Herschel's hydrargyro-cyanotype

Herschel made an ordinary cyanotype, washed it over with a mercury salt, and watched the whole picture disappear. He rinsed the mercury out, dried the sheet, and ran a hot flat-iron over what was now a blank piece of paper. The picture came back — brown. Left in a portfolio in complete darkness it faded away again over some weeks, and a fresh application of heat brought it back to full intensity. He reported all of this in the Postscript to his 1842 paper and never published the name he had given it; hydrargyro-cyanotype appears only in his manuscript Memoranda, and reaches this course through Mike Ware, who read them.

Two things make this entry unusual in the formulary. The first is that it is Level D: the salt is a mercury(I) compound and the last step of the process is to heat it. The second is an evidence problem that no amount of care can fix. Herschel, Robert Hunt in 1844, Hunt again in 1854 and Ernst Lietze in 1888 all describe the process, and not one of them states how strong the mercury solution was. Ware, who repeated it, does not either. So this page carries quantities for the print the process starts from, and none at all for the bath it is named after.

Ammonio-citrate of iron, one part of the salt to ten of water — the light-sensitive half of the starting print

Ingredient	Quantity	Form the source specifies
Ammonium iron(III) citrate	9 g	The brown salt, the only form that existed in 1842. Nine grams in 100 mL is Ware's "ca. 9% w/v" reading of Herschel's Memoranda entry "ACI 1/11 (1 salt + 10 water)".
Water	to make 100 mL	Herschel gives a proportion and no volume, in the Memoranda and in Article 223 alike. The make-up volume here is the one that turns his proportion into the per cent w/v Ware states; the formula is the strength, not the hundred millilitres.

Ferrosesquicyanate of potash, saturated cold — the precipitant that makes the Prussian blue

Ingredient	Quantity	Form the source specifies
Potassium ferricyanide	33 g	A solution saturated at room temperature, which Ware puts at about 33 per cent w/v. Herschel wrote "satd sol" and no figure.
Water	to make 100 mL	Saturation is a strength rather than a recipe. The 100 mL is what the schema requires; what Herschel specifies is that the solution holds all the salt cold water will take up.

Mixed in the ratio — the sensitiser of the starting print, as Article 223 states it

1 part Ammonio-citrate of iron, one part of the salt to ten of water + 1 part Ferrosesquicyanate of potash, saturated cold

If paper be washed with a mixture of the solutions of ammonio-citrate of iron and ferrosesquicyanate of potash, so as to contain the two salts in about equal proportions, and being then impressed with a picture, be thrown into water and dried, a negative blue picture will be produced.

Herschel's Memoranda record equal parts of the two solutions on 13 August 1842 and a change to five parts of citrate to three of ferricyanide on 16 August, thirteen days before this Postscript was added; which of the two he used for the mercury experiment is not recorded, and Article 223 says only "about equal proportions". Ware's Table 4.1 gives the one-to-one mixture as 4.7 per cent ammonium ferric citrate with 16.6 per cent potassium ferricyanide. **The mercury(I) nitrate bath that follows is not in this list of combinations**, because no source states a strength for it and the schema cannot hold a solution that has none.

!What this table is, and what it is not

It is the sensitiser of the starting print. Ware states that the hydrargyro-cyanotype begins from a cyanotype made in the usual way by Herschel’s standard “Method (a)”, and Method (a) is the formula on the Herschel 1842 cyanotype page. The two solutions above are that formula, repeated here because a process is not legible without its starting material.

It is not the mercury bath, and the mercury bath is not anywhere on this page as a quantity. Herschel writes “solution of proto-nitrate of mercury”. Hunt reprints that phrase twice, ten years apart. Lietze supplies the compound — mercurous nitrate, Hg₂(NO₃)₂·2H₂O — and still no strength. Ware’s own account gives none. Three separate silences and one modern confirmation, all agreeing that the number was never published, is about as strong as negative evidence gets, and the course does not manufacture one to fill the hole. See Mixing.

The safety level does not come from anything in the table. Both salts above are handled at Level A and Level B elsewhere in this course. The D belongs to the substance that is missing from it.

Purpose

To record and explain a documented transformation of a Prussian blue image by a mercury(I) salt and heat, and to say honestly how much of its chemistry is still unknown. That is the whole of the entry. It is here because it is real, because it is in the primary literature of the first years of photography, and because a reader who meets it — in Hunt, in Lietze, in Sutton and Dawson’s dictionary, in a museum caption or in a brown print that nobody can account for — needs somewhere to look it up.

It also earns its place for a second reason. The process is one of the clearest demonstrations anywhere in early photography that an image is a substance, not a picture. The blue that disappears and the brown that appears are not the same material seen in two moods; the first is destroyed and the second is built out of something else that was put into the paper afterwards. Very little else in the course makes that point so bluntly.

☠This page gives no procedure, and that is the rule rather than a caution

Under the Level D policy a page at this level carries no Preparation, no Procedure and no Build section, publishes no sequence to follow and no conditions to hold, and says plainly what is missing and why. What follows is the chemistry, the history and the hazard. The historical workflow appears only in outline, in the past tense, as what Herschel did — never as what you do.

Ware’s own verdict, from a chemist who reproduced the process and had every reason to want to recommend it: he “cannot recommend it generally owing to the high toxic hazard presented by the mercury salts, especially when heated.”

