Herschel's 1842 cyanotype

Every manual that prints “Herschel’s original cyanotype formula” prints it without a citation, and Mike Ware explains the omission bluntly: nowhere in Herschel’s published work are the strengths and proportions given. The 1842 paper says “about equal proportions” and stops. The numbers below come from the other half of the record — the manuscript Memoranda, where on 13 August 1842 Herschel wrote down what he had actually mixed. They are not the formula the twentieth century inherited, and the difference is the most interesting thing on this page.

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

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

Ferrosesquicyanuret of potash, saturated cold — the precipitant

Ingredient	Quantity	Form the source specifies
Potassium ferricyanide	33 g	A saturated solution at room temperature, which Ware gives as about 33 per cent w/v. Not a weighed figure of Herschel's: he wrote "satd sol".
Water	to make 100 mL	Saturation is a strength, not a recipe. The 100 mL is the schema's requirement; what Herschel specifies is that the solution holds all the salt the water will take up cold.

Mixed in the ratio — Herschel's mixing of 13 August 1842

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

Mixed sol of F3/2CP and Ammo Citr of Iron as in Paper 780 ie. equal parts of satd sol of F3/2CP and ACI 1/11 (1 salt + 10 water).

Ware's Table 4.1 gives the resulting mixed sensitiser as 4.7 per cent ammonium ferric citrate and 16.6 per cent potassium ferricyanide, the highest ferricyanide of any formulation in his survey of a hundred and fifty years of recipes.

Mixed in the ratio — Herschel's mixing of 16 August 1842, three days later

5 parts Ammonio-citrate of iron, one part of the salt to ten of water + 3 parts Ferrosesquicyanuret of potash, saturated cold

On 16 August 1842, Herschel modified the proportions of mixing his solutions of ammonium ferric citrate to potassium ferricyanide from 1 : 1 to 5 : 3, respectively, and there is no record of his making further changes after that.

Ware's table gives this mixture as 6.6 per cent citrate and 12.4 per cent ferricyanide. The ferricyanide figure follows exactly from a 33 per cent stock taken at three parts in eight; the citrate figure does not follow from a 9 per cent stock taken at five parts in eight, which would be about 5.9. The discrepancy is reported under Mixing and not reconciled here.

Purpose

To make a negative-working print in Prussian blue on plain paper, using no silver, and to fix it in water. That last clause is the point of the process and the reason its inventor cared about it. Herschel had solved silver fixing in 1839 with the hyposulphites, and he knew what residual thiosulphate does to a silver image over decades. A cyanotype needs no fixer: everything that is not image is soluble, so the wash that develops the print is also the wash that fixes it.

The formula is the second of Herschel’s two cyanotype methods, and the one that survived. The first, which Ware names the proto-cyanotype, was paper washed with potassium ferricyanide alone; it works, and it takes half an hour to an hour of sunshine. Adding the iron salt that Alfred Smee had sent him turned that hour into minutes.

Recommended uses

Contact printing from a negative, or photograms, in direct sun or under a UV source. This is a printing-out process with roughly one ten-millionth of the speed of an ISO 100 film. There is no camera use.

Learning what an iron process is, at its own historical starting point. The chemistry here is the whole of the iron-based family in its simplest form: light makes iron(II), and something else in the paper reacts with the iron(II) to make the image. Change the something else from ferricyanide to a platinum salt and you have the platinotype; to a silver salt and you have the argyrotype and the argentotype.

Where the historical proportions are the point. If you want to see what Herschel’s own prints looked like rather than what a 1900 blueprint looked like, this is the mixture, and Ware’s fading tests confirm it behaves differently from the later ones.

Not where consistency matters. Ammonium iron(III) citrate is an ill-characterised substance whose iron content ranges from 14 to 28 per cent by weight depending on the batch, and the brown form Herschel used is the more variable end of that range.

