The Siderotype Principle: Iron Reduces Silver

Take the two numbers a chemistry table gives you for this reaction and it should barely happen. The standard potential of the iron(III)/iron(II) couple is +0.771 V and that of silver(I)/silver(0) is +0.80 V; the difference is about 0.03 V, which is nothing. On those figures an iron(II) ion in water is only just able to reduce a silver ion, and any small change of conditions would stop it.

Yet Herschel got “a very sharp and beautiful picture … of great intensity” in two or three minutes, and every Van Dyke and every kallitype since has worked. The reason is that neither of those numbers describes the iron in your bottle. The iron is wearing a ligand, and the ligand moves the potential by three quarters of a volt. That single fact explains why the process works, why swapping citrate for oxalate changes the process rather than merely tuning it, why a kallitype can be toned with palladium and a cyanotype cannot, and why the developer in the next lab is not a developer in the sense Part VIII taught you.

The two steps, with silver in the second

Nothing about the first step is new. Part XXI established it, and Ware’s statement of it covers the whole family: under ultraviolet light, iron(III) held by an organic acid reacts with that acid to give iron(II) and carbon dioxide.

2 Fe³⁺ + C₂O₄²⁻ + UV → 2 Fe²⁺ + 2 CO₂

Step one, dissected: two electrons leave one oxalate and land on two iron(III) centres

Ware notes that this “dissected” form is written to show the redox bookkeeping rather than the actual species. In the dry solid the reaction is better written with the compounds themselves:

Fe₂(C₂O₄)₃ + UV → 2 FeC₂O₄ + 2 CO₂

The same reaction as it happens in a dry ferric oxalate coating

and in a solution of the well-characterised complex salt, with the complex ions:

2 [Fe(C₂O₄)₃]³⁻ + UV → 2 [Fe(C₂O₄)₂]²⁻ + C₂O₄²⁻ + 2 CO₂

And in solution, where the photoactive species is the trisoxalatoferrate(III) anion

Two things about that step are worth carrying forward. The colour change is slight — Ware describes it as pale yellow-green to pale yellow-brown — which is why the print-out image in a kallitype is nearly useless as a guide. And the result is not permanent on its own: atmospheric oxygen re-oxidises iron(II) back to iron(III), so an exposed sheet left long enough loses what the light did. The photoproduct has to be caught by something.

Fe²⁺ + Ag⁺ → Fe³⁺ + Ag

Step two: the electron goes on to silver, and the iron is exactly where it started

Count the electrons, because the count is the process. Silver(I) needs one electron to become silver metal, so one iron(II) makes one silver atom. Ware writes the ratio out explicitly when calculating siderotype absorbances: Fe:Ag is 1:1, where Fe:Pt(II) and Fe:Pd(II) are both 2:1 because those are two-electron reductions. Iron-silver is the cheapest member of the family in electrons as well as in money.

⚗What happened to the iron, and why the part is arranged around the answer

Look at the right-hand side of step two. The iron is back to iron(III), in the paper, in exactly the places where the picture is. It was a reagent, it was consumed by light and regenerated by silver, and it is now sitting in a sheet of cellulose next to a colloidal metal it is thermodynamically capable of oxidising again.

That is not a footnote. It is why the clearing lab exists, why this part’s break/fix page has “would not clear” in its title, and why Ware’s one-sentence diagnosis of the whole family is that its inherent problem “lies in the danger of leaving residual ferric iron in the print — to its ultimate undoing”.

Why the reaction goes, and the answer is the ligand

Here are the numbers, all from Ware’s Platinomicon, which collects them for exactly this argument.

Couple	Potential	What it means here
Fe³⁺/Fe²⁺, free aqua ions	+0.771 V	The textbook value. Iron(III) is a moderate oxidising agent
Fe(III)citrate/Fe(II)citrate	+0.372 V	The Van Dyke’s iron. Now a weak reducing agent
Fe(III)(C₂O₄)₃³⁻/Fe(II)(C₂O₄)₂²⁻	+0.02 V	The kallitype’s iron. A moderate reducing agent
Fe(III)EDTA/Fe(II)EDTA	−0.12 V	Not a sensitiser — this is the clearing bath, and it matters later
Ag⁺/Ag	+0.80 V	What has to be reduced
PtCl₄²⁻/Pt	+0.73 V	Part XXV
PdCl₄²⁻/Pd	+0.62 V	Part XXV
AuCl₄⁻/Au	+1.00 V	Herschel’s chrysotype
Hg²⁺/Hg	+0.85 V	Herschel’s kelainotype, and Part XXVI

Read the first and last rows together and you have the puzzle. A reaction between two couples runs in the direction that makes the cell potential positive, and the driving force is the difference between them. For the free ions:

E = 0.80 − 0.771 = +0.03 V

Free aqua ions

Thermodynamically allowed, and by a margin so slight that it is not something to build a photographic process on. Now put the iron in citrate, which is what a Van Dyke sensitiser does:

E = 0.80 − 0.372 = +0.43 V

Citrate-complexed iron

and in oxalate, which is what a kallitype does:

E = 0.80 − 0.02 = +0.78 V

Oxalate-complexed iron

Where the couples sit, and what the ligand does to the iron

1. 1Iron couples, left of the scale — the ligand decides where the iron sits, and the ligand is the only thing you choose
2. 2Noble-metal couples, right of the scale — the metal decides what has to be reached
3. 3The free-ion gap, 0.03 V — thermodynamically allowed and photographically useless
4. 4The oxalate gap, 0.78 V — the same reaction with the iron chelated, and it is not close

Values from Ware's Platinomicon, sections 10.5 and 11.3. Drawn to a linear scale in volts; positions are the published figures and not a measurement of this course's own.

Why does a ligand move a potential at all? Ware gives the reason in one line and it is worth having, because it generalises. A redox potential compares the stability of the two oxidation states. Oxalate binds iron(III) with an overall formation constant of about 10²⁰ and iron(II) with about 10⁵ — fifteen orders of magnitude apart. So chelation stabilises the higher oxidation state enormously more than the lower one, which makes iron(III) harder to reach and iron(II) easier to leave. The potential falls, and the iron(II) becomes a reducing agent.