Recommended uses

Reading the nineteenth-century literature without being misled. Hunt lists seven cyanotype processes in 1844 as “the most interesting of this class”, and this is one of them. Ware makes the sharper observation that Hunt does not list Herschel’s plain negative-working cyanotype in its own right at all, “except as a preliminary to this process” — so one of the earliest and most influential English manuals of photography presents the standard cyanotype mainly as a step on the way to bleaching it with mercury. Sutton and Dawson’s authoritative Dictionary of Photography was still doing something similar in its second edition of 1867, describing, on Ware’s account, only three of the more fallible cyanotype processes: the very slow proto-cyanotype, the unreliable positive process, and this one.

Understanding why the best cyanotype process was neglected for thirty years. Ware suggests that a run of “misdirected practical instruction” in the manuals — foregrounding the positive process, or complicating the negative one with a mercury bleach — “may well be responsible for the notable lack of take-up of Herschel’s best cyanotype process” over the three decades after he discovered it. The hedge is his and it is the right one; this is a plausible historical explanation rather than an established cause. This entry is one of the exhibits in that argument, and it is a lesson about technical literature that has not dated.

Learning what the Prussian blue lattice is by watching something get trapped in it. The conjectured mechanism below is the same chemistry as Prussian blue’s modern use as an antidote for thallium and caesium poisoning, and the same chemistry as the lead and thallium cyanotype toners this course also studies and does not use. If the lattice makes sense to you as a cage with cations rattling in and out of it, this process stops being a curiosity.

Reading Herschel on heat. Article 224, which follows immediately, uses this process as the evidence for a claim about radiant heat that he took up again the following year, in the paper whose title names the parathermic rays. It is a good demonstration of a first-rate experimenter reasoning correctly from a phenomenon whose chemistry he had no way to establish.

Not for making a brown print, or a print of any kind. There is no version of this that this course teaches, and the modern route to a brown cyanotype is a different chemistry entirely; see the next section.

When another formula is preferable

• For a print in Prussian blue, the Herschel 1842 cyanotype itself, which is the first three-quarters of this process and stops before the mercury. It is Level B and the course teaches it in full.
• For a modern print of the best quality, the classic cyanotype sensitiser or Ware’s New Cyanotype.
• For a brown or black image made from a cyanotype, the tannic acid toner. It reaches a comparable end by an honest and available route: an alkaline bath destroys the Prussian blue, and a tannin bath puts something else in its place. It is Level B, its quantities are published by a supplier, and the parallel with this page is exact — in both, the blue is destroyed first and the visible image is made of something that was not in the print to begin with.
• For understanding mercury in a photographic image at all, the Kodak IN-1 mercury intensifier, which is also Level D but is documented to a standard this process never was: two published baths, quantities, end points, and a mechanism that the sources actually establish.
• For the same lattice chemistry in another metal, the planned Level D entry on the lead, thallium and nickel cyanotype toners, where the colour shift is attributed to the same trapping of a heavy cation in the same cages.
• Never this, for any practical purpose. The classification is not a difficulty rating.

Mixing

This section gives no mixing procedure, and for the mercury bath it could not give one even if the policy allowed it. There is nothing to reproduce. What can be described without instructing is the shape of the thing, and in this case the shape is mostly a set of holes in the record.

The process is a treatment, not a sensitiser. Everything happens to a finished, washed, dried cyanotype. That puts it in the same structural family as the tannic acid toner — a sequence of baths applied to a completed print — and not in the family of coating solutions.

Nobody has ever published the strength of the mercury solution. This is worth setting out carefully, because it is the single most important fact about the evidence:

Source	Year	What it says about the mercurial bath
Herschel, Philosophical Transactions, Article 223	1842	“washed with solution of proto-nitrate of mercury”
Hunt, Researches on Light, Article 216	1844	“washed with a solution of the protonitrate of mercury”
Hunt, Researches on Light, 2nd edition, Article 281 item 3	1854	the same sentence, unchanged
Lietze, Modern Heliographic Processes, page 53	1888	“A solution of mercurous nitrate, Hg₂(NO₃)₂ + 2H₂O”
Ware, Cyanomicon, §4.1.3	2020	“coated with a solution of mercurous nitrate”

Not one gives a concentration, a volume, a temperature or a method of making the solution up. Herschel does give a strength for the proto-nitrate elsewhere in the same Postscript — his mercurial cream of Article 229 is made with a saturated solution — but that is a different process on a different paper, and reading it across would be the course inventing a quantity out of an adjacent paragraph. It is not done here.

The schema records that gap by leaving the bath out entirely. A formula object holds solutions, and a solution requires a quantity and a make-up volume. A bath whose only published description is the name of its solute cannot be entered without stating something no source states, so it is not entered, and this paragraph exists because a reader is entitled to know that the absence is deliberate. The same shortfall is recorded in the course’s schema notes, where several other pages have hit the general form of it: the schema can hold what a source said and cannot hold what the course did about it.

The sensitiser’s own quantities are second-hand and the course says so. Article 223’s statement is “about equal proportions” and nothing more. The 9 per cent and 33 per cent above are Ware’s reading of a manuscript notebook entry made sixteen days before the Postscript was written, and are attached to this process only because Ware attaches them — he names Method (a) as the starting point. Herschel had also changed his mixing ratio from 1 : 1 to 5 : 3 thirteen days before the Postscript, and there is no record of which he used here.