When another formula is preferable

• For a modern print of the best quality from the same two chemicals, the classic cyanotype sensitiser, which uses Valenta’s green salt of 1897 and a far lower ferricyanide concentration. Herschel’s own mixture is the slowest of all of them for a reason set out under The mechanism.
• For a short exposure, a smooth scale and a maximum density verging on black, Ware’s New Cyanotype of 1995, which replaces the citrate with an oxalate. It costs more, needs a preparation involving heating and filtration, and carries a dichromate.
• For any process that must not lose image density in the wash, almost anything else. Peptization of the image substance during washing is the classic process’s structural weakness.
• For a print that is not blue, either a toner such as the tannic acid toner, which begins by destroying the Prussian blue, or a different process entirely.
• Never for the mercury variant. Herschel’s hydrargyro-cyanotype begins from a print made by this formula and then applies a mercury(I) salt. That page exists to explain the chemistry, and it gives no procedure.

Mixing

Two bottles, mixed only when you are about to coat. The mixed sensitiser has a short life; the separate stocks keep.

1. The ammonio-citrate solution. Dissolve the salt in water at room temperature at Herschel’s proportion of one part of salt to ten of water — 9 g made up to 100 mL is that strength written as a weight and a volume. It dissolves easily; the solid is deliquescent and will have taken up water from the air, which is one more reason the strength of this bottle is approximate however carefully you weigh.
2. The ferricyanide solution. Saturated, cold: add potassium ferricyanide to water until no more dissolves at room temperature. Ware puts saturation at about 33 per cent w/v, so 33 g to make 100 mL is the same instruction with a number on it.
3. Mix equal volumes, as Herschel did on 13 August. Coat, in dim light, and dry in the dark.

The mixture does not precipitate. Herschel noticed that and thought it worth reporting: the two solutions mix without causing any precipitate, and produce a liquid of a brown colour, which washed over paper is green. That absence of a precipitate is chemically informative — see The mechanism.

🧮What the mixture actually contains, and where the published table does not add up

9 % ÷ 2 = 4.5 % citrate · 33 % ÷ 2 = 16.5 % ferricyanide

Mixed one to one

4.7 % citrate · 16.6 % ferricyanide

Ware's Table 4.1, Herschel original

Halving both stocks gives 4.5 and 16.5; Ware’s table prints 4.7 and 16.6, which implies he read the stocks as about 9.4 and 33.2 per cent rather than 9 and 33. That is a rounding, and it does not change anything.

The 16 August mixture is a different matter. Five parts of citrate stock to three of ferricyanide stock puts three-eighths of the ferricyanide into the mixture:

33.2 % × 3 ÷ 8 = 12.45 %, and Table 4.1 gives 12.4 %

Ferricyanide, 3 parts in 8

9.4 % × 5 ÷ 8 = 5.9 %, but Table 4.1 gives 6.6 %

Citrate, 5 parts in 8

The ferricyanide figure follows exactly. The citrate figure does not: to reach 6.6 per cent the citrate stock would have had to be about 10.6 per cent, not 9.4. This course cannot resolve that and does not guess. Either Herschel strengthened the citrate bottle at the same time as he changed the ratio and the Memoranda do not say so, or the two rows of the table were computed from different readings. Ware’s own later fading experiments round the modified formula to 6 per cent citrate and 12 per cent ferricyanide, which is the figure to work from if you want to reproduce his results rather than his table.

Nothing here is heated, and nothing is added in a special order, because there are only two solutions and each has one solute. That is unusual for this formulary and it is one of the reasons the process spread: Herschel accomplished the whole negative-working method, Ware notes, with three test papers.

Behaviour

It prints out. The image appears during exposure, which means it can be inspected and judged rather than timed blind — the practical advantage that made cyanotype the copying process of the drawing office for eighty years.

It reverses if you leave it too long. Herschel’s own Memoranda description is the best one written: the paper darkens to a dull bluish grey from an original green-yellow, reaches a maximum in about thirty or forty seconds in good sun, then whitens again to a pale greenish grey. A print withdrawn at different stages therefore looks completely different, and he sets out three cases: a short exposure gives a negative in bluish grey; a very long one gives a positive, the lit parts paler than the ground; an intermediate one gives an image in three tints at once. He named the reversal solarising, and the word is his.

The blue arrives in the water. Herschel again: when the picture so produced is thrown into water it gradually changes to a fine Prussian Blue Negative Photograph, and the process of the change is highly curious as to Pictorial effect. The wash is where the print-out tone becomes the image.

And some of the image leaves in the same water. This is the failure the classic process is known for. Ware lists it among the shortcomings: a significant proportion of the image substance is peptized and washed out, which truncates the tonal scale at the light end and leaves a print of artificially high contrast. The Prussian blue page explains why a substance that is nominally insoluble comes off in the wash at all.