🧮The stepwise constants, and one thing they predict

Ware gives the three stepwise formation constants for the oxalate complex: K₁ about 10⁹, K₂ about 10⁷, K₃ about 10⁴, so the overall Kf is their product, about 10²⁰. The last one is the small one, and that has a consequence he draws out:

[C₂O₄²⁻] ≈ √(1/K₃) = √(10⁻⁴) = 10⁻² mol/L

Free oxalate in a 1 molar tris-complex solution

About 0.01 molar free oxalate is floating around in a concentrated ferrioxalate solution, because the third ligand is only loosely held. That free oxalate is why a kallitype sensitiser throws a precipitate. Photographers’ Formulary tell you to expect it and to disregard it: silver(I) and oxalate make silver oxalate, which is sparingly soluble, and it comes out of the bottle the moment the two solutions meet.

The same free oxalate is a hazard note as well as a chemical one, and it belongs on the kallitype lab rather than here.

What the citrate figure is and is not. Ware’s +0.372 V is given for “the citrato-complex of iron(II)” with a reference, and he uses it to make exactly the point this section makes: the citrate complex “shows it to be a weaker reducing agent than the oxalato-complex … and it does not reduce platinum(II) or palladium(II), although it will reduce gold(III) and silver(I), under the printing conditions”. Read that sentence against the table and it becomes a prediction you can check: a Van Dyke sensitiser cannot be made to print in platinum or palladium, because +0.372 V does not reach +0.73 or +0.62 V; a kallitype’s oxalate can, and that is exactly what Part XXV’s chemistry is.

Ligand by ligand

Four ligands are named in the manifest for this page. Three of them are real iron ligands, and the fourth is a correction worth making carefully.

Citrate — the cheap one, and an ill-defined one

Ammonium iron(III) citrate is what the Van Dyke sensitiser, the argyrotype and the classic cyanotype all use.

It is not a compound. One supplier’s safety data sheet prints, in the field where a molecular formula should be, the sentence that it “is a complex salt of undetermined structure, composed of Iron, Ammonia, and Citric Acid”. Ware’s Cyanomicon tabulates the two commercial grades by iron content: the green form runs 14 to 18 per cent iron by weight, the brown 19 to 28 per cent. His argyrotype notes specify “the green variety: 16 % iron”. Two printers with different jars are not running the same experiment, which is why every practical page in this part asks for the supplier and the lot number in the batch record.

What it does to the process. A weaker reduction than oxalate gives, and therefore — on the potentials above — access to silver and gold and not to the platinum metals. Ware’s Cyanomicon puts the quantum yield of the citrate system at a maximum of 0.45 at 365 nm, falling to 0.28 at 436 nm: less than half the oxalate’s 1.2, so citrate is the slower ligand as well as the weaker one. Against that, it is cheap, it is stable as a solid, and the mixed Van Dyke sensitiser keeps for months where a mixed cyanotype keeps for hours.

And one thing about it is genuinely unknown. Ware states that the initial organic photoproduct of the citrate system is acetone dicarboxylic acid, and that the nature of the iron(II) photoproduct remains unknown. The oxalate case is worked out to the level of a named radical mechanism; the citrate case, which is the one your Van Dyke uses, is not.

Oxalate — the strong one, and the difficult one

Ferric oxalate is what makes a kallitype a kallitype, and it buys three things and costs three.

What it buys. The potential, at +0.02 V, which is a 0.78 V driving force against silver and enough to reach platinum and palladium as well. The quantum yield, about 1.2 iron(II) per absorbed photon between 250 and 420 nm against an ideal maximum of 2 — nearly three times the citrate figure. And the image: Sandy King reports that ferric oxalate “permits darker shadows, i.e. more Dmax, with kallitype than with either Vandyke or argyrotype”, while being careful to add that “the difference is not huge”.

What it costs. Ware’s assessment of the material is unusually blunt for a technical monograph: ferric oxalate “is a chemists’ nightmare: ill-characterised, evidently polymorphic, apparently uncrystallisable”, with a molecular structure that was unknown until one polymeric form was solved in 2015. Its composition and properties depend on the method of preparation, which is to say on the supplier. Sources differ over its state of hydration, so its formula weight is uncertain: Ware lists 375.76 for the anhydrous salt, 447.81 for a tetrahydrate, 465.83 in the CRC Handbook for a pentahydrate and 483.84 for the hexahydrate two catalogues sell. A weight of ferric oxalate is therefore not a known quantity of iron, and the kallitype formulary page says so in its own ingredient note.

And there are two different substances sold under the name. Photographers’ Formulary open their kit sheet with the warning that “the photographic term ‘ferric oxalate’ is a misnomer, which has given rise to a considerable amount of confusion in the photographic literature”. Tri-potassium ferric oxalate is thermally stable, keeps in the dark, and has photo-activity too low to be useful; the tri-hydrogen form, the acidic one, is what a photographic supplier means. They recommend against buying the green solid tripotassium salt at all. Ware’s parallel observation is that very few fine chemical houses list ferric oxalate, and that when they do “the price of ferric oxalate is usually about 100 times that of ferrous oxalate” — which is a good working reminder that the two are not interchangeable and that a cheap jar is a warning.

Its solutions do not keep. Ware records six to nine months in the dark as the figure some give, and notes that at least one well-respected platinum printer makes up fresh ferric oxalate solution the night before every session. King’s own limit for the kallitype is stricter and states the consequence: mix no more than you will use in two to three months, because the solution degrades “with a resulting increase in print fog”.

✕A database entry that is wrong, and how to tell

Look up ferric oxalate on PubChem and the record that comes back is titled Sensodyne, a proprietary toothpaste. The registry has resolved the query to a formulation rather than to the substance, and the computed properties on that page belong to neither.

What is still usable on it, and why. The GHS block underneath is aggregated from the European Chemicals Agency’s inventory under the entry Diiron trioxalate, EC 220-951-7, which is ferric oxalate, and it reports H302 and H312 at 100 per cent of a single notification. So the hazard block is evidence and the identity block is not — which is the general lesson. An aggregator’s title is a search result; the registry number underneath it is the identity. Check the EC or CAS number before you believe a page, and say which part of it you are relying on. This course’s ferric oxalate page records the defect rather than quietly working round it.

