The Mercury Daguerreotype: Iodine, Amalgam and a Mirror With a Memory

A cased plate in a museum drawer is an object made out of 1840s chemistry, and the chemistry is still sitting there. The globules making its highlights are an alloy of silver and mercury, and they are the same alloy now that they were when they condensed out of a warmed vapour onto a polished mirror in a small dark room. The Canadian Conservation Institute says it to owners in one sentence — the image consists of microscopic particles of silver amalgam, an alloy of mercury and silver, located on the silver surface — and that sentence is the whole of this lesson in compressed form. The picture is an alloy. The alloy was made by a vapour. The vapour was made by heating a liquid metal in a room with a person in it.

Everything worth knowing about the daguerreotype follows from that operation: why the plate had to be a mirror, why the exposures were minutes rather than hours, why the image reverses when you tilt it, why the object has to stay sealed, and why this course will describe the process at length and never once tell you how to carry it out.

☠Level D: what this page will not contain, and why the omission is a decision

The daguerreotype is classified Level D under the course’s safety rubric, and the reasoning is set out once, for the whole course, on the Level D policy. This page cites that policy rather than arguing it again.

No procedure appears here. No quantity to weigh, no charge for a vessel, no bath to make up, no apparatus, no sequence detailed enough to follow, and no substitution offered as a way of making the process acceptable. Where a step has to be named for the chemistry to make sense it is named in the past tense, as something people did, and nothing more.

One number is an exception, and it is the exception the policy explicitly allows: the bath temperature the 1839 manual specifies is used below as the ground of a hazard calculation, exactly as the formulary entry uses it, because the argument for the classification cannot be made without it. It is printed as evidence about a hazard and not as a direction, and no other operating condition accompanies it.

The omission is not caution about the chemistry. The chemistry is set out here more fully than anywhere else in the course. The omission is the conclusion of the assessment the page ends by showing you how to run.

Where this page sits

Four pages in this course carry parts of the daguerreotype, and they divide the subject by what question is being answered, which is worth having straight before you start.

Who owns which question about the daguerreotype

1. 1Part I: how did this come to exist?Daguerre the showman, the Niepce partnership, the cupboard story and its provenance, the 1839 announcement and pension, the English patent, the accelerator race and the gilding, and the place of the latent image in the history of the medium
2. 2The formulary: what does the manual actually say?The five operations of the 1839 book, the two aqueous solutions it publishes as quantities, what Arago and the chemists around him could and could not explain, and the fixing chemistry of the least soluble silver halide
3. 3The process atlas: what am I looking at?Identification, the image and its ambivalence, permanence, and the atlas fields a conservator would want filled
4. 4This page: what is happening, and what did it cost?The mechanism at the level of the layer and the particle; why the material was as slow as it was; the vapour as a physical developer that reduces nothing; the toxicology as arithmetic rather than as reputation; and what to do with a plate you own

So the history is Part I’s, the manual is the formulary entry’s, identification is the atlas entry’s, and the Becquerel variant has an atlas entry of its own that this page extends rather than repeats. Mercury acting on a developed silver gelatin image — the intensifiers and the mercury toning baths — is a different chemistry on a different material and belongs to Part XX’s Level D page and to the heavy-metal treatments. Cyanide as a whole working practice is its own lesson in this part.

The plate is the sensitive material

Every other photographic material in this course is a coating on a support. Gelatin holds silver halide crystals against paper or film; collodion holds them against glass; a salted paper print has silver chloride precipitated among the fibres. The support carries the sensitive substance and is not itself involved.

The daguerreotype has no coating at all. A sheet of copper faced with pure silver is polished until, in the Library of Congress’s words, it looks like a mirror, and then the outermost skin of that mirror is converted:

2 Ag + I₂ → 2 AgI

Iodine vapour meets metallic silver: the surface of the support becomes the sensitive layer

Nothing has been added to the plate except iodine. There is no binder between your eye and the image, which is why a daguerreotype resolves detail that nothing else of its century approaches, and it is also why the finished picture is a surface effect on a mirror rather than a stain in a layer. Polishing quality set image quality directly, with no intermediate to hide behind, which is why the manuals give more space to the polishing than to anything else and why the 1840 practitioner’s pamphlet bound after Daguerre’s own book is largely about tripoli, pumice and grease.

Silver iodide is a light yellow solid that darkens gradually in light, and it is by a wide margin the least soluble of the three photographic halides — 2.8 × 10⁻⁶ g per litre at 25 °C. Both properties matter later: the colour is how the operator judged the layer, and the insolubility is why fixing the plate was the hardest version of a problem that recurs through Part XI.

How thick was the yellow? Two 1839 measurements that disagree

Here is a question nobody asks about the daguerreotype and everybody should, because the answer decides how the process could possibly have worked.

The operator judged the sensitised layer by its colour. Daguerre’s specification for the operation is not a time and not a weight; it is a hue. A fine golden yellow, not carried on to violet, because a violet layer is less sensitive, and not stopped short of yellow, because a layer that has not reached it reproduces the image only with great difficulty. The Library of Congress describes the sensitised plate as taking on a yellow-rose colour. The manual says the colour takes five to thirty minutes to arrive and rarely longer, but the operation was not run to a time. It was run to an appearance.

And two chemists weighed it. Arago’s report, printed in Daguerre’s own book, records Dumas’s weighings, from which the thickness of the gold-yellow layer does not appear to reach a millionth of a millimetre — one nanometre — and adds that the plate gains no appreciable weight in taking the iodine on, though it gains very noticeably under mercury vapour.

Those two observations do not fit together, and saying so is not pedantry about an old book.

🧮What a colour tells you about a thickness, and every assumption in the argument

A thin transparent film on a reflecting substrate produces a colour by interference: light reflected from the top surface and light reflected from the bottom surface travel different distances, and which wavelengths reinforce and which cancel depends on that difference. OpenStax gives the bookkeeping — a half-wavelength phase shift on reflecting at an interface beyond which the refractive index is higher, none at an interface where it is lower, and conditions on twice the film thickness at perpendicular incidence.