Nothing is heated in the making of anything. The heat in this process arrives at the very end, applied to dry paper, and it is the step that both makes the picture and creates the hazard.

Behaviour

What follows is what the sources record happening, in the order they record it. It is history, not an instruction, and the numbers in it are observations rather than settings.

The blue disappears, and it takes hours. Herschel: the mercurial wash “in a little time entirely discharges it”. Ware, from his own repetition: “within a period of a few hours the blue image disappears”. The paper is then washed and dried and reads as blank — Herschel calls the picture “obliterated”, and the word is exact, because at this stage there is nothing to see.

Heat brings it back at once, and brown. Herschel’s flat-iron is “somewhat hotter than is used for ironing linen, but not sufficiently so to scorch or injure the paper”, and on its passage “the obliterated picture immediately reappears, not blue, but brown”. Ware describes the same step as heating strongly, short of scorching. Nobody publishes a temperature; a domestic iron of the 1840s has no setting to record.

It fades again in the dark, and the sources disagree about how fast. Herschel: kept “for some weeks in this state between the leaves of a portfolio, in complete darkness, it fades, and at length almost entirely disappears”. Hunt in 1844 is blunter — “however carefully kept, these photographs fade after a few weeks and disappear”. Ware, repeating it in the twenty-first century, reports fading “slowly in the dark over a matter of days”. Herschel and Hunt say weeks and Ware says days — roughly an order of magnitude apart — and not one of the three is a measurement.

And heat brings it back again. Herschel: “a fresh application of the heat revives and restores it to its full intensity.” Lietze, forty-six years later: “After a few weeks, the picture will disappear again, but can be re-developed by renewed ironing.” This is the property that makes the process interesting and the property that no proposed mechanism yet accounts for.

🕰Article 224: Herschel gets the right answer for a reason he could not check

Immediately after describing the process, Herschel used it as an instrument. He took a strip of paper carried as far as the blank stage, and exposed it to “a very concentrated spectrum (collected by a lens of short focus)”. The whiteness “is changed to brown over the whole region of the red and orange rays, but not beyond the luminous spectrum.”

From that he drew three conclusions, and they are worth reading as written:

1st, that it is the heat of these rays, not their light, which operates the change; 2ndly, that this heat possesses a peculiar chemical quality which is not possessed by the purely calorific rays outside of the visible spectrum though far more intense; and, 3rdly, that the heat radiated from obscurely hot iron, abounds especially in rays analogous to those of the region of the spectrum above indicated.

The first is plainly right and he had proved it himself in the same sentence — “a certain temperature must be attained, and that temperature suffices in total darkness”. The second is the seed of what he went on to call the parathermic rays, and it is a claim about radiant heat that he was reasoning towards from a chemistry nobody could then describe. He notes that the same three conclusions had come out of his experiments on guaiacum in Articles 158 to 160, which is exactly how a careful experimenter builds confidence in a result: the same conclusion from an unrelated material.

What this course takes from it. The observation is established: heat alone, in the dark, browns the sheet, and the effective radiation lies inside the visible spectrum rather than beyond the red. The interpretation is Herschel’s own, made in 1842 with no means of establishing the chemistry, and the course does not endorse “a peculiar chemical quality” of heat as a modern statement about anything.

Image characteristics

Colour. Blue, then nothing, then brown. That is the whole of the published description, from every source, and there is no more specific account of the hue anywhere the course has looked.

Density and scale: no measurement exists. Nobody has published a densitometric reading, a characteristic curve, a maximum density or a step-tablet comparison for a hydrargyro-cyanotype. Every statement about how strong the image looks is an adjective in a nineteenth-century sentence, and Ware’s own repetition is reported qualitatively. A page in this formulary would normally carry the numbers here; this one carries the fact that there are none.

Permanence: the worst in the cyanotype family, and worse than the plain cyanotype by a long way. A cyanotype’s characteristic failure is a slow fade in light, over years, which Ware has measured on step tablets. This image fades in darkness, over days or weeks, and darkness is exactly where prints are stored. Whatever the brown substance is, it is not stable at room temperature in a portfolio.

But the fading is not a loss of the image. This is the strangest property of the process and the one that no other entry in this formulary shares. The picture that has faded to nothing can be recovered in full by heating it again, apparently repeatedly. Something survives in the paper that is not visible and is not destroyed, and heat converts it back into whatever is brown.

Surface: none. As with any cyanotype there is no binder; the whole history happens in and on the paper fibres, which is one reason the substances involved are so hard to characterise after the fact.

The mechanism

The honest summary is that nobody knows. Ware states it in as many words — “we can only conjecture what chemistry may be involved here, for the observations deserve further scientific investigation” — and this course is not going to improve on a chartered chemist who repeated the experiment. What can be done is to set out what is established, then mark clearly where the established chemistry stops and the conjecture begins.