Tartrate — real, historical, and under-documented

Herschel names it in the same breath as the citrate. Article 210 records that the effect “is not, it should be observed, peculiar to the ammonio-citrate of iron. The ammonio- and potasso-tartrate fully possess, and the perchloride exactly neutralized partakes of, the same property”; Article 218 offers the reader either the ammonio-citrate or the ammonio-tartrate for the silver process. Ware’s list of photosensitive iron(III) carboxylates is citrate, malonate, tartrate and glycollate, and he adds that “no clear criterion has yet emerged for deciding what structural feature of such complexes is necessary for photosensitivity”.

What this course could not find. No redox potential for an iron(III)/iron(II) tartrate couple, and no quantum yield for the tartrate system, appears in any source read for this page. The direction is established — a chelating polycarboxylate lowers the iron potential, and tartrate is one — and the magnitude is not. Where tartrate appears in this part it is in a quite different role: it is the potassium sodium tartrate of a kallitype developer, where it is a ligand for the iron(II) that is already there rather than a photosensitiser, and it is the tartaric acid of the Van Dyke sensitiser, whose function the Van Dyke lab takes up because the answer is less obvious than it looks.

Sulfamate is not an iron ligand, and the distinction matters

The manifest for this page lists sulfamate among the ligands, and the sources do not support that, so the page says what they do support instead.

In Ware’s argyrotype the iron is still ammonium iron(III) citrate — the green form, 22 g of it per 100 cc. What the sulfamic acid does is make silver sulphamate, NH₂SO₃Ag, in situ from silver(I) oxide, and the sulfamate ends up as the silver’s counter-ion, not on the iron. Ware’s reasoning is a completely different argument from the ligand argument, and it is set out under the argyrotype below.

The confusion is easy to fall into and it is worth naming, because it is the same shape as the ferric-oxalate misnomer: two substances in one bottle, and the name of one attached to the role of the other. In this family the anion on the iron sets the photochemistry and the anion on the silver sets the wet processing. They are independent choices, and the argyrotype changes the second while leaving the first alone.

Print-out and develop-out, in the same family

Both processes in this part make the same two steps happen. What differs is where the second step happens and when, and that difference is what separates a Van Dyke from a kallitype.

The same chemistry, arranged two ways

1. 1Van Dyke: silver is in the coating, so step two happens where step one doesThe sheet carries iron(III) citrate and silver nitrate together. As the exposure proceeds, each iron(II) made by light finds a silver ion within a molecular distance and reduces it. The image builds while you watch, and the exposure is judged by looking
2. 2And it keeps going after the light stopsThe suppliers agree on the practical consequence: stop when the print is about half as dark as you want, because it darkens through the wash, the fixer, any toning and the drying
3. 3Kallitype: silver is in the coating too, but the photoproduct is stuckIron(II) oxalate is only sparingly soluble — Ware gives 0.022 g per 100 cc — so in a dry sheet the ions are immobile and cannot meet. The exposure makes a photoproduct and very little picture
4. 4The developer mobilises the iron(II), and then the reaction runs in secondsWall's 1924 statement of it is still the clearest: "the ferrous salts dissolve in the developer, and reduce silver nitrate to the metallic state at the points where the ferric salt has been reduced by light"

Two arrangements of one reaction. Nothing in either route reduces anything the light had not already reduced.

What a Van Dyke actually does, and when

The manifest asked whether the reduction proceeds during exposure, after exposure, or in the wash, because sources differ. Read together, they do not so much disagree as describe different parts of one curve, and the honest answer has three pieces.

Most of it happens under the light. Bostick & Sullivan state it plainly: the image “darkens and develops as it is exposed, producing a ‘printing out’ image, negating the need for a separate development stage”. Wynn White’s practitioner figure is that after exposure a faint image is visible “with the print showing approximately half of its final density”. Bostick & Sullivan give the same number as an instruction: stop when the print “appears approximately half as dark as the desired final image”.

Some of it happens in the wet baths, and it is a small amount. Ware, describing the argyrotype, gives a figure rather than an impression: “A little development (half to one stop) can subsequently be expected to occur in the high values during wet processing.” Half to one stop in the high values is real and is not most of the picture.

And a large part of what you see in the fixer is not more silver at all. This is the piece that sources describe and do not explain, and Ware does explain it. The dramatic darkening in the thiosulfate bath — Photographers’ Formulary say the print “will darken and become brown during this fixing period”, Bostick & Sullivan say it comes out about three-quarters as dark as the final image — Ware attributes to partial sulphide toning: the silver nanoparticles acquire a coating of silver sulphide, perhaps only a few atoms thick, which changes their surface plasmon resonance and therefore their colour “profoundly”. Energy-dispersive X-ray analysis of argyrotypes by Ellie Young at the Royal Melbourne Institute of Technology found sulphur as well as silver in the image. So the print does not get much more silver in the fixer; it gets a different-coloured silver. The same mechanism, taken too far, is how a fixer destroys the print, and the labs give it as the reason for a short, weak bath.

What no source read for this course establishes is the split between those three contributions as a measurement — how much of the final density is exposure, how much is wet development and how much is the colour shift. That is a thing your own densitometer can settle for your paper, and the Van Dyke lab asks you to measure it.

What the kallitype developer is doing, which is not what a developer does in Part VIII

This is the most important mechanism claim in the part, and it is a claim the sources make rather than one the course infers.

There is no developing agent in it. Open a bottle of sodium citrate kallitype developer and there is sodium citrate and water. No hydroquinone, no metol, no phenidone, no sulfite, no alkali beyond what the salt itself provides. Nothing in the bath can reduce silver. Compare Part VIII, where the whole subject is which molecule donates the electrons and how fast.