It goes on darkening after it comes out. Prussian white formed by over-exposure is oxidised back to Prussian blue by air over several hours of drying. Hydrogen peroxide shortens the wait and, Ware states, makes no difference to the final densities.

Image characteristics

Colour: Prussian blue, and on Herschel’s own proportions a deep one, because there is a large reserve of ferricyanide to precipitate against.

Maximum density: limited by solarisation rather than by exposure. Once the shadows begin to reverse, more light stops adding density. Ware makes the useful point that the reversal is not simply a loss — the self-masking of a printed-out blue is reduced when the shadows turn white, which lets more light in and can end in a higher final density after reoxidation.

Scale: shortened in the wash. What comes off the drying line has fewer separated high values than what came out of the printing frame.

Fading: measurable, and this formula is in the middle of the pack. Ware exposed step tablets made with five sensitisers to about 2 kilolux hours of artificial daylight and measured the loss of reflectance density. He expresses it as the fade, 100 times the density change, and reports the maximum fade for each:

Sensitiser tested	Composition, final mixture	Maximum fade (100 ΔD)
Smee — the ferricyanide-only proto-cyanotype	no citrate, 16 % ferricyanide	13
Herschel — his own modified recipe, on Aquapel-sized paper	6 % brown citrate, 12 % ferricyanide	11
Herschel — the same, on gelatin-sized paper	as above	12
Lietze — the usual nineteenth-century recipe	14 % brown citrate, 12 % ferricyanide	10
Valenta — the usual twentieth-century recipe	14 % green citrate, 12 % ferricyanide	10
Ware — New Cyanotype	16 % ferrioxalate, 10 % ferricyanide	22

Two things in that table are worth carrying away. Herschel’s own proportions fade slightly more than the later ones, and the ferricyanide-only process more again. And the newest and best-looking sensitiser fades twice as much as any of them, which is a caution against reading image quality as permanence. Ware also found that the maximum fade occurs across a broad plateau of mid-tones rather than in the deepest blue, so the loss of picture information is greatest exactly where a viewer is most likely to notice it.

Surface: none. There is no binder. The image sits in and on the paper fibres, which is why the paper’s sizing changes the result — note that the same formula fades measurably differently on gelatin-sized paper than on a synthetic-sized conservation stock.

The mechanism

⚗Two reactions, and only one of them needs light

Step one: light reduces the iron. The photochemistry belongs entirely to the citrate. Herschel proved this himself in Article 210 by the elegant expedient of leaving the ferricyanide out: paper washed with the ammonio-citric solution alone, held four or five seconds in the sun with no visible effect, then washed in the shade with ferricyanide, gives a strong blue where the sun fell and nothing anywhere else. In Article 211 he ran a graduated exposure series and watched the blue increase with time and then diminish again. His conclusion is exact and modern: the photographic influence is confined to the first stage, and the ferrosesquicyanate acts as a mere precipitant on the nascent compounds resulting from that influence.

What the light does is reduce iron(III) to iron(II) at the expense of the citrate ligand, which is oxidised and loses carbon dioxide. Ware writes the change for the dimeric citratoferrate(III) anion [Fe₂(C₆H₄O₇)₂(OH₂)₂]²⁻ as a scheme rather than a balanced equation, and the reason is stated in the same sentence: the organic photoproduct has been identified as acetone dicarboxylic acid, CO(CH₂COOH)₂, which may decarboxylate further to acetone, but the nature of the iron(II) photoproduct remains unknown. This course does not print it as an equation, because an equation that does not balance is not one, and completing it would mean inventing the iron species Ware declines to name. What is established is the direction and the cost: absorbed light, iron(III) reduced to iron(II), citrate oxidised, carbon dioxide lost.

Note also that the dimer is the photoactive species — the monomer that predominates below pH 1.5 is photo-inactive — which is why the pH of a cyanotype sensitiser matters at all.

Step two: the iron(II) precipitates the blue. No light is needed, and this is a plain electron transfer between two iron centres:

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

The oxidation states switch

The iron(III) and the hexacyanoferrate(II) then precipitate together as Prussian blue. Ware is careful about the naming: the intermediate ferrous ferricyanide is highly unstable and transforms very rapidly to ferric ferrocyanide, so what looks like a Turnbull’s blue process gives Prussian blue, and the two names are the same substance. The Prussian blue page carries that argument and the compositions.