⚗What is established: the print before the mercury touches it

Light reduces the iron; the ferricyanide precipitates the blue. The photochemistry belongs entirely to the ammonium iron(III) citrate: absorbed near-ultraviolet and blue light reduces iron(III) to iron(II) at the expense of the citrate ligand. The potassium ferricyanide takes no part in that and simply waits. Then a plain electron transfer between two iron centres:

Fe²⁺ + [Fe(CN)₆]³⁻ → Fe³⁺ + [Fe(CN)₆]⁴⁻

The oxidation states switch

and the iron(III) and hexacyanoferrate(II) precipitate together as Prussian blue. The full argument, including why the unstable ferrous ferricyanide of a nominal Turnbull’s blue process ends as ferric ferrocyanide anyway, is on the Herschel 1842 cyanotype page and on the Prussian blue page.

Prussian blue is a defect lattice with holes in it. Ware’s account of the structure is the piece that matters here: the ‘insoluble’ variety has one quarter of its hexacyanoferrate groups missing with water molecules in their place, and the form the cyanotype actually makes — nominally Turnbull’s blue — has one third of them missing, replaced by water molecules and anions. Crystallographers call these systematic vacancies, and the resulting open framework “gives it a porous quality that explains why the substance is a very useful ‘chemical sponge’ for soaking up and trapping cations similar to potassium, such as those of the metals caesium and thallium.” This is not an analogy. It is why Prussian blue is a licensed medical antidote for caesium-137 and the preferred antidote for thallium poisoning: the thallium(I) cation “can enter the Prussian blue lattice in place of potassium. There it becomes trapped.”

Prussian white is what Prussian blue becomes when it is reduced, and it is colourless. Ware gives its composition as M₂Fe(II)[Fe(II)(CN)₆], where M is a singly charged cation — usually potassium or ammonium — sitting in the cube centres of the lattice. The reduction cannot happen on its own:

light + Prussian blue + reductant → Prussian white + oxidation products

Ware's general form, written for the photochemical case

Ware writes it that way because the case he is explaining there is solarisation, the light-driven fading of a cyanotype. What his conjecture proposes for this process is the same reduction driven by a chemical reductant instead; no source read for this course states whether the mercurial discharge happened in light or in darkness, and the course does not fill that in either. The reverse reaction, in air, is what makes a solarised cyanotype come back blue as it dries:

4Fe[Fe(CN)₆]²⁻ + O₂ + 2H₂O → 4Fe[Fe(CN)₆]⁻ + 4OH⁻

Reoxidation by air, as Ware writes it

Ware adds the detail that decides everything below: the electrochemical reduction of Prussian blue “is fully reversible provided an adequate supply of suitable cations is available to diffuse in and out of the lattice to maintain charge neutrality.”

⁇Where the chemistry stops and Ware's conjecture begins

What Ware proposes, in his own words and marked by him as conjecture:

the reversibility of the phenomenon may be a consequence of the mercury(I) being trapped in the Prussian blue lattice in the same way as thallium(I) is known to be, and then reducing the substance to Prussian white. The effect of heat may be to cause decomposition of the mercury ions to mercury metal, which in a finely divided state would appear brown.

Read against the established chemistry above, it is a coherent proposal and it does two jobs at once. A mercury(I) cation entering the cube centres in place of potassium is exactly the “supply of suitable cations” the reversible reduction requires; and mercury(I) going to mercury(II) is an oxidation, which is the “something else” that has to be oxidised for Prussian blue to be reduced. The same species is both the counter-ion and the reductant. That is elegant, and it is not evidence.

What is missing. There is no analysis of a hydrargyro-cyanotype — no spectroscopy, no elemental map, no diffraction — published anywhere the course has looked. No balanced equation appears on this page because no source writes one, and completing one would mean choosing between mercury(I) species and oxidation products that nobody has identified.

Why the brown is the harder half. Two candidate brown substances are in play and Ware’s own text argues against one of them. Finely divided mercury metal is his suggestion. The other obvious candidate is Prussian brown, the iron(III) hexacyanoferrate(III) of his Appendix II.11 — but the same appendix reports that the brown mixture of iron(III) and hexacyanoferrate(III) “rapidly turns blue when spotted onto cellulose filter paper or any other oxidisable substrate”, because the pair has a redox potential around +1.5 V and oxidises almost anything. A brown that survives on paper for weeks is very unlikely to be that one. This comparison is the course’s, drawn between two passages of Ware’s own book; he does not make it, and it is an argument against a candidate rather than evidence for his.

And the fading in the dark may be the same fact as another of Ware’s. He explains the failure of Herschel’s mercury printing process — the kelainotype — by suggesting that “the metal of the image is sufficiently volatile to evaporate completely within a few days”, and records that the surviving Herschel specimens of that class in Austin are completely blank. If the brown here is mercury metal, its disappearance over days in a closed portfolio is the same phenomenon. That inference is the course’s, joining two of Ware’s statements; he draws it for the kelainotype and not for this process. It also leaves the revival by heat unexplained, which is the fact any successful mechanism has to account for and none yet does.

Deeper: why an equation would be dishonest here

It is tempting to write mercury(I) disproportionating — Hg₂²⁺ giving mercury metal and mercury(II) — because that reaction is real, well known, and would produce brown finely divided mercury on heating. The course does not write it, for three reasons that are worth naming because they generalise.

First, no source read for this page states it happens here. Ware writes “decomposition of the mercury ions to mercury metal” and deliberately does not specify the route. Filling in a plausible one and setting it in an equation block would convert his hedge into the course’s assertion, which is precisely the move Rule 1 forbids.