What it does instead is supply a ligand and a solvent. Ware’s account of the platinotype case is the mechanism, and King’s statement that “the developers and clearing agents used for platinum can be used for kallitype” is the licence to transfer it. The iron(II) oxalate that light produced is sparingly soluble — 0.022 g per 100 cc — so in a dry coating the ions are immobile and “cannot encounter one-another”. Add oxalate, citrate or tartrate ions in solution and the iron(II) dissolves as a complex:

FeC₂O₄ + C₂O₄²⁻ → [Fe(C₂O₄)₂]²⁻

The developer's first job: get the photoproduct into solution so it can move

and once mobile it does the thing it was always able to do:

Fe²⁺ + Ag⁺ → Fe³⁺ + Ag

The developer's second job is not a job at all — this reaction needs no help once the ions can meet

Wall’s 1924 dictionary states the whole of it in one sentence, seventy years before the redox potentials were tabulated for this purpose: “The ferrous salts dissolve in the developer, and reduce silver nitrate to the metallic state at the points where the ferric salt has been reduced by light.”

And it has a third job, which is why development runs for minutes rather than seconds. King records that development is “visually complete in about 15-30 seconds, but a development time of 5-10 minutes is important for archival purposes: much of the residual ferric iron, which if left in the print could cause loss of permanence, is removed at this stage”. The bath is the first clearing bath as well as the developer, and the two functions compete for its capacity: King’s replenishment rule exists because “the accumulation of ferrous iron will make it increasingly difficult to clear the print”.

!Three words that do not mean what Part VIII meant by them

Developer. In Part VIII a developer contains a reducing agent that donates electrons to silver. In this part it contains a ligand that mobilises a reducing agent the light already made. Nothing in a kallitype developer is oxidised by silver.

Development. In Part VIII development amplifies a latent image by a factor of 10⁷ to 10⁸. Here it completes a reaction whose extent light has already fixed, atom for atom. A longer development does not build more image after the first half-minute; it clears more iron.

Physical development. Part IV’s term for silver deposited from solution onto existing nuclei rather than grown in place from a crystal. That is a much better description of what happens here, and the next section is about what it does to the print.

The absence of amplification, quantified

Everything odd about these processes descends from one ratio, so it is worth deriving rather than asserting.

🧮Photons in, silver atoms out, three ways

A developed silver halide emulsion. Part IV’s arithmetic is the course’s own, from published coating weights and grain sizes, with its assumptions written out — and it exists because no source the course holds states an amplification factor. One gram of silver per square metre is 5.6 × 10²¹ atoms; a square metre of 0.2 μm crystals in a single layer is 2.5 × 10¹³ crystals, and at four silver atoms per latent-image cluster that is 1.0 × 10¹⁴ atoms of latent image:

5.6 × 10²¹ ÷ 1.0 × 10¹⁴ ≈ 6 × 10⁷

Amplification by development

An iron-silver print. No amplification step exists, so the count is the photochemistry times the stoichiometry. Ware’s measured quantum yield for ferrioxalate photoreduction is about 1.2 iron(II) per absorbed photon between 250 and 420 nm; the ideal maximum for its two-electron mechanism is 2, and the shortfall is the competing dark back-reaction. At Fe:Ag of 1:1:

1.2 × 1 = about 1.2

Silver atoms per absorbed photon, oxalate

0.45 × 1 = about 0.45

Silver atoms per absorbed photon, citrate at 365 nm

The ratio. About 6 × 10⁷ against about 1. Seven to eight orders of magnitude, and it is the whole of the difference between an enlarging paper and this one.

What the figures are and are not. The 10⁷ is the course’s own calculation with a stated shaky step — the monolayer assumption — and it is a ratio of atom counts. The 1.2 and the 0.45 are published measurements of quantum yield, which is a ratio of molecules changed to photons absorbed. They are not the same quantity and dividing one by the other is a comparison of magnitudes rather than a derivation. Ware makes the same order-of-magnitude comparison from the other direction: working from the Stark-Einstein law alone, the maximum possible sensitivity of any proto-photographic material is about 34 J/m² for a just-perceptible image, which on a photographic scale is a speed of about 10⁻⁵ ISO — one ten-millionth of a 100 ISO film. Two independent routes to the same exponent is a reason to trust the exponent and not the digits.

Four consequences follow immediately, and each is a practical fact you will meet in the next four pages.

Exposures are in minutes. Photographers’ Formulary give the Van Dyke’s optimum as “in the 10-30 minute range” under a 500 or 1000 W photoflood, and the kallitype’s as 10 to 20 minutes under a 275 or 300 W sunlamp, falling to “4 to 6 minutes” in a UV light box. Bostick & Sullivan say kallitype print times are “about 1-2 stops faster than traditional palladium”. Every one of those figures belongs to somebody else’s lamp, which is why both labs publish a calibration and not a time.

The light must be ultraviolet. The photoactive species absorbs where the exposure has to be made and nowhere else useful, and there is no amplification to make up a shortfall.

There is no enlarger. Projection spreads the light over the square of the magnification. With 10⁷ of amplification you can afford that; with 1, you cannot. Every print in this part is a contact print the size of its negative.

A lot of silver has to be coated. In an emulsion, a few atoms per crystal become the picture. Here, every atom of image silver has to be put on the paper as a salt and reduced individually — which is why the overview’s cost arithmetic lands where it does, and why Wynn White double-coats.

⚗One thing iron-silver gets back, and it is not nothing

Ware quantifies a penalty that cyanotype pays and this family does not. Many alternative-process coatings contain coloured substances that absorb the printing light and “give nothing in return” — the pigment in gum bichromate, the potassium ferricyanide in cyanotype, the noble-metal salts in platino-palladiotype. He calls it the internal filter effect: a coloured filter built into the coating and inescapable. In a classic cyanotype it is severe, because hexacyanoferrate absorbs strongly in the same near-ultraviolet band the iron needs.

His statement about this family is flat: “The internal filter effect is substantially absent in iron-silver sensitizers, and gold(I).” Silver nitrate is colourless and does not compete for the ultraviolet. So an iron-silver coating uses its light more efficiently than a cyanotype does, at the same hopeless absolute speed. That is worth knowing when you are choosing a lamp for the cluster: a source whose output is nearly wasted on a cyanotype is not necessarily wasted here.

Why it must be a contact print, and what that asks of the negative

Two independent reasons converge, and it is worth keeping them apart because only one of them is about speed.

The speed argument. At 10⁻⁵ ISO, an enlarger exposure is not slow, it is impossible. Nothing about reciprocity or lamp power changes an eight-order-of-magnitude shortfall.