Step three, if the exposure runs on: the blue is reduced in its turn.

light + Prussian blue + ferrous ions → Prussian white + ferric ions

Solarisation, as Ware states it

Herschel worked this out in 1842 and put it in a footnote to Article 207: the whitening is due to the deoxidation of the precipitated Prussian blue and the formation of the proto-ferrocyanuret of iron; the resumption of colour in the shade, to the re-oxidizement of this compound, which is well known to absorb oxygen from the air with avidity. He then adds the observation that troubled him — Simple Prussian blue, however, is not whitened by the violet rays. Its state must be peculiar. Ware supplies the answer: pure Prussian blue relaxes thermally and comes back unchanged, but if an oxidisable impurity is in intimate contact with it, trapped in the lattice, the excited state has time to accept an electron. That same photoreduction is why cyanotypes fade in light at all.

Deeper: why Herschel’s own formula is the slowest one ever published

Two numbers explain it, and they are both quantum yields — the fraction of absorbed photons that produce a chemical change.

The citrate is efficient. Ware gives a quantum yield of 0.45 at 365 nm at pH 4, falling to 0.28 at 436 nm. Hexacyanoferrate(III) is not: its photolysis quantum yields are of the order of 0.01. The mixed sensitiser is about forty times more efficient per absorbed photon than ferricyanide alone, which is precisely the difference between Herschel’s half an hour or an hour for the proto-cyanotype and thirty or forty seconds to maximum print-out for the mixture.

But the ferricyanide does not merely fail to contribute. It absorbs blue and ultraviolet light strongly while contributing almost nothing to the sensitivity, so it acts as an internal filter, attenuating the printing light on its way to the molecules that can use it. Ware sets this out as the central design conflict of the whole blueprint industry: high speed demands a low ferricyanide concentration, and an intense blue demands a high one.

Now look at where Herschel’s mixture sits. Ware’s survey of some sixty published recipes across a hundred and fifty years finds them clustered at about 10 per cent brown citrate with 8 per cent ferricyanide in the nineteenth century, and about 13 per cent green citrate with 6 per cent ferricyanide in the twentieth. Herschel’s first mixture is 4.7 per cent citrate with 16.6 per cent ferricyanide — the least sensitiser and the most filter of any formulation in the survey. His own revision three days later moved in exactly the direction the whole subsequent history moved: more citrate, less ferricyanide.

Formulation	Citrate, % w/v in the mixture	Ferricyanide, % w/v
Herschel, proto-cyanotype	none	about 16
Herschel, 13 August 1842	4.7 (brown)	16.6
Herschel, 16 August 1842	6.6 (brown)	12.4
Alexander Herschel’s recollection, 1870s	14.7	12.8
Later nineteenth-century average	about 10 (brown)	about 8
Usual twentieth-century average	about 13 (green)	about 6

⁇Whose formula is the 'Herschel formula'?

The disagreement. Herschel’s son, Professor Alexander Stewart Herschel, publicised in the 1870s what he recalled as his father’s working recipe. It is in the fourth row above, and it is nothing like either entry in the Memoranda: three times the citrate of the 13 August mixture. Ware calls the divergence puzzling and reports that Herschel made no recorded change after 16 August.

Who holds what. The manuals that print “Herschel’s original formula” are, in practice, printing either Alexander’s recollection at second hand through Lietze’s 1888 Modern Heliographic Processes, or a twentieth-century average, and they cite nothing. Ware, having read the manuscript, holds that the Memoranda entries are the only direct evidence of what Herschel mixed.

What this course uses and why. The Memoranda quantities, because they are contemporaneous notes in the inventor’s own hand recording what he had that day mixed, and because everything else in the chain is a recollection at thirty years’ distance or an uncited restatement. Alexander’s figures are printed above and are not dismissed — they may well record his father’s later practice — but a son’s memory reported to a photographic newspaper is a weaker document than a dated notebook.