The consequence that matters here needs no algebra beyond one line. For a film to reinforce or cancel a particular visible wavelength, its optical thickness has to be comparable with that wavelength. Taking green light at about 550 nm and a refractive index somewhere between 1.5 and 2.5, which brackets any plausible value for a solid salt:

t ≈ λ / (4n) → 550 nm ÷ (4 × 1.5) = 92 nm, down to 550 nm ÷ (4 × 2.5) = 55 nm

Film thickness for a first-order interference colour at normal incidence

Of the order of fifty to a hundred nanometres, then, for the first order, and thicker still for the orders above it. OpenStax states the opposite limit explicitly: where a film is very thin and the path difference is negligible, the two reflections are exactly out of phase and destructive interference occurs at all wavelengths — the film shows no colour at all.

So a layer one nanometre thick cannot produce an interference colour. One nanometre is about one five-hundredth of the wavelength of green light. If the yellow-to-rose-to-violet progression that every period operator used as a gauge is interference, the layer is at least fifty times thicker than Dumas’s weighing implies.

Assumptions, stated so you can attack them. Perpendicular incidence, which a plate held up to a window is not. A refractive index bracketed rather than sourced, because the course has read none for silver iodide. And a metal substrate, which complicates the phase bookkeeping in a way this argument does not attempt — reflection from a metal does not give a clean zero or half-wavelength shift. None of those changes the exponent, which is why the course states an order of magnitude and not a thickness.

Could the colour be something other than interference? Partly. Silver iodide is itself a light yellow solid, so a first yellow could be the substance’s own body colour rather than a film effect. But an absorbing layer that grows thicker gets deeper in the same hue; it does not run yellow, then rose, then violet. A colour sequence through the spectrum as thickness increases is the signature of interference, and it is the same Newton’s-scale progression OpenStax describes for a wedge of air.

⁇What the course concludes, and what it marks open

Established: the operator judged the layer by a colour sequence; Dumas’s weighing gave under a nanometre; the plate gained no weight the balance of 1839 could resolve from the iodine and gained noticeable weight from the mercury.

The course’s own inference, labelled as one: the two cannot both be right about a film that shows interference colours, and the physics favours the colour. The likeliest reading is that the balance was at the edge of what it could do — Arago’s own wording is careful, does not appear to reach — and that the layer was tens of nanometres rather than one.

Open: the course has read no modern measurement of the sensitised layer’s thickness on a daguerreotype plate. That measurement exists somewhere in the conservation-science literature; it is not in this course’s corpus, and nothing here should be read as substituting for it.

Why it is worth the trouble: this is the whole light-absorbing volume of the material. A material’s speed cannot be discussed without knowing roughly how much of it there is.

Why it was so slow

The daguerreotype was, in 1839, an enormous gain in sensitivity — the 1837 contract with Isidore Niépce claims that it reproduced objects sixty to eighty times more rapidly than the process his father had invented, and Part I works through why. Set beside any material this course teaches you to make, it is desperately slow: three to thirty minutes in the camera at Paris, by the manual’s own account, three or four minutes in June and July against seven or eight in April and September, and twenty minutes for a subject entirely in half-tone even in the best months.

Four reasons, and they are worth separating because only two of them are about the developer.

There is almost no sensitive material. Even on the generous reading above, the silver iodide is a film of tens of nanometres on a mirror. A modern gelatin emulsion is micrometres thick, with crystals stacked through its depth rather than spread across a surface, so it presents far more absorbing material over the same area. Photons that are not absorbed do nothing at all, and a film that thin transmits or reflects most of what reaches it.

It responds to almost none of the spectrum. The AIC’s conservation record states it flatly: sensitivity is confined to blue and ultraviolet. Everything from green to red passes through the process as if the camera were shut. A modern panchromatic film uses the whole visible band; the daguerreotype used the short end of it, which is also the end where the sky is bright and a face in shadow is not. That single fact explains more about daguerreotype portraiture — the strained expressions, the necessity of direct sun, the difficulty of shadows — than any other on this page.

Nothing has been done to the halide beyond making it. Part V spends a whole part on what turns freshly precipitated silver halide into a fast material: ripening to grow the crystals, digestion with sulfur and gold to build sensitivity centres, dye sensitisation to extend the response past blue. A layer of silver iodide grown in situ by fuming a metal has had none of it. There is no chemical sensitisation because there is no emulsion to sensitise.

And there is no binder, which may matter more than it sounds. Ware’s account of silver photolysis makes the general point that the halogen liberated when light splits a silver halide has to be taken up by something, or it recombines with the photolytic silver and the exposure is undone; gelatin is the halogen acceptor that does that job in a modern emulsion. A daguerreotype plate has no binder of any kind. What became of the iodine liberated during a daguerreotype exposure, and whether recombination limited the plate’s speed, is a question the course poses and does not answer — no source read here addresses it, and the obvious candidate acceptor is the slab of metallic silver underneath, which would simply make more silver iodide. That is reasoning about a real gap, not a finding.

⚗The accelerators: what is established and what is not

Bromine from late 1840 and chlorine from 1841 cut the exposure from minutes to seconds, and the record of that is solid. Goddard published his result in a letter dated 12 December 1840; Eder insists the credit is shared with Kratochwila in Vienna, who had it in September 1840 and reported at least a fivefold gain, exposures of a few seconds and portraits made on cloudy days in eight seconds. Claudet described iodo-chloride sensitising in May 1841.

Why a mixed halide surface should be faster than a pure iodide one, this page does not explain, because no source in the course’s corpus explains it for the daguerreotype. What the course can do is point at where the analogous question is treated with evidence: the behaviour of mixed silver halides in an emulsion, and why iodide in a bromide crystal changes its sensitivity, is Part IV’s and Part V’s material, and it is not transferable to a fumed metal surface without an argument nobody here has made. Period claims that a particular accelerator formula was a trade secret with a special virtue are salesmanship and are not repeated as chemistry.

Eder records one effect that runs the other way and is worth holding, because it constrains any account of the latent image on these plates: exposing a plate that already carries a latent image to iodine, bromine or chlorine vapour destroys it, though the plate will take a new exposure and develop from that. Gaudin reported it in 1841; Shaw and Percy two years later.

The developer that reduces nothing

This is the section the rest of the course needs, because mercury development sits outside the scheme Part IV builds for everything else.

Part IV divides development in two. In chemical development the image silver comes out of the crystal it lands on, reduced in place by a developing agent that gives up electrons. In physical development the image silver comes out of the solution, which contains silver ions that plate onto the latent-image speck; a developing agent still reduces them. Both are redox. Both add silver to silver.