The spectral argument. Even a light source powerful enough would be the wrong colour. Enlarger optics, condensers and the negative’s own base absorb in the near ultraviolet, and Part XVI’s contact frame is glazed in glass rather than acrylic for exactly that reason.

So the print is the size of the negative, and the negative has to be made for the process. This is Part XXI’s subject and is not re-taught, but the figures for this part belong here because they differ from the cyanotype’s. Photographers’ Formulary say Van Dyke Brown “is capable of an extremely long tonal range. Negatives with a density range up to 1.85 can be used”, and print the same sentence with the same figure for their kallitype kit. Sandy King’s recommendation for kallitype is a density range “of about log 1.8”, which he calls “a very contrasty negative that will not print well even on a grade #0 or #1 paper” and reaches by developing film about 50 per cent longer than normal. Ware’s argyrotype notes ask for “at least 2 and as much as 2.4” in the ultraviolet, which is a different measurement from a visual density range and the reason Part XXI’s ultraviolet-density section exists.

Those three figures agree well enough to plan by. A negative made for platinum will print in this part, a negative made for grade 2 enlarging paper will not, and a negative made for the classic cyanotype is a different negative again. The standard negative Part XXI asked you to choose is the right one here.

Where the silver ends up, and what that does to the print

An enlarging print has its image in a coated layer that can be lifted off the base. These prints do not. The sensitiser is a solution that soaks into cellulose, and the metal is deposited among and inside the fibres.

An iron-silver print, in section

Drawn to show the arrangement, not the depth. The particle size is Ware's figure for a brown silver image; the paper weights are his argyrotype recommendations. Whether the silver sits predominantly in the fibres or in the sizing is discussed below and is not settled by any source read for this course. Layer depths are drawn to be readable, not to scale: on real film the base is roughly a hundred times the emulsion, and drawn honestly the emulsion would vanish. Any thickness given in the labels is the real one.

Three consequences, all of them visible.

The surface is the paper’s surface. There is no gloss to speak of and no baryta beneath to raise the maximum density by reflecting light back through the image. That is part of why these prints look the way they do and part of why their Dmax is lower than an enlarging paper’s.

The silver is unprotected. Ware’s structural point is the one that governs the permanence discussion: brown silver images consist of particles of colloidal dimensions, about 20 nm — “far smaller than the wavelengths of visible light” — and “such small particles are inevitably more vulnerable to chemical attack: they present a relatively large surface area and are rapidly dissolved by reagents that ‘etch’ or ‘bleach’ (i.e. oxidise) silver”. There is no gelatin between that metal and the air, and no supercoat between it and a finger.

And the sizing is part of the chemistry, not just the handling. Photographers’ Formulary state the observation directly: “Prints made with arrowroot starch-sized-paper will have a brown color while those sized with gelatin will have a blue tone.” Ware’s platinum work gives the mechanism for the same family of effects — gelatin, like a surfactant, “protects” nanoparticle metals and favours smaller particles and therefore warmer colours. So the sizing changes where the silver goes and how big the particles are, and both change the colour.

⁇Does the silver sit in the fibres or in the sizing? The course cannot say

The manifest asked for the image structure “as conservators describe it, including whether the silver sits in the fibres or the sizing”, and named the American Institute for Conservation’s Photographic Materials Group entry as the source.

That entry is empty. Read on 7 September 2026, the page titled VanDyke Brown, Kallitype, Brown Print, Sepia Print, Ferro-Gallic, Argentotype, Agyrotype carries the standard headings for a completed process entry — Historical Facts, Identification Characteristics with Image material, Colour and Support beneath it, Conservation, Housing and Storage, Exhibition, Emergency Recovery — and no text under any of them. Its footer records that it was last edited on 31 August 2020. The equivalent entries for platinum and palladium, for cyanotype and for the salted paper processes are written.

What the course has instead is Ware’s particle size, the sizing observations above, and the general statement that the image substance is precipitated in the paper. That is enough to say the image is in the sheet and not on it, and it is not enough to say what fraction is held by cellulose and what fraction by sizing.

What would settle it is a cross-section or an elemental map of a Van Dyke or kallitype print in the conservation literature. If you find one, the claim in this section can be sharpened; until then this page states the limit rather than filling it.

Image colour, and the four things that set it

The same negative can print warm brown, cold brown or nearly neutral, and none of that is a dye. It is particle size, and Ware lists the properties that govern the colour of a nanoparticle image: the chemical composition of the image substance, the particle size distribution and shape, adsorbates on the surface, and the refractive index of the host matrix.

For a silver siderotype, four things you control move those properties.

The ligand, through the speed of the reaction. A vigorous reduction makes fewer, larger nuclei grow; a sluggish one makes many small particles. That is the general shape of the argument and it is why the two processes of this part do not look alike from the same negative.

The water in the paper at the moment of exposure, which is the one Ware puts most weight on for a print-out process: “A large ‘reservoir’ of fibre water will permit greater amounts of substances to react locally during exposure and therefore make it possible for the particles of metal to grow larger. Such particles will appear more neutral in hue. A very restricted ‘pool’ of fibre water will constrain the chemistry to producing small metal particles only, which can show quite marked colours — brown or sepia for silver.” His argyrotype practice puts that to work: humidifying the coated sheet over water at 100 per cent relative humidity for thirty minutes before exposure shifts the print-out colour to “an attractive purplish-grey”, and he includes glycerol in the sensitiser as a humectant for the same reason, because without it “the image may be a more yellowish-brown”.

The developer, in a kallitype only, which is the whole craft argument of the kallitype lab and is left there.

The toner, which changes the composition of the image substance rather than the size of its particles, and which the clearing and toning lab owns. The mechanism for a printed-out silver image is Part XXII’s.

A fifth thing changes the colour and is not a control: the sulphide the fixer puts on the surface of the particles, discussed above. It is one of the reasons a Van Dyke that has been over-fixed looks different as well as thinner.

What is left in the paper that should not stay

Take a printed and washed sheet and list what is in it besides the picture. This is the inventory the next three pages work from.