Second, the iron half would still be unaccounted for. Any equation for the discharge has to say what happened to the Prussian blue and where its electrons came from and went. Ware’s proposal implies mercury(I) is oxidised and the blue is reduced, but the products are not established and the stoichiometry is unknown. An equation that balances only because its author chose the unknowns is not a balanced equation.

Third, the reversibility is a constraint no proposal has satisfied. A picture that fades to nothing and is fully restored by heating, more than once, requires that whatever went away is still in the sheet. Evaporating mercury does not obviously satisfy that; neither does an irreversible decomposition. This is the sharp end of Ware’s remark that the observations deserve further investigation, and the sharp end is where a course stops writing chemistry and says so.

Function of every ingredient

Ammonium iron(III) citrate, 9 g per 100 mL of its own solution, about 4.5 per cent in the mixed sensitiser. What it is: the ammonium salt of an iron(III)–citrate complex, and an ill-characterised substance: Ware’s Table 4.2 puts the iron content of the brown salt at 19 to 28 per cent and of Valenta’s green salt of 1897 at 14 to 18. Herschel had the brown, which is the more variable end and the only form that then existed. Why it is here: it is the only light-sensitive thing in the starting print, and without the print there is nothing for the mercury to act on. What it does: absorbs near-ultraviolet and blue light and reduces its iron(III) to iron(II), the citrate being oxidised and losing carbon dioxide. Photographic consequence: it sets the speed of the exposure and, through the amount of iron(II) it delivers, the quantity of Prussian blue available to be destroyed later. More of it: a faster print and more blue in the shadows, which on this page means more of the substance whose fate the mercury decides. Less of it: a slower, weaker print; carried far enough there is too little Prussian blue for the transformation to show. What it interacts with: the ferricyanide, which it must not meet until the sensitiser is wanted; and any alkali in the paper, which attacks the finished blue directly.

Potassium ferricyanide, 33 g per 100 mL as a cold saturated solution, about 16.5 per cent in the mixed sensitiser. What it is: red prussiate of potash, potassium hexacyanoferrate(III), and in 1842 a novelty: Alfred Smee had described in 1840 how to prepare it by electrolytic oxidation of the commonplace and widely available potassium ferrocyanide, and sent Herschel a specimen early in 1842. Why it is here: to be the precipitant that turns the light’s iron(II) into Prussian blue, which is the image substance this whole page is about. What it does: accepts an electron from iron(II) and becomes hexacyanoferrate(II); the pair precipitates. Photographic consequence: it sets the maximum blue available and, because it absorbs blue and ultraviolet strongly while contributing almost nothing to sensitivity, it also acts as an internal filter that slows the print. Herschel’s own one-to-one mixture carries more ferricyanide than any other formulation in Ware’s Table 4.1, and roughly twice the 8 per cent that his survey of about sixty published recipes finds as the nineteenth-century average. More of it: deeper potential blue, longer exposure. Less of it: faster printing, less maximum density, and a large excess of iron salt that bleeds into the highlights. What it interacts with: acid, which must never reach it — see Incompatibilities — and light, which slowly decomposes it in the bottle.

Water, twice, and it is doing more than dissolving. In the ferricyanide bottle the water is the strength, because “saturated” means whatever this water at this temperature will hold; warm water takes up more and deposits crystals on cooling. In the citrate bottle it is a stated proportion, one part of salt to ten. And in the process itself water appears a third time, as the bath that both develops and fixes the starting cyanotype and later as the wash that removes the mercurial salt — Herschel’s instruction that “the nitrate being thoroughly washed out” is a step whose thoroughness he evidently thought worth stating. Distilled or deionised water is the sensible choice for any iron process, because tap water brings its own iron and its own carbonate hardness, and carbonate is the one thing a Prussian blue image cannot tolerate.

Mercury(I) nitrate — the ingredient this formula is named after, and the one this page cannot put in the table. What it is: mercurous nitrate, Hg₂(NO₃)₂, which Lietze gives as the dihydrate and Ware identifies in his endnote; Herschel and Hunt call it the proto-nitrate of mercury. PubChem records it under CAS 10415-75-5, with 7782-86-7 for the dihydrate. Why it was there: to destroy the Prussian blue image, entirely and in a few hours, leaving a sheet that reads blank. What it does: not established. Ware’s conjecture, marked as such above, is that the mercury(I) enters the Prussian blue lattice as thallium(I) does and reduces the blue to colourless Prussian white; the brown that heat later produces he attributes to decomposition of the mercury ions to finely divided mercury metal. Photographic consequence: the destruction of one image and the eventual appearance of another, in a different substance and a different colour, with a permanence far worse than the blue it replaced. More or less of it: unknown, and unknowable from the record, because no source states how much was used. That is not a rhetorical flourish: the ordinary formulary question “what happens if you use twice as much” has no answer here, and a page that answered it would be making one up. What it interacts with: the Prussian blue, the wash water that has to remove it, reducing agents of every kind — the reactivity profile on its PubChem record warns that mixtures of metal nitrates with phosphorus, tin(II) chloride or other reducing agents may react explosively — and heat, which is both the last step of the process and the reason the process is classified Level D.