Mercury development is neither, and the difference is not a technicality.

Three ways to turn a latent image into a picture, by where the image metal comes from and what happens to it

further from the ordinary account of development

1. Chemical development — the metal comes from the crystalA developing agent hands electrons to silver ions in the crystal the latent image sits on. Silver(I) becomes silver(0); the agent is oxidised. Part IV puts the amplification factor at of the order of 10⁷ to 10⁸, about 2 × 10⁸ silver atoms per crystal from a latent-image cluster of about fourredox; silver onto silver
2. Physical development — the metal comes from the bathThe developer itself carries silver ions, which are reduced onto the latent-image speck from outside. Still redox, still silver onto silver, but the particle shape and the image colour differ, which is why physically developed plates read grey or tanredox; silver onto silver
3. Mercury development — a second metal arrives as a vapourMercury vapour condenses on the plate and alloys with the silver where light had acted. Mercury enters as mercury(0) and stays mercury(0); the silver it joins is already silver(0). No electron changes hands and nothing is reduced. There is no developing agent, because there is nothing to develop in the chemical senseno redox; a different metal

Say it plainly, because it is the sentence this page exists to put on the record. The first developer in the history of photography was not a reducing agent, and mercury development is not a redox reaction at all. Both metals are in the zero oxidation state before the amalgam forms and both are in it afterwards. What happens is alloying: mercury dissolves into silver, as it dissolves into many metals, and the alloy is the picture. NIOSH’s own incompatibility list for mercury names calcium with the reason in brackets — amalgam formation — alongside lithium, rubidium and copper. The reaction that makes a daguerreotype is the reaction that ruins laboratory equipment.

⚗Why no equation for the amalgam appears on this page

Every other mechanism in this course that can be written as a balanced equation is written as one. This one is not, and the reason is Rule 1 rather than delicacy.

A silver amalgam is an alloy of variable composition, not a compound of fixed stoichiometry, and no source this course has read gives the composition of daguerreotype image particles — not the AIC’s conservation record, which says silver and mercury amalgam and gives a particle size, not the CCI’s, which says the same thing in fewer words, and not Arago’s report, whose chemists could weigh the plate but not analyse the globule. Writing Ag + Hg -> AgHg would look like chemistry and would be an invention: it asserts a one-to-one compound nobody has established.

What can be stated without inventing anything is the direction of the weighings. Arago records that the plate gains no appreciable weight from the iodine and gains very noticeably under mercury vapour, so a substantial mass of mercury is arriving. And Pelouze’s result is stranger and better: after the hyposulfite wash the plate weighs less than before the whole operation began, amalgam and all, and chemical examination of the liquid shows the hyposulfite is removing silver. The fixer takes more away than the developer put on.

Why the mercury goes where the light went

This is the question the nineteenth century could not settle and the course does not settle either. Eder records four accounts from people with plates in front of them: Arago’s chemical one, in which light reduces silver iodide to metallic silver which then takes up mercury; Donné’s physical one, in which the exposed layer goes powdery so mercury reaches the metal beneath; Moser’s condensation argument of 1842, supported by his demonstration that a fully exposed plate could be developed with steam; and Choiselat and Ratel’s silver subiodide of 1843. The atlas entry and the mercury page set them out; they are not restated here and this course adjudicates between none of them.

What this page can add is the shape of the problem in modern vocabulary, offered as framing rather than as an answer. A vapour condensing preferentially at particular sites on an otherwise uniform surface is a nucleation problem: condensation needs somewhere to start, sites differ in how easily they let it start, and a site that starts first grows at the expense of its neighbours by depleting the vapour around it. Moser’s steam experiment is exactly the observation that would make a nucleation account attractive, because water and mercury share nothing chemically and everything about condensation. Whether the nucleating site is photolytic silver, a disrupted iodide layer or something else is precisely the question the four accounts disagree about, and putting a modern word on the disagreement does not resolve it. The course states this as the shape of an open question and not as a mechanism, and it will not say more than the sources support merely because a plausible story is available.

Arago’s closing sentence on the subject is the model for how to hold it: thousands of fine drawings may be made with the daguerreotype before its mode of action has been completely analysed.

One detail has never been explained at all and is worth keeping visible for that reason. The plate is carried at forty-five degrees to the rising vapour, and Arago records that if it were horizontal when the mercury precipitated, the picture would have to be looked at at forty-five degrees to show its maximum effect. He speculates about crystalline needles orienting themselves and claims nothing. A page that quietly dropped that sentence would make the process look better understood than it was.

What the image is made of, and why it has no densities

Two measurements of the image particles, separated by the better part of two centuries, and they are not the same measurement.

Dumas and Adolphe Brongniart put a microscope on a fresh plate in 1839 and found the lights and half-tones formed of spherules very regularly one eight-hundredth of a millimetre across — 1.25 µm. The AIC’s modern conservation record gives image particles ranging from 0.1 to 50 µm. The second is a range over a whole population of surviving plates, gilded and ungilded, made by many hands; the first is what two chemists saw in the first year, on plates made by one. The course reports both and reconciles neither.

🧮How much metal is in one globule, and what that says about amplification

Part IV computes the amplification of chemical development from sourced numbers and gets of the order of 2 × 10⁸ silver atoms per developed crystal, from a latent-image cluster of about four. It is worth doing the equivalent sum here, because the answer contradicts the obvious guess.

Take Dumas and Brongniart’s 1.25 µm spherule as a sphere, and NIOSH’s specific gravity for mercury, 13.6, as a stand-in for the amalgam’s density, which no source gives:

V = (4/3)π(0.625 µm)³ = 1.02 × 10⁻¹² cm³

Volume of one image globule

N = (1.02 × 10⁻¹² × 13.6 ÷ 200.6) × 6.022 × 10²³ ≈ 4 × 10¹⁰

Atoms in it, at 13.6 g/cm³ and a relative atomic mass of 200.6

Of the order of ten billion atoms in one image particle, then — some two orders of magnitude more metal per developed site than Part IV’s figure for a developed gelatin crystal.