Species	Where it came from	Why it must go	Which bath takes it
Iron(III), free	Unexposed sensitiser, and every iron(II) that reduced a silver ion	Oxidises the image silver; hydrolyses to a yellow-brown stain	Water first, then a chelating clearing bath
Iron(III), bound to cellulose	Chemisorbed to the hydroxyl groups of the glucose units	The same, and water will not touch it	A chelating bath, and a reducing bath before it
Iron(II)	Exposed sensitiser that found no silver	Air oxidises it back to iron(III), and then see above	The developer, in a kallitype; the wash, in a Van Dyke
Unreduced silver salt	Everywhere the negative was dense	Still light-sensitive, so the print goes on printing	The fixer, and only the fixer
Silver-thiosulfate complexes	Made by the fixer, out of the silver salt	Decompose over years to silver sulphide, staining and fading	The final wash, and a hypo-clearing bath
Thiosulfate itself	The fixer	Attacks the image silver directly over time	The same

Three of those six have tests, and the tests are the subject of the clearing and toning lab: a hexacyanoferrate spot test for residual iron, and Part XII’s residual-silver and residual-thiosulfate tests, which this part has to check for validity on a paper with no gelatin in it.

And the first two are the ones that give the family its reputation. Ware’s argyrotype account puts the reason the clearing is hard in one sentence, and it is a trap rather than a difficulty: “The difficulty with these processes lies in clearing the print of iron salts, without dissolving the image silver in the presence of the oxidizing nitrate ion, for which alkaline developers were necessarily recommended, but are not very effective in removing the excess iron(III) salts, which is better done in acid.”

Read that twice, because it is a genuine bind. Iron(III) comes out best in acid. The image silver survives best in alkali, because nitrate is an oxidising anion and “tends to dissolve the colloidal image silver during wet processing, especially under acidic conditions”. The traditional answer — the borax and Rochelle salt developers of the kallitype — protects the silver and, in Ware’s words, “cause hydrolysis of the excess iron(III) in the sensitizer and the deposition of insoluble ferric hydroxide in the image, which ultimately causes it to fade”. Every classical iron-silver process is a compromise between two failures, and that is the problem the next section’s reformulation was designed to remove.

The argyrotype, and what changing the silver salt buys

Ware re-visited the chemistry in 1990 and published the result in the British Journal of Photography in June 1991. The reasoning is a good model of how to attack a process rather than a recipe.

The diagnosis. Not the iron, which works. The nitrate, which is an oxidising anion and is present in every historical iron-silver sensitiser because silver nitrate is the cheap soluble silver salt.

The requirement. A soluble silver salt with a non-oxidising anion, so that the sensitiser can be worked acid — where the iron clears well — without the acid costing you the image.

The candidate. Ware notes that few such salts exist and that most, silver fluoride among them, have properties or toxicity that “debar them from ‘home chemistry’”. Silver sulphamate, NH₂SO₃Ag, fits: it cannot be bought but is easily made in the beaker from sulfamic acid and silver(I) oxide.

Ag₂O + 2 NH₂SO₃H → 2 NH₂SO₃Ag + H₂O

Silver sulphamate, made in situ: the acid dissolves the oxide and the anion that results does not oxidise silver

The working pH, and it is chosen rather than accepted. Ware’s sensitiser carries about a 20 per cent excess of sulfamic acid over the 5.87 g stoichiometrically needed for 7 g of silver oxide, which brings the solution to about pH 3.5. He gives three reasons in one sentence: the excess acid suppresses hydrolysis of the iron(III), keeps the silver in solution — otherwise silver citrate precipitates — and “this pH is also the optimum for the photosensitivity”. Ware’s Cyanomicon adds the photochemical reason behind the third: in the citrate system the monomeric complex below pH 1.5 is photo-inactive and the dimer above pH 2 is the photoactive species.

What it buys downstream. A single-bottle sensitiser that keeps for a year or more; wet processing in de-chlorinated water with a little citric acid rather than a compromise developer; and contrast adjustable by acid alone, since a further 1 g of sulfamic acid per 100 cc makes a more contrasty sensitiser because the acid “tends to dissolve silver in the highlights”.

⁇1990 or 1991?

Ware’s own two accounts give different years for the same invention. His web article says “I re-visited the underlying chemistry in 1990, and came up with an easier … process, which I called Argyrotype”, and its footer records first publication in the British Journal of Photography on 13 June 1991. His workshop handout says “The Argyrotype version was devised in 1991”.

Nobody else is involved in the disagreement, and it is small: an invention made in one year and published in the next is the ordinary shape of such a thing. The course writes it as “devised about 1990 and published in 1991” and gives both of Ware’s statements rather than choosing. The formulary entry carries the same note.

Why this part does not teach it as a lab. The argyrotype needs silver(I) oxide or silver carbonate, which are not stocked by photographic suppliers, or a precipitation from silver nitrate and sodium hydroxide that adds a step and a hazard; and its whole advantage is a processing convenience the two labs here are meant to make you feel the absence of first. It is written in full in the formulary, it is the process to move to when the clearing of a Van Dyke has annoyed you enough, and Bostick & Sullivan’s Van Dyke sheet notes that “Argyrotype follows the same instructions”, so the bench technique transfers.

The comparison this part exists for

Three ways of putting a silver image on a sheet of paper, in one table, with every entry from a source.

