Yellen, W. Macdonald, Eds., Experi- of simple gold compounds is still J. W. Laist, “Gold” in M.
mental Archeology, Columbia University Press: New York, C. Sneed, J. L. Maynard, R. C. Brasted, Eds., Comprehensive
NY, 1977. Inorganic Chemistry, Vol. 2, Van Nostrand: New York, NY,

2.
In this article I use the term alchemy as it has been used 1954, Chapter 3. This is because more recent reviews of gold
by historians for the past 300 years to denote either attempts to chemistry focus instead almost exclusively on the organo-
transmute base metals into gold or discover the elixir of life. I am metallic chemistry of gold.
fully aware that Newman and Principe have declared this usage
10.
English translations of both accounts may be
incorrect and have instead named the first of these practices found in H. S. Redgrove, Alchemy: Ancient and Modern,
chrysopoeia and argued that the term alchemy should be used as a Rider & Son: London, 1922, Chapter 5.
blanket descriptor for all early chemical practice, be it technologi-
11.
L-G. Zhen, W. Gup, M. Bai, X-W. Yang, “Prepara-
cal, metallurgical, pharmaceutical, or iatrochemical in nature. Their tion of Chloroauric Acid and Its Thermal Decomposition,”
primary argument is that this is how the term was used in the past Chinese J. Nonferrous Met., 1982, 16, 1976-1982.
by many of its actual European practitioners. However, I am un-
12.
See reference 5. Boyle spends considerable space
aware that the term chrysopoeia was ever widely used by Euro- attempting to rationalize how it was possible for a small
pean alchemists nor that most of the distinctive technological, amount of the philosopher’s stone to completely transform a

10
 THE CHEMICAL INTERPRETATION OF ALCHEMY

much larger amount of base metal.

from the German edition of 1604.

13.
J. C. Schroeder, “A Chemical Interpretation of
29.
Z. Zivkovic, N. Strbac, D. Zivkowic, D. Grujicic,
Alchemy,” J. Chem. Educ., 1987, 64, 994-995. B. Boyanov, “Kinetics and Mechanism of Sb2S3 Oxidation

14.
A. R. Butler, C. Glidewell, J. Needham, S. Prit- Process,” Thermochimica Acta, 2002, 383, 137-143.
chard, “Mosaic Gold in Europe and China,” Chem. Brit.,
30.
N. N. Greenwood, A. Earnshaw, Chemistry of the
1983, 19, 132-136. Elements, Pergamon: Oxford, 1984, p. 672.

15. C. Glidewell, “Ancient and Medieval Chinese Pro-
31.
H. Rawson, Inorganic Glass-Forming Systems,
tochemistry,” J. Chem. Educ., 1987, 66, 631-633. Academic Press: London, 1967, pp. 203-204.

16.
C. H. Sorum, Introduction to Semiqualitative
32.
O. C. Smith, Identification and Qualitative Analy-
Chemical Analysis, 4th ed., Prentice-Hall: Englewood Cliffs, sis of Minerals, Van Nostrand: New York, NY, 1946, pp. 157-
NJ, 1967, Chapter 4. 158. Interestingly Valentine also describes the formation of a

17.
The free energy of formation of SnS2(s) was ap- white B2O3-Sb2O3 glass in reference 28.
proximated using the enthalpy of formation reported by
33.
W. Weyl, Colored Glasses, Dawson: London,
Karapet’yants in reference 6. 1959, Chapter 7.

18.
R. Opferkuch, “Metallurgische Aspeckte der Al-
34.
M. Y. Rodygin, “On the Modeling of an Alchemi-
chemie,” in W. Sawodny, R. Opferkuch, A. Schunk, Eds., cal Process,” J. Chem. Educ., 1997, 74, 949-950.
Chemie im Spiegel der Jahrhundertem, Universitätsverlag
35.
J. B. Dumas, Lecons sur la philosophie chimique,
Ulm: Ulm, 1992, pp. 57-65. Socièté Belge de Librairie: Brussels: 1839, pp. 25-28.