Assumptions, all of them arguable. A sphere rather than the flattened cap a condensing droplet probably forms. Pure mercury’s density for an alloy that contains silver, which pushes the atom count in one direction and the atomic mass in the other. One period diameter rather than the modern range, whose top end is forty times wider and so encloses some sixty thousand times the volume. And, decisively, no latent-image cluster size for a daguerreotype exists in this course’s corpus, so this is metal per particle and not an amplification ratio; Part IV’s number is a true ratio and this one is not.

What survives the assumptions is the qualitative conclusion, and it is the useful one. The daguerreotype’s slowness is not a shortage of amplification. Whatever starts a globule, an enormous amount of metal ends up on it. The speed problem is upstream: too little absorbing material, responding to too little of the spectrum, with nothing done to it to make it more sensitive.

The particle size is also why the finished plate has no tonal scale in the sense the rest of this course uses the phrase. Ware’s account is the exact one: virtually none of the light falling on a daguerreotype is absorbed by it. The amalgam globules scatter light diffusely; the bare polished silver reflects it specularly. A print is a pattern of densities — of light absorbed and not returned. A daguerreotype is a pattern of two different ways of returning light, which is why the same plate reads positive from one angle and negative from another, and why photographing it destroys the property by choosing one angle. Part I draws the geometry and owns the explanation; the consequence that belongs here is a conservation one, and it appears below: the mat, the cover glass and the case are not decoration but the apparatus that gives the plate a controlled dark surround to reflect.

Taking the iodide off, and putting gold on

Two operations close the process, and both are owned in detail elsewhere. What belongs here is why each one was harder than it looks.

Fixing. Daguerre states the purpose of his last operation in a clause that could define the word for the whole century that followed: to remove the iodine which, if the picture were exposed too long to light, would go on decomposing and destroy it. The difficulty is that silver iodide is the least soluble halide in the subject, so the first fixing problem anyone faced was the worst one available. Warm saturated salt was the first answer and Eder is blunt about it — an imperfect fixation which gave the plates a mottled appearance. Hyposulphite of soda, on Herschel’s 1819 chemistry, was the second, and is in Daguerre’s own published manual as the preferable alternative. The complex chemistry is Part XI’s and the manual’s own account is the formulary entry’s. Potassium cyanide entered later daguerreotype practice as a fixer and a cleaner, and that whole family of working habits is the cyanide lesson’s and the cyanide formulary entry’s.

Gilding. An ungilded plate is so fragile that the AIC states, without hedging, that the image can easily be wiped off. Fizeau’s bath of hyposulphite of soda containing gold chloride, which Eder dates to 1840 and the AIC to 1841, was adopted almost universally because it improved both beauty and permanence; Fordos and Gélis identified the double salt shortly afterwards, and it became sel d’or, the ancestor of the combined toning-and-fixing baths of Part XX.

The mechanical half of that is obvious once you know what the image is: loose metal globules resting on a polished surface, bound down by a deposit laid over them. The optical half is where care is needed. Any deposit laid over a particle whose entire function is to scatter light must change how it scatters, so gilding cannot be neutral for the way the plate reads. But the sources say beauty and permanence, which is an aesthetic judgement and a durability claim, and the course has read no measurement of what gilding does to a daguerreotype’s tonal range or to the contrast between its scattering and specular areas. The reasoning is offered as reasoning; the measurement is marked absent.

The room the pictures were made in

The hazard argument that follows is not about a substance on a shelf. It is about an operation, and the operation happened in a particular kind of room, so the room is worth describing.

Eder’s account of commercial daguerreotypy gives the developing apparatus as a wooden box with a saucer-like iron bottom, an alcohol lamp under it and a thermometer standing inside. The mercury sat in an open capsule in the bottom of that box, deep enough to cover the thermometer’s bulb; the plate was held above it, face down, at forty-five degrees. The sensitising apparatus was a second closed box with iodine in the bottom. Both were opened by hand, repeatedly, to look at the plate — that is how the process was controlled, because neither operation had a time, only an appearance. Both were in a small dark room, because the plate is light-sensitive from the moment the iodine touches it. Neither box was connected to anything that took air out of the building.

By 1841 that room was a business. Petzval’s portrait objective, calculated in Vienna in 1840, and the bromine and chlorine accelerators between them turned an exposure of minutes into an exposure of seconds, and Eder dates the boom in portraiture from the lens. The commercial pressure ran entirely one way: a faster plate meant a sitter who could hold still, which meant a portrait, which meant a fee. Every improvement in the 1840s made the accelerator step more elaborate and the mercury step more frequent.

🕰Nobody in that room was being careless by the standards of their trade

It matters to say this before the arithmetic, because the arithmetic is severe and it is easy to read it backwards. Elemental mercury had been in apothecaries’, gilders’ and instrument-makers’ hands for centuries, and the operation looked, to anyone standing in the room, like warming a dish of a heavy liquid that does not smell of anything. The course has read no period source warning daguerreotypists about mercury vapour, and will assert neither that such a warning existed nor that none did; what it can say is that the vapour hazard was not characterised in anything like the terms used below until long afterwards, because the terms themselves — a workplace exposure limit, a biological monitoring value, an IDLH — are twentieth-century inventions.

Rule 5 of this course’s brief exists exactly here. Safety overrides historical authenticity, and the judgement it licenses is about what we now know and will therefore not do, not about what they were.

The harm: the arithmetic, and then the evidence

What the vapour does at the concentrations the operation produced

Two regulators, two figures, and they say different things because they are answering different questions.

HSE’s EH40 gives mercury and its divalent inorganic compounds, measured as mercury, a long-term workplace exposure limit of 0.02 mg/m³, with no short-term figure at all — and then does something it does for only a dozen substances in the whole list: it sets a biological monitoring guidance value, 20 µmol of mercury per mol of creatinine in urine. A limit measured in the air says what the control must achieve. A limit measured in the worker says the regulator is not willing to rely on the air measurement alone.

NIOSH’s entry for mercury compounds other than the organo-alkyls sets a recommended limit for mercury vapour of 0.05 mg/m³ with a skin notation, a ceiling of 0.1 mg/m³ for other mercury compounds, and a concentration immediately dangerous to life or health of 10 mg/m³ as mercury. Its exposure routes are all four — inhalation, skin absorption, ingestion and contact. Its symptom list is mostly neurological: tremor, insomnia, irritability, indecision, headache and lassitude, with the central nervous system and the kidneys among the target organs.