One contemporary check, from a different pen. Robert Hunt’s Researches on Light of 1844 — the first manual of photographic processes, written two years after the paper — restates the negative cyanotype in Article 216 as a mixture which may contain equal proportions of ammonia-citrate of iron and ferrosesquicyanate of potash, with no strengths, exactly as Herschel published it. But in Article 217, describing Herschel’s mercury variant, Hunt does print a strength for the iron bottle: one part by weight of the ammonia-citrate of iron is dissolved in eleven parts of water. That is one in twelve where the Memoranda are one in eleven. The two are not the same number and neither is quoted as the other here; what they agree on is the order of dilution, which is what makes the Memoranda figure credible rather than a transcription slip.

⁇Invention or discovery, and who saw it first

The disagreement. Ware titles his essay on the subject John Herschel’s Cyanotype: Invention or Discovery? and leaves the verdict to the reader. Two prior observations complicate the usual story. Döbereiner reported in 1831 that irradiating an aqueous solution of ferric oxalate precipitates ferrous oxalate and evolves carbon dioxide — the photochemistry of an iron salt of an organic acid, eleven years early. And in 1828 John Mercer, a Manchester calico printer, noted in his notebooks the formation of Prussian blue on cotton by a light-induced reaction, with red prussiate of potash on cloth spotted with a persalt of iron.

Who holds what. The Getty Conservation Institute’s atlas states flatly that the process was invented by Herschel and featured in the 1842 paper. Ware holds that “discovery” fits the 23 April 1842 ferricyanide observation better than “invention”, that Mercer is a candidate for the unsung precursor, and that Herschel’s failure ever to use or cite Döbereiner’s ferric oxalate is a remarkable omission — one Herschel himself later brushed aside, in an 1864 priority defence, as differing from his own processes only by the substitution of oxalic for citric acid.

What this course uses and why. Herschel is the inventor of the cyanotype process: he made pictures, published a method, named the class and put the specimens before the Royal Society. Mercer and Döbereiner observed the underlying photochemistry earlier and did not make a photographic process of it, which is a real distinction and not a slight. The course states both and avoids “X invented photography” phrasing in either direction.

Function of every ingredient

Ammonium iron(III) citrate, 9 g per 100 mL of its own solution, giving about 4.5 per cent in the mixture. What it is: the ammonium salt of an iron(III)–citrate complex, and not really a compound at all — its iron content varies from 14 to 28 per cent by weight between batches and no single molecular formula can be written for it. Herschel had the brown form, the only one that existed; the green form is Valenta’s of 1897 and is what every later recipe means. Why it is here: it is the only light-sensitive thing in the formula. Smee sent it to Herschel as a chemical curiosity — two salts, in his own words, “which of late have been used in Medicine having been vamped up by the Chemists and Druggists”, offered because they gave very dark solutions — and Herschel discovered their photosensitivity immediately. What it does: absorbs near-ultraviolet and blue light and reduces its iron(III) to iron(II) at the expense of the citrate, which is oxidised to acetone dicarboxylic acid with loss of carbon dioxide. Photographic consequence: it sets the speed of the process, and it is the reason this formula prints in minutes where the ferricyanide alone takes an hour. More of it: faster printing, and — on Whitaker’s 1883 testing, reported by Ware — a shorter storage life for the coated paper; carried far enough, excess iron(II) migrates into the unexposed areas during washing and blues the highlights. Less of it: slower, and less blue available in the shadows, which is Herschel’s own mixture. What it interacts with: the ferricyanide, which it must not meet until you are ready to coat; and any alkali in the paper, which will attack the finished image.

Potassium ferricyanide, 33 g per 100 mL as a saturated cold solution, giving about 16.5 per cent in the mixture. What it is: the red prussiate of potash, potassium hexacyanoferrate(III), and in 1842 a novelty — Smee had just shown how to make it in quantity and purity by electrolysing the ordinary yellow prussiate. It is the reason there is a cyanotype at all. Why it is here: to be the precipitant. It waits, unchanged by the exposure, and reacts with the iron(II) the light has made. What it does: accepts an electron from iron(II), becoming hexacyanoferrate(II), while the donor becomes iron(III); the pair precipitates as Prussian blue. It has a small photochemistry of its own — Herschel’s proto-cyanotype depends on it — but with a quantum yield around 0.01 it contributes almost nothing here. Photographic consequence: it sets the maximum blue available, and simultaneously slows the print, because it absorbs the blue and ultraviolet light strongly and acts as an internal filter. More of it: deeper potential blue, longer exposure, and — the reason it matters on this page — Herschel’s own mixture carries the most ferricyanide of any published cyanotype, which is why it is the slowest. Less of it: faster printing but a limited maximum density, and a large excess of iron salt which bleeds into the highlights on over-exposure. What it interacts with: acid, which must never reach it, and light, which slowly decomposes it in the bottle.