What the record does not contain, and why the omissions matter. There is no stated volume, no temperature, no time for the mercurial bath beyond “a little time” and “a few hours”; no statement of whether the solution was acidified, which matters for a mercury(I) salt in water; no account of how much mercury remained in the paper after the wash; and no analysis of the finished object. Each of those absences is a fact about the 1842 record rather than a decision of this course, and together they are the reason a page about a fully described phenomenon can still be a page about ignorance.

Interactions

Mercury(I) with Prussian blue, which is the process. Complete within hours on Ware’s account, and total: not one of the five sources reports any residual blue.

Heat with the treated paper, which makes the picture. Here the developer is temperature: no reagent is added at that stage at all. Herschel’s own demonstration that it is the heat rather than the light is that “a certain temperature must be attained, and that temperature suffices in total darkness”, and he then measured which part of the spectrum could do the same thing.

Heat with a mercury salt, which is the hazard. Inhalation is the route the occupational limits are written against: NIOSH’s entry for mercury compounds sets its tightest figure on mercury vapour and carries the skin notation on every limit in it. PubChem’s fire-hazard block for mercurous nitrate states that smoke from a fire involving it “may contain toxic mercury vapor and oxides of nitrogen”. An ironing step is not a fire, but it is deliberate heating of a mercury salt held in paper, held close to the operator, with no extraction. Ware’s refusal to recommend the process names heating specifically.

A nitrate with paper, which is a fire risk on its own terms. The same PubChem record notes that these substances “will accelerate burning when involved in a fire” and “may explode from heat or contamination”. Herschel’s warning not to let the iron scorch the paper reads, in modern terms, as a warning about an oxidiser-impregnated sheet under a hot iron.

Acid with ferricyanide, the absolute rule of every cyanotype workflow. It is not part of this process, and it is stated here because the process sits inside a cyanotype workflow: acid must never reach a ferricyanide bath or its waste.

Mercury(II) with cyanide, which is a different Herschel process and a useful contrast. Ware notes that mercury(II) “has an enormous affinity for cyanide, forming a very stable complex”, and uses that to explain why Herschel’s positive-working cyanotype could be fixed by including corrosive sublimate in the sensitiser: the mercury(II) destroys the Prussian white in the highlights so that it cannot revert to blue. Two mercury oxidation states, two entirely different actions on the same iron–cyanide chemistry, in the hands of the same man inside a year.

Alkali with any finished Prussian blue image. Ware reports Holtzman’s finding that a buffer at pH 9.4 completely decolourises Prussian blue by irreversible hydrolysis in one to ten minutes. Anything that survives on this page as a blue image survives only on unbuffered paper.

Variants

Herschel’s other mercurial papers, from the same Postscript. Article 228 describes paper washed first with the ammonio-citrate and then with the mercurial proto-nitrate, which “darkens to a very deep brown, nay to complete blackness” on moderate exposure and yields “very sharp and intense photographs of a negative character”. They were, he says, “difficult to fix”; his only partial success was a bichromate wash and a twenty-four-hour soak that left an obstinate yellow tint and still faded on keeping. Hunt gives strengths for a version of this paper in 1844, and this page does not reprint them: it is a different formula from the one this entry is about, and it is a mercury sensitiser rather than a mercury bleach.

The kelainotype, Article 229 — the most extravagant paragraph Herschel ever wrote about a photograph. Mixing the proto-nitrate with the ammoniacal double salts gives a precipitate which, worked to the consistency of cream and spread on paper, produced pictures “of such force and depth of colour, such velvety richness of material, and such perfection of detail… as infinitely to surpass any photographic production I have yet seen, and which indeed it seems impossible to go beyond.” Ware, quoting it, warns the reader to note the next sentence: “Most unfortunately, they cannot be preserved.” Herschel named the class from the Greek for dark, tried it as celaenotype and then, on Talbot’s suggestion, as amphitype; he never made it work, and Ware’s judgement is that the mercury process came to nothing “probably because the metal of the image is sufficiently volatile to evaporate completely within a few days”. The specimens of that class among the Herschel material in Austin are blank sheets. Neither the composition nor the manipulation is reproduced here.

The corrosive sublimate positive cyanotype, from the 1843 paper. Herschel’s second attempt at fixing the positive-working cyanotype put mercury(II) chloride into the sensitiser so that the print would be fixed without any washing at all — “ipso facto fixed as soon as dry”. Ware’s verdict is that “it cannot be recommended, in view of the very high toxicity (and expense) of mercury(II) chloride”, and he adds that ingestion of only 0.05 g of that salt can be fatal. See mercury(II) chloride for the hazard record.

No modern variant of this process exists, and the course offers none. There is no safer mercury and no substitute cation that anybody has published for this specific transformation. The honest modern route to a brown image on a cyanotype is a different chemistry with its own page, the tannic acid toner, and the course recommends that instead rather than pretending this one can be made acceptable.

A note on what is not a variant. Ware records a 2007 proposal that inverted Herschel’s positive-working method and was announced by its author as the “Herschelotype”; Ware calls the attachment of Herschel’s name to it “a clumsy misappropriation”. It has nothing to do with this process and is mentioned only because a reader searching for Herschel’s cyanotype variants will meet it.

Safety

Level D, and the classification comes entirely from the mercury(I) nitrate and from the fact that the last step of the process heats it. The Level D policy owns the rule; the figures below are quoted as a measure of how much control the substance is judged to need, never as a threshold to work to.