Now put a number on the source term, because a limit only means something beside the concentration the operation actually generates.

🧮Saturated mercury vapour, computed from NIST's fitted data, with every assumption named

No source this course holds states the airborne mercury concentration in a daguerreotypist’s room. Nobody measured it and nobody could have. What can be computed is the saturation concentration: the most a closed volume of air in equilibrium with liquid mercury can hold at a given temperature. That is an upper bound on the source, not a prediction of a dose, and the distinction is the whole of the caveat at the end.

Step 1 — the vapour pressure. NIST’s Chemistry WebBook gives Antoine parameters for liquid mercury, fitted by NIST to Hicks’s 1963 evaluation, valid from 298.14 to 749.99 K:

log₁₀ P = A − B / (T + C), A = 4.85767, B = 3007.129, C = −10.001

Antoine equation, with P in bar and T in kelvin

At 298.15 K (25 °C) that gives 0.264 Pa. Two independent checks that this is the right order: PubChem carries the ILO-WHO safety card’s figure of 0.26 Pa, and NIOSH’s 0.0012 mmHg works out at 0.16 Pa, which is what the same equation returns at 20 °C. The three agree once the difference in temperature between them is allowed for — though NIOSH’s Pocket Guide does not print the temperature its value belongs to, and the course does not supply one on its behalf.

Step 2 — pressure to concentration. The ideal gas law, with mercury vapour taken as monatomic at a relative atomic mass of 200.59:

c = PM / RT = (0.264 Pa × 200.59 g/mol) ÷ (8.314 J mol⁻¹ K⁻¹ × 298.15 K) ≈ 0.021 g/m³ = 21 mg/m³

Saturated vapour concentration

Step 3 — read it against the limits. Twenty-one milligrams per cubic metre is about twice NIOSH’s immediately-dangerous-to-life-or-health concentration and about a thousand times EH40’s long-term limit. That is at 25 °C, with nothing heated and nothing happening — a dish of mercury standing open in a sealed room.

Step 4 — warm it. The 1839 manual specifies a bath brought to 60 °C, the figure the formulary entry prints once for this exact purpose. The same equation gives 3.56 Pa there, and the same gas law gives:

c(60 °C) = (3.56 × 200.59) ÷ (8.314 × 333.15) ≈ 0.258 g/m³ = 258 mg/m³

At the bath temperature the manual specifies

Twenty-six times the IDLH, and about thirteen thousand times the long-term workplace limit. Forty degrees of warming multiplies the vapour pressure by 13.5 and the saturated concentration by 12.

Now the assumptions, because a bound is only as good as they are. Saturation is an equilibrium and a real room is not at equilibrium: air moves, the surface area of the pool is finite, and evaporation takes time, so the concentration a person actually breathed was lower — how much lower, nobody can now say. The ideal gas law at these pressures is not the weak link. The Antoine fit’s stated range begins at 298.14 K, so the 20 °C check above is a short extrapolation below it, which is why the 25 °C figure is the one quoted. And a bound is not a dose: this arithmetic says what the source was capable of, not what entered anybody.

What survives all of that is the exponent, and it is not close. The saturated concentration over an unheated pool already exceeds the concentration NIOSH calls immediately dangerous to life or health, and warming it to the temperature the process requires puts it more than an order of magnitude above that again. There is no arrangement of gloves, goggles and an open window that closes a gap of four orders of magnitude.

Saturated mercury vapour concentration against temperature, computed from NIST's Antoine parameters

• Saturated mercury vapour — the most air in equilibrium with liquid mercury can hold at that temperature

Show the numbers behind this plot

A curve rising steeply from left to right. At 25 degrees Celsius the saturated concentration of mercury vapour is 21 milligrams per cubic metre; it reaches 95 at 45 degrees, 258 at 60 degrees and 639 at 75 degrees. A horizontal guide line at 10 milligrams per cubic metre marks the concentration NIOSH calls immediately dangerous to life or health; the curve begins above that line at the left-hand edge of the plot and never returns to it. The long-term workplace exposure limit of 0.02 milligrams per cubic metre is about a thousand times lower than the lowest point on this curve and cannot be drawn on the same axis at all.

Series	Temperature (°C)	Saturated concentration (mg/m³)
Saturated mercury vapour	25.00	21.00
Saturated mercury vapour	30.00	32.00
Saturated mercury vapour	35.00	46.00
Saturated mercury vapour	40.00	67.00
Saturated mercury vapour	45.00	95.00
Saturated mercury vapour	50.00	134.00
Saturated mercury vapour	55.00	187.00
Saturated mercury vapour	60.00	258.00
Saturated mercury vapour	65.00	352.00
Saturated mercury vapour	70.00	477.00
Saturated mercury vapour	75.00	639.00

Computed by the course from NIST's fitted Antoine parameters and the ideal gas law, not measured. EH40's long-term limit of 0.02 mg/m³ is not drawn because it is a thousand times below the bottom of the curve; on this axis it is indistinguishable from zero, which is the clearest statement of the problem available.

Two properties of the substance finish the argument, and both are on the mercury page.

The vapour has neither colour nor smell. Every other Level D substance in this course announces itself somehow — cyanide by the acid two trays away, dichromate by its colour, bromine by being impossible to ignore. Mercury announces nothing. Nothing about the room tells you what is in it, which is why NIOSH sets its respirator ladder against the vapour rather than against the liquid, and why the regulator asks for urine samples rather than trusting the air.

It does not dissolve in water and it does dissolve, to some extent, in fat. A metal that will not go into water and will go into lipids has a route into a body that intuition does not predict.

And now the evidence, which is thinner than the story

Everyone who writes about daguerreotypes says that the operators were poisoned by mercury. It is one of the most-repeated claims in the popular history of photography. This course has looked for the evidence behind it and has to report what it found.

⁇What the historical record supports, and what it does not

This course has read no study, survey or conservation-department account establishing the incidence of mercury poisoning among daguerreotypists. Not a contemporary medical report, not a modern epidemiological reconstruction, not an institutional history. The historical sources behind this page — Eder’s history, Daguerre’s manual with Arago’s report on it, two museum accounts of the medium and two conservation records — make the claim nowhere at all, let alone document it.