Water, twice, and nothing else in the formula. Why it matters: both stocks are simply the salt and water, so the water is doing two jobs — it is the solvent, and in the case of the ferricyanide bottle it is the definition of the strength, because “saturated” means as much salt as this particular water at this particular temperature will hold. Warm water will take up more; a solution saturated warm and used cold will deposit crystals. Distilled or deionised water is the sensible choice for an iron process, because tap water carries its own iron and its own carbonate hardness, and carbonate is the one thing this image cannot tolerate. What happens with more or less: the proportions are the formula; a stock made up at the wrong strength changes both the speed and the maximum density, and there is no development step later to compensate.

What the formula does not contain, and why the omissions are informative. There is no wetting agent, so on a modern internally sized paper the sensitiser sits in the coarse pores rather than penetrating the fibres, and much of it washes away. There is no acid in the processing bath, so the tonal scale is the long, soft one that plain water gives. There is no preservative, so the citrate bottle grows mould. And there is no fixer, which is the whole point: everything unexposed is soluble.

Interactions

The two solutions with each other, before you want them to. They mix without precipitating — Herschel remarked on it — because the iron(III) is locked up in the citrate complex and is not available as free Fe³⁺ to react with ferricyanide. That is the same fact that makes the sensitiser possible and makes it short-lived: nothing is stopping a slow dark reaction, only slowing it.

Iron(II) with ferricyanide, which is the image. Instant, and it does not need light, which is what Herschel demonstrated by developing a latent iron image with a ferricyanide wash after the exposure.

Light with the Prussian blue already formed, which is solarisation during the exposure and fading afterwards. They are the same reaction seen on two timescales.

Alkali with the finished image, which destroys it. Ware reports Holtzman’s finding that a buffer at pH 9.4 completely decolourises Prussian blue by irreversible hydrolysis in one to ten minutes — and pH 9.4 is the pH of a saturated calcium carbonate solution, which is the buffer routinely added to archival board. A cyanotype demands an unbuffered paper and an unbuffered mount.

Air with the print, which finishes it. The Prussian white made by over-exposure oxidises back to Prussian blue over several hours of drying.

Hydrogen peroxide with the print, which finishes it faster. Ware gives 0.3 per cent for about half a minute before the final wash, and states that it does not change the final densities — it only saves the wait.

The paper with everything. More than in any silver process, because there is no binder to hold the image away from the sheet.

Variants

Herschel’s own second mixture, 16 August 1842: 5 parts citrate stock to 3 of ferricyanide. It is printed in the table above and in the data of this page, because it is the same formula at a different ratio and Herschel made no further change. If you print one of these, print this one: it is his settled proportion.

The proto-cyanotype, of 23 April 1842: potassium ferricyanide alone. Article 203 gives it — paper simply washed with a solution of the salt, after half an hour or an hour’s exposure to sunshine, a very beautiful negative photograph is the result, fixed by soaking in water in which a little sodium sulfate is dissolved to ensure the fixity of the Prussian blue deposited. Ware recommends 10 to 16 per cent rather than saturation, so the salt does not crystallise on the surface before the paper absorbs it, and gives one warning that does not apply anywhere else on this page: perceptible amounts of hydrogen cyanide and cyanogen are released during the exposure of ferricyanide alone, so it is not done in a confined space. He states explicitly that the same warning does not apply to the much faster processes using ammonium ferric citrate or oxalate. This course does not publish the proto-cyanotype as a working procedure.

The gum-arabic positive, Article 220. Herschel’s positive-working method washes ferrocyanide over an exposed ammonio-citrate paper, and the image runs. Adding gum arabic to the wash stops it spreading in the pores and lets the Prussian blue agglomerate on the fibres. His own verdict: the manipulations of this process are, however, delicate, and complete success is comparatively rare.

The ferric chloride developer. In the Memoranda of 24 April, Herschel found that very weak iron(III) chloride acidulated with sulphuric acid develops a deep and most superb blue on a ferricyanide paper, and noted in the same breath that the ground is blued too. Ware’s summary is the same: greater speed and higher density, and invariable blue fog in the highlights.