	Silver halide emulsion	Printed-out silver chloride	Iron-silver
Taught in	Parts IV, V and XVIII	Part XXII	this part
Light-sensitive substance	Silver halide crystals in gelatin	Silver chloride with excess silver nitrate, in the paper	An iron(III) carboxylate; the silver salt is not light-sensitive
What light makes	A latent image of a few silver atoms	Metallic silver directly, plus a halogen the acceptor must take up	Iron(II), plus carbon dioxide from the ligand
Amplification	10⁷ to 10⁸, by development (course’s own arithmetic, Part IV)	None	None. About 1.2 silver atoms per absorbed photon at best
Speed	ISO 25 to 3200 for film	Of order 10⁻⁵ ISO — Ware’s limiting estimate for any proto-photographic material	The same order; Ware’s estimate covers both
Spectral response	Blue-sensitive natively; extended by dyes to the red and infrared	Blue and near-ultraviolet	Near-ultraviolet only. Ware puts the ferrioxalate charge-transfer band at λmax 260 nm and its quantum yield near 1.2 out to 420 nm
Exposure source	Enlarger, camera, contact	Sun or a UV source, by contact	Sun or a UV source, by contact
Image structure	Filamentary silver in a gelatin binder, over baryta on a paper base	Colloidal silver in the paper, or in the sizing	Colloidal silver, about 20 nm, in the paper and its sizing; no binder
Image colour	Neutral to warm, set by grain size, developer and toner	Warm, self-masking, shifted by gold or platinum toning	Warm, set by ligand, humidity, developer and toner
What must be removed	Unexposed halide, then thiosulfate	Unreacted silver salt, then thiosulfate	Iron(III) and unreacted silver salt, then thiosulfate
The characteristic failure	Residual thiosulfate and silver sulphide staining	Fading of unprotected colloidal silver; sulphiding and mirroring	Residual iron oxidising the image, with yellow-brown highlight stain

📷What the table does not have, and why the standard negative matters

Every row above is sourced, and not one of them is a measurement made on your paper. The process comparison matrix says the same thing about itself in terms: what is published there is assembled from different photographers, papers and decades, so it describes what the processes are like rather than what they do to one picture.

The standard negative is the fix. Print it in every process you can reach, measure the exposure scale, the maximum density and the image colour each time, and the last column of that table becomes yours. This part adds two rows, and Part XXV’s comparison atlas closes it.

Alternative route

This page needs no facility. It is reading, arithmetic and a table, and a reader with no wet space and no ultraviolet source can work through all of it — which is why the front matter’s requiresUV flag is about the practice the lesson prepares for rather than about the lesson.

What that reader should do instead of the labs. Read the four formulary entries this page keeps sending you to — Van Dyke Brown, kallitype, Herschel’s argentotype and Ware’s argyrotype — side by side, and put the four sensitisers in a table of your own with the iron salt, its ligand, the silver salt, the working pH and where the second step happens. The four differences that table exposes are the argument of this part, and it can be made with paper and a pencil.

And the one experiment that has no substitute. Nothing here tells you the colour your paper gives. Ware’s own remark about humidity and image colour — that leaving a coated sheet over water for thirty minutes shifts a print-out image to a purplish-grey — is the kind of claim that costs one sheet to test and cannot be settled by reading. If you can print at all, test it.

Recap

The reaction is barely favourable for the free ions and comfortably favourable for the complexes. Fe³⁺/Fe²⁺ is +0.771 V and Ag⁺/Ag is +0.80 V, a gap of 0.03 V; citrate takes the iron to +0.372 V and oxalate to +0.02 V, and the gap becomes 0.43 and 0.78 V. The ligand is the design decision.

The ligand also decides what else you can reach. Citrate reduces silver, gold and mercury and does not reduce platinum or palladium; oxalate reduces all of them. That single fact is why a kallitype can be toned in palladium and why Part XXV is an oxalate process.

One iron(II) makes one silver atom, and there is no second stage. At a quantum yield of about 1.2 for the oxalate and 0.45 for the citrate, the process delivers of order one silver atom per absorbed photon against 10⁷ to 10⁸ for a developed emulsion. Everything else — the minutes, the ultraviolet, the contact frame, the cost — follows from that ratio.

Print-out and develop-out are the same chemistry differently arranged. In a Van Dyke the silver is beside the iron and the reaction runs under the light; in a kallitype the iron(II) oxalate is insoluble and immobile, and the developer’s job is to dissolve it, not to reduce anything.

The image is colloidal silver in the paper, about 20 nm, with no binder over it — which sets the surface, the maximum density, the colour and the vulnerability all at once.

And the iron is still there when you have finished. It ends the reaction as iron(III), some of it chemisorbed to the cellulose, and iron(III) oxidises silver. That is the clearing problem, and it is the subject of two of the four pages after this one.

Check your understanding

Question 1. Standard potentials: Fe³⁺/Fe²⁺ is +0.771 V and Ag⁺/Ag is +0.80 V. Why does putting the iron in a citrate or oxalate complex make the process work?

• It makes the iron(III) more soluble, so more of it can be coated
• It lowers the iron couple - to +0.372 V for citrate and +0.02 V for oxalate - so the iron(II) becomes a much stronger reducing agent and the driving force against silver rises from 0.03 V to 0.43 or 0.78 V
• It raises the silver couple, so silver becomes easier to reduce
• It protects the iron(II) from atmospheric oxidation, so more of it survives to reach the silver

Show the answer and why

Answer: It lowers the iron couple - to +0.372 V for citrate and +0.02 V for oxalate - so the iron(II) becomes a much stronger reducing agent and the driving force against silver rises from 0.03 V to 0.43 or 0.78 V

Ware gives the mechanism as well as the numbers: oxalate binds iron(III) with an overall formation constant of about 10²⁰ and iron(II) with about 10⁵, so chelation stabilises the higher oxidation state by fifteen orders of magnitude more than the lower one. That makes iron(III) harder to reach and iron(II) easier to leave, and the potential falls accordingly. Protection from aerial oxidation is a real and separate issue - Ware notes that the exposure alone leaves nothing permanent because oxygen re-oxidises the iron(II) - but it is not what the ligand is doing to the potential.

Question 2. You have a Van Dyke sensitiser and you want to print in palladium. What do the redox potentials say?

• It will work but more slowly, because palladium is harder to reduce than silver
• It will not work: the citrato-iron(II) couple at +0.372 V does not reach PdCl₄²⁻/Pd at +0.62 V, and Ware states in terms that the citrate complex does not reduce platinum(II) or palladium(II)
• It will work only if the sensitiser is made more acid
• It will work, because palladium at +0.62 V is easier to reduce than silver at +0.80 V

Show the answer and why

Answer: It will not work: the citrato-iron(II) couple at +0.372 V does not reach PdCl₄²⁻/Pd at +0.62 V, and Ware states in terms that the citrate complex does not reduce platinum(II) or palladium(II)

The last option inverts the argument, which is the easy mistake: a lower metal potential means the metal is harder to reduce, not easier, because the reducing couple has to sit below it. Ware states the consequence directly, that the citrato couple "does not reduce platinum(II) or palladium(II), although it will reduce gold(III) and silver(I), under the printing conditions". Change the ligand to oxalate and the iron couple falls to +0.02 V, which reaches all of them - which is why the kallitype and the platinotype share a sensitiser chemistry and the Van Dyke does not.