That is worth setting against a case where the record does deliver, because the contrast shows the absence is not simply an artefact of what survives from the nineteenth century. Towler’s Silver Sunbeam of 1864 records the occupational harm from cyanide in plain terms: silver stains can be washed off the hands with cyanide of potassium, he says, but doing so “might entail upon the operator incurable ulcers”, and the health of operators is much impaired, “especially in those large printing establishments where a number of females are employed in this department, who by continual manipulation in the toning and fixing baths are frequently in a suffering condition.” A working manual naming a harm, a mechanism and an affected group. Nothing of that kind for mercury has reached this page.

Three things follow, and only three.

The absence of a study is not evidence that nobody was harmed. Chronic elemental mercury poisoning presents as tremor, insomnia, irritability, indecision, headache and lassitude — NIOSH’s own list — which is a description of a difficult middle-aged man in any century and would not have been attributed to a trade. A hazard that is invisible, odourless and produces non-specific neurological symptoms over years is close to the worst possible case for nineteenth-century recognition, and its under-reporting is what one would expect whether or not it occurred.

The absence of a study is also not permission to assert. This course does not repeat a claim as established history because it is repeated everywhere. If the studies exist — and they may well — they are not in this corpus, and a reader who finds one has found something this page needs.

And the classification does not depend on the answer. Not by a little: not at all. The arithmetic above is built from a 1963 vapour-pressure evaluation and two present-day regulators, and it would read exactly the same if the daguerreotypists of the 1840s had all lived to ninety. Level D is assigned on the hazard and the available controls, never on a body count. Making the classification hang on the historical record would be the wrong argument even when it happens to point the same way.

The Becquerel variant, and why removing the mercury does not remove the level

Named in Part I, classified in its own atlas entry, and taken up here because this is the lesson a reader arrives at asking about it.

The AIC records a variant dating from 1840 in which the exposed plate is brought out by red light instead of by mercury vapour, needing no mercury and no bromine or chlorine sensitising, at the cost of plates around ten times slower — which makes portraiture difficult. Eder records that the same red-light action on an already-exposed iodised plate was found independently by Draper in 1842, Lerebours in 1846 and Claudet in 1847, so the effect is real and not isolated. The course still gives no procedure for it, for a reason its atlas entry states and this page will repeat once: sensitising a plate means iodine vapour in a box that has to be opened by hand, EH40 gives iodine a short-term limit of 0.1 ppm, and removing the worst hazard from an operation is not the same as controlling it.

That is worth dwelling on for one paragraph, because it is the trap this whole part is built to teach a reader to avoid. Faced with a Level D process, the natural move is to hunt for the substitution that rescues it, and Becquerel’s variant is precisely the substitution one would hunt for: same plate, same picture, no mercury. The Level D policy forbids “a substitution that makes it safer” as a route to a procedure, and the reason is visible here. A partial removal of a hazard is not a control. The remaining operation is a vapour-phase halogen step in a small dark room with no extraction, and it is not made acceptable by being the smaller half of what used to be there.

There is a second reason, and it is an evidence gap rather than a hazard. The atlas entry cannot say what a Becquerel plate’s image substance is, how a conservator distinguishes one from a mercury-developed plate, or whether it is as stable. Neither can this page. A process whose image material is unknown to the course is not one the course could discuss the permanence of, let alone teach.

The object in your hands

The most likely way a reader of this course meets a daguerreotype is not in a museum. It is a small hinged case in a box of somebody’s effects, and the question is what to do with it.

!A named institution's advice for the owner of a cased photograph

The Canadian Conservation Institute publishes a leaflet for exactly this situation — CCI Notes 16/1, Care of Encased Photographic Images, by Klaus B. Hendriks, revised by Joe Iraci, 1986 and revised 1995 and 2007. Its substance, as the course reads it:

What the object is. Inside the case is a package that can be removed in one piece, and it usually has four parts: the image-bearing plate, a window mat or spacer, a cover glass, and a gilt frame of pliable brass. The image itself is microscopic particles of silver amalgam on the silver face of a silver-plated copper sheet.

Handling. Cased photographs should be handled only with protective lintless nylon or cotton gloves. The Library of Congress adds the general rule for all photographs: freshly washed hands, clean lint-free cotton or inert plastic gloves, and never touch the image surface. CCI’s reason for the second is stark — the unprotected surface of a daguerreotype is sensitive to the slightest touch.

Cleaning. Loose surface dust can be removed with a soft brush or with pressurised air. No other cleaning method can be recommended. Glass and frame, once separated, can be cleaned with a cotton swab and a mild soap solution, rinsed and dried; the plate cannot, and the object must not be immersed in water.

Storage. Relative humidity between 30 and 50 per cent and never above 60; temperature no higher than 24 °C and ideally below 21 °C, not fluctuating by more than 4 °C in a day; individual document boxes.

Read that beside the chemistry of this page and the advice stops sounding like a list of rules. The image is loose metal particles lying on a mirror, held down by a gold deposit if the plate was gilded and by nothing at all if it was not, and the AIC’s sentence — the image can easily be wiped off — is literal. The sealed package is not packaging. It is the instrument that keeps the image surface away from everything, and that gives the plate the controlled dark surround it needs in order to read as a positive at all. Tarnish is a chemical process at the exposed edge of a silver mirror, and the seal is what slows it. Opening a sealed package to look, or to clean, is the single most damaging thing an owner can do, and it is irreversible in the most exact sense: there is nothing to put back.

⁇The mercury and cyanide question, answered as honestly as the sources allow

A reader who has followed the arithmetic above will reasonably ask whether an object made of amalgam is a hazard to the person who owns it. The honest answer has three parts and the course will not pad it.

The mercury is certainly in the object. Two independent conservation bodies say so as a matter of identification, not of speculation: the image particles are a silver–mercury amalgam. That is not a residue or a contaminant; it is the picture.

Whether a housed plate presents a health risk to a person, this course cannot say, and neither CCI’s leaflet nor the AIC’s conservation record raises the question. CCI’s leaflet for owners — the one written for precisely this reader — gives handling, cleaning and storage advice and says nothing at all about mercury as a hazard to a person; it asks for gloves for the object’s sake. The course has read no measurement of mercury emission from a daguerreotype plate and states nothing about it in either direction. Amalgamated mercury is a different physical situation from a pool of the liquid metal, and the arithmetic on this page is about a pool of the liquid metal; it does not transfer, and it is not offered as though it did.