19.
T. A. Schwartz, G. B. Kauffman, “Experiments in
36.
I have, however, located a version in the so-called
Alchemy,” J. Chem. Educ., 1975, 53, 136-138, 235-239. Bosom Book of Sir George Ripley as reprinted in Collectanea

20. I. D. Brown, B. D. Cutforth, C. G. Davies, R. J. Chemica, Being Certain Select Treatises on Alchemy and
Gillespie, P. R. Ireland, J. E. Vekris, “Alchemist’s Gold, Hermetic Medicine, The Alchemical Press: Edmonds, WA,
Hg2.86AsF6: An X-Ray Crystallographic Study of A Novel 1991, pp. 121-147. As with reference 28, this is an unac-
Disordered Mercury Compound Containing Metallically knowledged reprint of a volume published by Arthur Waite in
Bonded Infinite Cations,” Can. J. Chem., 1974, 52, 791-793. 1893. The treatise by Ripley that it contains was apparently

21.
B. D. Cutforth, PhD Thesis, McMaster University: first published in London in 1683. Since it is not listed in
Hamilton, Ontario, 1975. most reputable bibliographies of Ripley, I do not know if it

22.
A. R. Butler, C. Glidewell, J. Needham, “The So- was actually written by him or by an anonymous author
lutilization of Cinnabar: Explanation of a Sixth-Century Chi- under his name. Its use of the various allegorical terms dif-
nese Alchemical Recipe,” J. Chem. Res., 1980, (S)47, fers somewhat from the version paraphrased by Rodygin and
(M)0817-0832. Dumas.

23.
Ts’ao T’ien-Ch’in, Ho Ping-Yu, J. Needham, “An
37.
H. Wang et al, “Thermal Decomposition of Lead
Early Medieval Chinese Alchemical Text on Aqueous Solu- Acetate Trihydrate in Air Atmosphere,” Chem. Ind. Engr.
tions,” Ambix, 1959, 7, 122-155. Prog., 2015, 34, 3570-3373.

24.
Lu Gwei-Djen, J. Needham, D. Needham “The
38.
W. B. Jensen, “Whatever Happened to Homberg’s
Coming of Ardent Water,” Ambix, 1972, 19, 69-112. Pyrophorus?,” Bull. Hist. Chem., 1989, 3, 21-24.

25.
A. R. Butler, J. Needham, “An Experimental
39.
W. B. Jensen, “Alfred J. Hopkins and the Color
Comparison of the East Asian, Hellenistic and Indian (Gand- Theory of Greco-Egyptian Alchemy,” to be submitted.
harian) Stills in Relation to the Distillation of Ethanol and
40.
M. C. Usselman, “Wollaston’s Microtechniques
Acetic Acid,” Ambix, 1980, 27, 69-76. for the Electrolysis of Water and Electrochemical Incandes-

26.
A. R. Butler, C. Glidewell, S. M. Glidewell, S. E. cence,” in J. T. Stock, M. V. Orna, Eds., Electrochemistry,
Pritchard, J. Needham, “The Solubilization of Metallic Gold Past and Present, ACS Symposium Series No. 390, ACS:
and Silver: Explanation of Two Sixth-Century Chinese Pro- Washington, DC, 1989, pp. 20-31.
tochemical Recipes,” Polyhedron, 1987, 6, 483-488.
41.
M. C. Usselman, “Multiple Combining Propor-

27.
L. Principe, “‘Chemical Translation’ and the Role tions: The Experimental Evidence,” in F. L. Holmes, T. H.
of Impurities in Alchemy: Examples from Basil Valentine’s Levere, Eds., Instruments and Experimentation in the History
Triumph Wagen,” Ambix, 1987, 34, 21-30. See also, L. Prin- of Chemistry, MIT Press: Cambridge MA, 2000, pp. 243-271.
ciple, The Secrets of Alchemy, University of Chicago Press:
42.
M. C. Usselman, A. Rocke, C. Reinhart, K.
Chicago, IL, 2014, pp. 138-143. Foulser, “Restaging Liebig: A Study in the Replication of

28.
B. Valentine, The Triumphal Chariot of Antimony, Experiments,” Ann. Sci, 2005, 62, 1-55.
The Alchemical Press: Edmonds, WA, 1992. This is appar-

ently based on an English translation dating from 1685 and is

an unacknowledged reprint of an edition first published by
Arthur Waite in 1893. Principe provides his own translation

11