The nineteenth- and twentieth-century averages, in the table above. They are not variants of Herschel’s formula in the schema’s sense — nobody derived them from it — but they are where the process went: more citrate, less ferricyanide, and after 1897 the green salt.

The classic sensitiser and the New Cyanotype have their own entries.

No course variant of Herschel’s own proportions is offered. There is nothing to make safer by changing them, and the honest modern alternative already exists as a separate, better-documented formula.

Safety

Level B, and the classification comes entirely from the potassium ferricyanide. Ammonium iron(III) citrate is Level A and Prussian blue is Level A. Ware’s own assessment of the classic sensitiser is that the chemicals are not dangerous and can be safely handled by children under supervision; this course keeps the ferricyanide’s own page as the governing document and its level as the formula’s.

The specific points that follow, none of which replaces the chemical page:

• Ferricyanide is a cyanide complex, not a cyanide salt, and the distinction is real: the cyanide is bound to iron and is not free. It is not the hazard that potassium cyanide is, and the two are never to be confused.
• It becomes one in contact with acid, which is the reason acid must never enter this workflow. See Incompatibilities.
• The proto-cyanotype exception matters. Ware records that exposing ferricyanide alone to strong light liberates perceptible hydrogen cyanide and cyanogen, and states that this does not apply to the citrate or oxalate processes. That is a specific finding about a specific case, and it is not generalised here in either direction.
• The exposure is the hazard people forget. This is a UV process. Direct sun for the length of a printing exposure is a skin and eye exposure; an artificial UV source is worse, because it is closer and it does not feel warm. Skin cover, and eye protection or an enclosure for a lamp.
• Gloves and no dust. Both solids are handled dry when the stocks are made up. Weigh them without raising dust, and keep the work out of the kitchen: nothing that has held either salt goes back to holding anything you eat or drink from.

Storage

Two brown bottles, in a cupboard or a box, never one bottle. The mixed sensitiser has a short life; that is why the formula is two solutions at all, and it is the single most-repeated practical instruction in the literature.

The citrate bottle grows mould. Ware’s plain statement is that in many environments the surface will be covered by a thick furry growth within a week or two. A few crystals of thymol floated on the surface inhibit it; they do not dissolve, and you avoid them when drawing a sample. Thymol is itself harmful and its own page carries the handling.

The solid citrate is deliquescent, becomes sticky in humid air, and on long storage may compact into an intractable mass. Keep it stoppered and dry.

The ferricyanide solid keeps well; its solution keeps less well, decomposing slowly in light, which is another reason for the brown bottle.

Coated paper is not covered by any figure on this page. Ware reports Whitaker’s finding that an excess of citrate shortens the storage life of coated paper, and nothing in the sources read here gives a keeping time for paper coated with Herschel’s own proportions. Coat what you will print.

Label both bottles with the strength and the date. Neither solution announces its age, and the citrate’s strength was approximate on the day you made it.

Incompatibilities

Acid, with the ferricyanide, absolutely and without exception. No acid stop bath, no acid clearing bath, no vinegar on the bench, nothing acidic in the sink the trays drain into. This is the rule that turns a Level B process into a serious one if it is broken.

Alkali, with the finished print. Chalk-buffered board, alkaline mountboard, alkaline sizing, alkaline tap water. See Interactions for the pH 9.4 figure.

Buffered paper, before you even coat. The same chemistry, arriving earlier: an alkaline reserve in the sheet will attack the Prussian blue as it forms.

Metal, in trays, tongs, clips and sinks. An iron–cyanide chemistry is exactly the thing that finds a trace of iron and shows it as a stain.

Silver, and every silver bath in the darkroom. Ferricyanide is the bleach in Farmer’s reducer for a reason; a contaminated tray or a shared measuring cylinder will attack a silver print.

Strong oxidising agents and strong reducing agents, as for the salt itself; see its own page.

Household bleach and anything chlorine-releasing, in the same room and the same drain, for the same reason as the acid rule.

Waste

Two streams, and neither of them is a silver stream. The exhausted stock solutions are dilute iron(III) citrate and dilute potassium ferricyanide; the wash water carries unreacted sensitiser plus peptized Prussian blue, which is the blue you can see in the tray.