Cyanide is a possibility and not a finding. Potassium cyanide entered later daguerreotype practice as a fixer and a cleaner, so a given plate’s unknown history may include it, and there is no way to tell by looking. The course has read no measurement of cyanide residues on historic plates and asserts none.

And the practical conclusion is the same on every reading. Leave the package sealed. Keep it dry, dark, cool and stable. Handle it by the case, gloved. Do not clean the plate, do not immerse it, do not open it to see what is underneath, and take conservation advice before doing anything that is not on CCI’s short list. That is the right advice for the picture’s sake, and it is also what a precautionary reading of the mercury would give — which is a convenient coincidence and not the reason for it.

Running the assessment yourself

Part XX sets out the four questions this course actually applies when it puts a process at Level D, and the closing assignment of this part will ask you to run them on a process you have not met. Here they are run on this one, so you can see what a completed assessment looks like when the answer is not in doubt.

1. What are the hazards, from a named source with a date? Mercury: harmonised CLP classification, signal word Danger, H330 fatal if inhaled, H360D, H372, H400, H410. EH40, 0.02 mg/m³ long-term as mercury, with a biological monitoring guidance value. NIOSH, 0.05 mg/m³ recommended for the vapour with a skin notation, IDLH 10 mg/m³. Iodine: EH40 short-term limit 0.1 ppm. Bromine: EH40 0.1 ppm long-term, 0.2 ppm short-term. Not a reputation and not a period manual’s reassurance — four classifications and a regulator’s table, all dated.

2. What control addresses each one, and can a domestic reader assemble it? The governing hazard is an invisible, odourless vapour whose saturated concentration over the warmed bath is of the order of 10⁴ times the long-term limit. The control that addresses it is engineered extraction with monitoring — air sampling to show the control is working, and, on the regulator’s own reading, biological monitoring because the air sampling is not trusted alone. A domestic reader can assemble neither.

3. Is any hazard one the sources cannot characterise? Not for mercury, which is unusually well characterised. But note where the gaps are anyway, because the assignment will reward finding them: the amalgam’s composition, the image layer’s thickness, the mechanism of selective condensation and the Becquerel plate’s image substance are all unknown to this course, and a process with that many open questions is one whose behaviour cannot be predicted even where its hazards can.

4. Is there a route to the same photograph that this course can teach? No, and this is the one place the daguerreotype differs from most Level D entries. Uranium toning has a substitute; the chromium(VI) bleach has one; the raw selenium has a bottle. There is no route to a daguerreotype that is not a daguerreotype, because the object’s whole character — a unique direct positive, ambivalent, on a mirror — comes from the operation the hazard lives in. The Becquerel variant removes the mercury and keeps the halogens and the classification. That is the honest shape of this case: the course is not withholding a process it could give you in a milder form. It is declining to hand you the only form there is.

Recap

• The plate is the sensitive material, not a support for it. Iodine vapour converts the outermost skin of a polished silver mirror to silver iodide, so there is no binder between the eye and the image — which is why the detail is extraordinary and why the picture can be wiped off.
• The layer’s thickness is an unsettled question with real consequences. Dumas’s 1839 weighing gives under a nanometre; the interference colours the operators steered by require tens of nanometres at least. The course states the conflict, favours the optics, and marks the modern measurement as one it has not read.
• The slowness was a light-capture problem, not an amplification problem. Very little absorbing material, sensitivity confined to blue and ultraviolet, and none of the ripening, chemical sensitisation or dye sensitisation that Part V exists to teach — while a single image globule holds of the order of 10¹⁰ atoms of metal.
• Mercury development is not a redox reaction. No agent, no electron transfer, no reduction: a metal vapour condenses and alloys with a metal surface where light had acted, and the alloy is the picture. Why it goes where the light went was argued four ways in the nineteenth century and is not settled here.
• The image has no densities. Scattering globules against a specular mirror, which is why the plate reads positive or negative according to the angle, why the case and mat are functional, and why a photograph of a daguerreotype cannot show you the thing itself.
• The hazard argument is arithmetic, not reputation. Saturated mercury vapour is about 21 mg/m³ at 25 °C — already twice the IDLH — and about 258 mg/m³ at the bath temperature the manual specifies, some thirteen thousand times EH40’s long-term limit, from a vapour with no colour and no smell.
• The claim that daguerreotypists were poisoned is not established by anything this course has read. Towler’s cyanide testimony of 1864 shows that period manuals did record occupational harm where they saw it, so the silence is worth noticing; but silence is not evidence of absence either, and the classification does not depend on it in either direction.
• A daguerreotype you own should stay sealed, dry, dark and unhandled. CCI’s advice is dust removal and nothing else; the mercury is in the object by construction and no source read here characterises a risk from a housed plate in either direction, which is a reason for advice rather than alarm.

Check your understanding

Question 1. Why does this course refuse to print a balanced equation for the formation of the daguerreotype image, when it prints equations for almost every other mechanism it teaches?

• Because the reaction is too dangerous to describe in detail
• Because a silver amalgam is an alloy of variable composition rather than a compound of fixed stoichiometry, and no source the course has read gives the composition of daguerreotype image particles
• Because the mechanism involves free radicals that cannot be written in ordinary notation
• Because the amalgam decomposes before it can be analysed

Show the answer and why

Answer: Because a silver amalgam is an alloy of variable composition rather than a compound of fixed stoichiometry, and no source the course has read gives the composition of daguerreotype image particles

Writing something like Ag + Hg to AgHg would look like chemistry and would be an invention: it asserts a one-to-one compound that nobody has established. The conservation records that identify daguerreotypes say silver and mercury amalgam and give particle sizes; they do not give a composition, and Arago's chemists could weigh the plate but not analyse a globule. What the course can state without inventing anything is the direction of the weighings: no appreciable gain from the iodine, a noticeable gain under mercury, and then Pelouze's stranger result that after the hyposulfite wash the plate weighs less than it did before the whole operation began.

Question 2. In what sense is mercury vapour development neither chemical development nor physical development as Part IV defines them?