The notified classification. PubChem aggregates the GHS classification of mercurous nitrate from 27 reports across 5 notifications to the ECHA C&L Inventory: signal word Danger, pictograms GHS06, GHS08 and GHS09, with H300 fatal if swallowed, H310 fatal in contact with skin and H330 fatal if inhaled, together with H373 for organ damage through prolonged or repeated exposure and H400 and H410 for aquatic toxicity. Fatal by all three routes is the profile potassium cyanide carries; mercury(II) chloride, the salt of the Kodak intensifier and of Herschel’s other mercurial cyanotype, carries H300 and H310 but not H330. The sample here is small — five notifications — and a small sample is worth saying out loud, but it is not a reason to read the classification down.

The occupational limits. NIOSH’s entry for mercury compounds [except (organo) alkyls] (as Hg) explicitly covers all inorganic mercury compounds, which is where a mercury(I) salt sits. It gives a recommended limit of 0.05 mg/m³ as an eight-hour average for mercury vapour with the skin notation, a ceiling of 0.1 mg/m³ for other mercury compounds, also with the skin notation, and an IDLH of 10 mg/m³ as mercury. Its exposure routes are inhalation, skin absorption, ingestion and contact, and its target organs are the eyes, skin, respiratory system, central nervous system and kidneys. HSE’s EH40 gives 0.02 mg/m³ for mercury and its divalent inorganic compounds measured as mercury; the word divalent means that entry does not, on its face, name a mercury(I) salt, and EH40’s own introduction says that absence from the list does not indicate that a substance is without risk. The course reports that gap rather than quietly reading the limit across.

Why heating is the point, and not a detail. Every one of those limits is written against airborne mercury. The process ends by pressing a hot iron onto a sheet impregnated with a mercury salt, at close range, in whatever room the worker is standing in — and then, when the picture has faded, doing it again. Ware’s sentence is the correct one to end on: he “cannot recommend it generally owing to the high toxic hazard presented by the mercury salts, especially when heated.”

The oxidiser is a second, independent hazard. A nitrate held in paper under an iron is an oxidiser in intimate contact with fuel. The record warns of accelerated burning, of explosion from heat or contamination, and of mercury vapour and nitrogen oxides in the smoke.

What is not a hazard here, and why saying so is part of the assessment. The two salts in the quantity table are not the reason for the classification. Ware’s own view of the classic cyanotype sensitiser is that its chemicals are not dangerous and can be handled safely by children under supervision; the ammonium iron(III) citrate page is Level A and the potassium ferricyanide page is Level B, on its incompatibility with acid rather than on its own toxicity, because the cyanide in a ferricyanide is bound to iron and is not free. Prussian blue itself is Level A, and Ware describes it as non-toxic, unreactive and tasteless, safely eaten in gram quantities without harmful effects — which is why it is a licensed antidote for caesium and thallium poisoning. Everything that makes this entry Level D arrives with the mercury.

And a note on objects rather than reagents. The course has found no conservation guidance addressed specifically to hydrargyro-cyanotypes, which is unsurprising: they were rare, they fade in the dark and the surviving population may be close to zero. A print in a collection is a different question from a bench reagent, and this page does not extend a bench assessment to an object, or an object assessment to a bench. Where an institution publishes on the handling of mercury-bearing photographic material, that guidance governs.

Storage

Nothing on this page is stored, because nothing on this page is made. What the historical record does say about keeping is worth recording, because it is the most unusual property of the process.

The finished picture does not keep, and darkness is where it fails. Herschel gives some weeks in a portfolio in complete darkness before it almost entirely disappears; Hunt says a few weeks; Ware says days. Every other image in this course is stored in the dark precisely to preserve it. This one is destroyed there.

The loss is recoverable, which means something is still present. Repeated heating restores the image to full intensity, on Herschel’s account and Lietze’s. Whatever leaves the picture does not leave the paper, or does not leave it quickly.

No keeping figure is published for any solution. Not for the mercurial bath, which has no published composition to keep, and not for the sensitiser as used in this process. The two sensitiser stocks keep separately and the mixture does not, which is the general rule for cyanotype sensitisers and is set out on the Herschel 1842 cyanotype page.

Storing the literature is the only storage instruction this page has. Note the source, the edition and the article number when you record something from it: Hunt’s account is Article 216 in 1844 and Article 281 item 3 in 1854, and a citation that says only “Hunt” cannot be checked.

Incompatibilities

Mercury(I) nitrate with reducing agents. Its PubChem reactivity profile, from Bretherick, warns that mixtures of metal and nonmetal nitrates with alkyl esters may explode, and that mixtures with phosphorus, tin(II) chloride or other reducing agents may react explosively. A darkroom is full of reducing agents; every developer on the shelf is one.

Mercury salts with heat, in any quantity. Stated above under Safety and repeated here because it is the incompatibility that decides the classification.

Mercury with the rest of the darkroom. Mercury contaminates what it touches and is not removed by ordinary cleaning. A tray, a rod, a sink or a pair of tongs that has held a mercury salt is not returned to general use.

Acid with any ferricyanide, absolutely, as everywhere else in this course: no acid stop bath, no acid clearing bath, nothing acidic in the sink the trays drain into.