Do not acidify anything, at any stage, including in the waste bottle. This is the disposal consequence of the incompatibility above, and it is why cyanotype waste is not simply “iron salts”.

Collect it rather than discarding it in the wash. The volumes are small but the ferricyanide content is real, and the visible blue in the first wash is image substance, not dye.

Local regulation decides what happens next, and this course cannot tell you what it says where you are. Label the containers with what is in them and follow the general chemical waste SOP and the course’s disposal ruling.

Troubleshooting

No blue at all, or barely any. Most likely under-exposure, since this formula is the slowest published. Check second that the citrate bottle has not been standing so long that its mould has consumed part of the citrate, and third that the paper is not buffered — an alkaline reserve will take the image apart as fast as light makes it.

The blue is pale and flat, and the highlights are empty. The classic peptization failure: too much of the image washed out. It is worse on a heavily sized modern paper that keeps the sensitiser on the surface, and it is what a wetting agent, absent from this historical formula, exists to fix.

The print looks positive, or partly positive. Solarisation, and it is not a fault — it is Herschel’s own observation and he found the three-tint intermediate stage remarkable. If you did not want it, expose less.

The image reversed and then would not come back. Prussian white needs air, and several hours of drying, to reoxidise. Peroxide will do it in half a minute if you cannot wait.

Blue fog across the highlights and the paper base. Either the coated paper was fogged before exposure, or excess iron(II) migrated in the wash — Ware’s “bleeding”, which happens when the citrate is in large excess over the ferricyanide. On Herschel’s proportions the excess runs the other way, so suspect fogging first.

Blue stains in patches with no relation to the negative. Iron picked up from a tray, a clip or a sink; or ferricyanide left standing on the paper where a brush rested twice. Herschel warned about the second in Article 219 — the wash cannot be in too thin a film, and a brush passed twice over any part gives the effect of a coarse and ill-printed woodcut.

A greenish or greyish cast in the deepest tones. Ware calls this “burning out” and attributes it to excess ferrous iron in the heavily exposed regions.

Crystals in the ferricyanide bottle. A solution saturated at a warmer temperature and stored cold. Warm it gently and use it at a consistent temperature, or make it up at the temperature you will work at.

The print keeps changing colour for a day after drying. Expected. Judge nothing until it is dry and has had its several hours.

Experiments

Print Herschel’s two mixtures side by side. One coating at 1 + 1 and one at 5 + 3, from the same two bottles, on the same paper, under the same negative, exposed together. Three days separate them in the historical record and nobody has published what the difference looks like. Record exposure to a fixed step on a step tablet, and the maximum density after drying.

Measure the internal filter. Coat three papers at the same citrate strength with ferricyanide at one-half, one and twice Herschel’s, and expose all three through a step tablet in one frame. The prediction from Ware’s account is a clear trade: the weakest ferricyanide prints fastest and reaches the lowest maximum density. This is the single most useful experiment on the page because it turns a sentence about absorption into a set of numbers.

Test the two-stage claim as Herschel did. Coat paper with the citrate solution alone, expose a strip for five seconds in sun, and wash it over with the ferricyanide solution in the shade. Article 210 is a demonstration you can repeat in ten minutes, and it separates the photochemistry from the image formation more clearly than any diagram.

Run the solarisation series. A strip marked into compartments, exposed in an arithmetic progression as in Article 211, washed and dried. You will produce Herschel’s own figure, and the point at which density stops increasing is the practical ceiling of the process on your paper.

Compare plain water against Ware’s 1 per cent citric acid bath. Two identical prints, one processed in water as Herschel says and one in the dilute acid, judged for maximum density and for highlight clearing. Herschel’s own instruction is water; the acid development is a modern refinement, and the comparison is the honest way to see what it buys and what it costs.

Fade a pair and measure it. Two prints from the same negative, one stored in the dark and one in a window, densitometered at a mid-tone rather than at maximum density — Ware’s finding is that the loss peaks in the mid-tones, and a test that measures only the deepest blue will report a permanence the picture does not have.

Coat one sheet of buffered paper on purpose. Alongside an unbuffered control. It is the fastest demonstration in this course of why an archival material can be the wrong material, and it is the same chemistry as the first bath of the tannic acid toner.