• It is physical development, because the metal comes from outside the plate
• It is chemical development, because the silver iodide is reduced by the mercury
• It is neither: no electron changes hands. Mercury arrives as mercury(0) and joins silver(0), so nothing is oxidised or reduced and there is no developing agent at all
• It is a special case of physical development in which the solvent happens to be a gas

Show the answer and why

Answer: It is neither: no electron changes hands. Mercury arrives as mercury(0) and joins silver(0), so nothing is oxidised or reduced and there is no developing agent at all

Part IV's two categories both involve a developing agent handing electrons to silver ions: in chemical development the ions come from the crystal, in physical development from the bath. Mercury development has no ions and no agent. Both metals are in the zero oxidation state before the amalgam forms and both are in it afterwards; what happens is alloying, the same process NIOSH lists among mercury's incompatibilities under the heading amalgam formation. Calling it physical development is defensible as loose shorthand because the image metal arrives from outside, but the redox half of the definition is simply absent, and that is worth knowing before you try to reason about it using developer chemistry.

Question 3. Daguerre's operators judged the sensitised layer by its colour, and Dumas weighed the same layer at under a nanometre thick. Why can both not be right?

• Because silver iodide is colourless, so any colour must come from the substrate
• Because a film that thin has a negligible optical path difference, so its two reflections cancel at every wavelength and it shows no interference colour at all; producing a visible colour needs an optical thickness comparable with the wavelength, of the order of tens of nanometres
• Because a nanometre-thick layer would be transparent and therefore invisible
• Because Dumas measured a gilded plate and the operators judged an ungilded one

Show the answer and why

Answer: Because a film that thin has a negligible optical path difference, so its two reflections cancel at every wavelength and it shows no interference colour at all; producing a visible colour needs an optical thickness comparable with the wavelength, of the order of tens of nanometres

OpenStax states the very-thin-film limit explicitly: with a negligible path difference the two reflections are exactly out of phase and destructive interference occurs at all wavelengths, so the film is simply dark. To reinforce or cancel a particular visible wavelength the optical thickness has to be comparable with that wavelength, which puts the film at roughly fifty to two hundred nanometres for the first orders whatever plausible refractive index is used. One nanometre is about a five-hundredth of the wavelength of green light. Silver iodide does have its own pale yellow body colour, so a first yellow could be absorption, but an absorbing layer deepens in the same hue as it thickens rather than running yellow to rose to violet, and that spectral progression is the signature of interference.

Question 4. The saturated concentration of mercury vapour at 25 °C works out at about 21 mg/m³, against a NIOSH IDLH of 10 mg/m³. What exactly does that calculation establish, and what does it not?

• That anyone in a room containing mercury receives a dose immediately dangerous to life or health
• That the source is capable of producing concentrations well above the IDLH even unheated, which is an upper bound on the air rather than a dose to a person, because a real room is ventilated and not at equilibrium
• That the NIOSH figure is wrong, since ordinary rooms containing mercury thermometers were not lethal
• That mercury vapour is only hazardous when the liquid is deliberately heated

Show the answer and why

Answer: That the source is capable of producing concentrations well above the IDLH even unheated, which is an upper bound on the air rather than a dose to a person, because a real room is ventilated and not at equilibrium

Saturation is an equilibrium: it is the most that air in a closed volume can hold at that temperature, and it says what the source is capable of, not what entered anybody. A real room has moving air, a finite pool surface and a finite time, so the concentration breathed was lower by an amount nobody can now reconstruct. What the bound does establish is the size of the gap the controls would have to close, and warming to the bath temperature the manual specifies takes the saturated figure to about 258 mg/m³, some thirteen thousand times EH40's long-term limit of 0.02 mg/m³. No combination of gloves, goggles and an open window closes four orders of magnitude, which is the argument for the classification.

Question 5. The claim that daguerreotypists were poisoned by mercury appears in almost every popular history of photography. How does this course handle it?

• It repeats the claim, since it is universally accepted
• It rejects the claim, since no evidence was found for it
• It reports that no study or institutional account establishing the incidence is in its corpus, notes that the symptom pattern would have been almost impossible to attribute in the nineteenth century, and states that the classification does not depend on the answer either way
• It treats the claim as settled because the exposure calculation makes harm inevitable

Show the answer and why

Answer: It reports that no study or institutional account establishing the incidence is in its corpus, notes that the symptom pattern would have been almost impossible to attribute in the nineteenth century, and states that the classification does not depend on the answer either way

Rule 7 requires established chemistry, historical evidence and interpretation to be kept apart, and this is one of the places where the popular literature runs them together. The corpus behind this page contains Eder, Daguerre with Arago's report, two museum accounts and two conservation records, and none of them makes the claim at all. The contrast with Towler's 1864 record of cyanide harm, which names a mechanism and an affected group, shows the absence is not simply an artefact of what survives. But absence of a study is not evidence that nobody was harmed either: chronic elemental mercury poisoning presents as tremor, insomnia, irritability, headache and lassitude, which no nineteenth-century observer would have connected to a trade. And the Level D classification rests on a 1963 vapour-pressure evaluation and two present-day regulators, so it would read identically if every daguerreotypist had lived to ninety.

Question 6. Why does the Becquerel variant, which develops the exposed plate by red light and needs no mercury at all, remain Level D in this course?

• Because red light of that intensity is itself an eye hazard
• Because the sensitising step is unchanged: it is a vapour-phase iodine operation in a small dark room with no extraction, and removing the worst hazard from a process is not the same as controlling it
• Because the plates are ten times slower and the course only teaches efficient processes
• Because the variant was never published in enough detail to be reproduced

Show the answer and why

Answer: Because the sensitising step is unchanged: it is a vapour-phase iodine operation in a small dark room with no extraction, and removing the worst hazard from a process is not the same as controlling it

The Level D policy forbids offering a substitution as a route to a procedure, and the Becquerel variant is the clearest case in the course of why. It is a genuine improvement — the AIC records that it needs no mercury and no bromine or chlorine sensitising — and it still leaves an operator opening a box of iodine vapour by hand, repeatedly, in a dark room, with EH40 setting a short-term limit of 0.1 ppm for iodine. A partial removal of a hazard is not a control. There is a second reason too: the course has read no account of what a Becquerel plate's image substance actually is, so it could not discuss the material's behaviour or permanence even if the hazard were answered.

Check my answersClear