Alkali with a Prussian blue image, including chalk-buffered board and alkaline mountboard. See Interactions.

Silver, and every silver bath. Ferricyanide is the bleach in Farmer’s reducer, and a contaminated cylinder will attack a silver print.

Waste

The only mercury waste question that can arise from this page is an object, not a bath, because no procedure is given and therefore no solution is made. That is worth stating rather than assuming.

Mercury is its own waste stream, and the mercury page owns that record for this course, including the waste-classification entries it cites. It is not a heavy metal that joins the silver stream, and it is not a dilute iron salt that joins the cyanotype stream. Nothing on this page adds to or qualifies what that page says.

The cyanotype rule still governs anything blue. Do not acidify cyanotype waste at any stage, including in the waste bottle; the visible blue in a first wash is image substance rather than dye, and the ferricyanide content is real.

Local regulation decides what happens next and this course cannot tell you what it says where you are. Label containers with what is in them and follow the general chemical waste SOP and the course’s disposal ruling. A reader who finds mercury-bearing material in a collection or a bequest should treat it as a hazardous-waste question for the relevant authority and not as a darkroom one.

Troubleshooting

On a Level D page this section cannot be about fixing what you made, because you are not going to make it. It is about identifying what you meet.

A nineteenth-century “cyanotype” that is brown rather than blue. Consider this process before concluding that a blue print has discoloured. The distinction matters for conservation: a faded cyanotype and a hydrargyro-cyanotype are different materials with different hazards, and one of them contains mercury.

A blank sheet in an album of early photographs, annotated as a picture. Ware records that some of the Herschel specimens in Austin — especially those derived from mercury — are now completely blank, while others are obviously surviving cyanotype and chrysotype images. A blank sheet with a contemporary caption is evidence, not an empty page.

A brown image that changes when handled or displayed warm. The one thing every account agrees on is that heat is the developer. An object of this class should not be assumed to be stable against warmth, and nothing about it should be tested with heat.

A source that gives a strength for the mercurial bath. If one is ever found, it will be new information, and the first question to ask is whether it is a nineteenth-century printing or a modern reconstruction presented as one. This course has looked at Herschel 1842, Hunt 1844, Hunt 1854, Lietze 1888 and Ware 2020, and found the same silence in all five.

A reference to Sutton and Dawson, or to the 1845 Practical Manual of Photography. Both are reported by Ware as carrying this process, and neither has been read for this course. Anything attributed to them here is attributed to Ware’s account of them, and a reader with access to either would be doing the course a service by checking what they actually say.

Experiments

Nothing involving mercury is proposed, at any scale, in any form. What follows are experiments that address the same questions with materials the course does teach, plus the questions that remain genuinely open.

Heat an ordinary cyanotype, as a control nobody appears to have published. Take two identical prints from the Herschel 1842 cyanotype formula, iron one at the temperature Herschel describes — hotter than for linen, short of scorching — and keep the other. There is no mercury anywhere in this test; it is a domestic iron on plain paper, and the only cautions are the ordinary ones about a hot iron. Compare them for colour, density and any subsequent change. Every account of this process assumes the brown depends on the mercury; the control that shows heat alone does nothing to a plain cyanotype is not in any source read here, and it is a morning’s work.

Reduce Prussian blue to Prussian white and bring it back, without leaving Level A. Solarise a cyanotype deliberately by gross over-exposure, watch the shadows go pale as Prussian white forms, then follow the reoxidation in air over several hours of drying, or accelerate it with the 0.3 per cent hydrogen peroxide bath of about half a minute that Ware gives, which he states makes no difference to the final densities. This is the reversible half of the lattice chemistry the mercury conjecture depends on, and it is directly observable.

Repeat Herschel’s Article 224 experiment on a plain cyanotype. Focus a solar or lamp spectrum onto a strip with a short-focus lens and record which region does what. Herschel’s claim was about a sheet that had been through the mercury bath, so this is not the same experiment — but establishing what the spectrum does to an ordinary cyanotype is the missing baseline against which his result would have to be read.

Read the primary sources rather than the summaries. Herschel’s Postscript of 29 August 1842 runs from Article 217 to Article 230; reading it whole shows how the cyanotype, this process, the mercury papers and the kelainotype sit in one continuous run of experiments over a single extraordinary summer. Hunt’s two printings, ten years apart, show what the manual literature did with it.

🧪The questions a properly equipped laboratory could still settle

None of these is a home experiment and none should be attempted with mercury outside a laboratory with the controls Ware’s warning implies. They are recorded because Ware states that “the observations deserve further scientific investigation”, and a reference work should say what the investigation would be.

What is the brown substance? Elemental analysis and diffraction on a heated sample would separate finely divided mercury metal from any iron–cyanide candidate in a single afternoon, and that question has been open since 1842.

Where does the mercury sit before heating? If the lattice-trapping conjecture is right, a mercury(I) species should be detectable in the Prussian white phase in a cation position, and its disappearance on heating should be measurable.

What survives the fade? The revival by reheating is the constraint no proposal satisfies. A faded sample that can be restored contains everything it needs; identifying what that is would settle the mechanism more directly than any measurement on a fresh one.

How much mercury remains in the paper after washing? This is a conservation question as much as a chemical one, and it applies to any surviving object of this class in a collection.