Question 3. What is a kallitype developer doing chemically?

• Donating electrons to silver, as a developing agent does in Part VIII
• Dissolving the insoluble iron(II) oxalate the light made, so the iron(II) can move and reduce the silver - and, over five to ten minutes, carrying much of the residual iron out of the sheet
• Raising the pH so that the silver nitrate hydrolyses to silver oxide, which is then reduced
• Removing the unexposed silver salt, which is why the print must not be over-developed

Show the answer and why

Answer: Dissolving the insoluble iron(II) oxalate the light made, so the iron(II) can move and reduce the silver - and, over five to ten minutes, carrying much of the residual iron out of the sheet

There is no reducing agent in the bath: sodium citrate and water cannot reduce silver. Iron(II) oxalate is only sparingly soluble - Ware gives 0.022 g per 100 cc - so in a dry coating the ions are immobile and cannot meet. Adding citrate, oxalate or tartrate ions dissolves the photoproduct as a complex, and the redox reaction then runs on its own. Wall put the whole mechanism in one sentence in 1924. King adds the second function: development is visually complete in 15 to 30 seconds but is continued for 5 to 10 minutes because much of the residual iron leaves at that stage. Removing unexposed silver is the fixer's job and a quite separate bath.

Question 4. A Van Dyke sensitiser is about 3.8 g of silver nitrate in 100 mL, and a supplier gives about 4 mL to coat an 8 × 10 inch sheet. How much silver metal is that per square decimetre, and how does it compare with an enlarging paper at about 1.0 g of silver per square metre?

• About 0.19 g/dm², roughly two hundred times the coating weight of an enlarging paper
• About 0.019 g/dm² of silver metal, roughly twenty times the coating weight of an enlarging paper
• About 0.0019 g/dm², roughly twice the coating weight of an enlarging paper
• About the same as an enlarging paper, which is why the maximum densities are similar

Show the answer and why

Answer: About 0.019 g/dm² of silver metal, roughly twenty times the coating weight of an enlarging paper

An 8 × 10 inch sheet is 20.3 × 25.4 cm, or about 5.2 dm². Four millilitres of sensitiser carries 3.8 × 0.04 = 0.152 g of silver nitrate, and silver is 107.9 of the salt's 169.9 g/mol, so 0.152 × 107.9/169.9 = 0.0965 g of silver applied. Over 5.2 dm² that is about 0.019 g/dm², or 1.9 g/m². An enlarging paper at 1.0 g/m² is 0.01 g/dm². So the iron-silver coating puts roughly twice the silver on the paper - and Ware's Argyronomicon records that only about 2 per cent of the applied silver survives in a finished salt print, which is a far larger difference than the coating weight and is where the real comparison lies. The maximum densities are not similar: no baryta, no binder and no filamentary growth all work against this print.

Question 5. Predict what changes if the citrate in a Van Dyke sensitiser is replaced by oxalate. (Select all that the sources support.)

• The iron(II) becomes a stronger reducing agent, so platinum and palladium toning become chemically possible
• The print-out image becomes much fainter, because the iron(II) oxalate photoproduct is insoluble and immobile in a dry sheet
• A developer becomes necessary, and it works by dissolving the photoproduct rather than by reducing anything
• The sensitiser becomes more stable in the bottle and easier to store

Show the answer and why

Answer: The iron(II) becomes a stronger reducing agent, so platinum and palladium toning become chemically possible, The print-out image becomes much fainter, because the iron(II) oxalate photoproduct is insoluble and immobile in a dry sheet, A developer becomes necessary, and it works by dissolving the photoproduct rather than by reducing anything

The first three are what the kallitype is, and the next two labs test all three: the potential falls from +0.372 to +0.02 V, the print-out image becomes a guide rather than a picture, and the developer bath appears. The fourth is the opposite of what happens. Ferric oxalate is, in Ware's words, ill-characterised, polymorphic and notoriously variable, its solutions are said to decompose in six to nine months in the dark, and King limits a mixed batch to two or three months because it degrades with a resulting increase in print fog. Whereas a mixed Van Dyke sensitiser, on the Formulary's own sheet, "will remain active for months".

Question 6. Ware replaced silver nitrate with silver sulphamate in the argyrotype. What problem was he solving?

• Silver nitrate is too expensive at the concentrations these processes need
• Nitrate is an oxidising anion that dissolves the colloidal image silver in wet processing, especially in acid - so a nitrate process must be worked alkaline, and alkali hydrolyses the excess iron(III) into the image
• Silver nitrate is not soluble enough to give a sensitiser of useful strength
• Nitrate reacts with the citrate ligand and destroys the photosensitivity

Show the answer and why

Answer: Nitrate is an oxidising anion that dissolves the colloidal image silver in wet processing, especially in acid - so a nitrate process must be worked alkaline, and alkali hydrolyses the excess iron(III) into the image

This is the bind Ware names as the difficulty of the whole family: iron(III) clears best in acid, and colloidal silver survives best where an oxidising anion cannot reach it, which in a nitrate process means alkali. The kallitype's borax and Rochelle salt developers are the traditional answer and they buy the silver at the cost of hydrolysing the iron into the sheet. Substituting a non-oxidising anion removes the conflict and lets the sensitiser work at pH 3.5, where the iron does not hydrolyse, the silver stays in solution, and - Ware adds - the photosensitivity is at its optimum.

Question 7. A Van Dyke print comes out of the frame at about half its final density and is much darker after five minutes in the fixer. What is happening in the fixer?

• The fixer is reducing the remaining silver salt to metal, which is why over-fixing costs density
• Partial sulphide toning of the silver nanoparticles, which changes their colour rather than their quantity - and taken too far, converts them entirely to silver sulphide and fades the print
• The thiosulfate is complexing the residual iron, which removes a yellow cast and makes the image look darker by contrast
• The alkalinity of the bath is completing the reduction that the exposure began

Show the answer and why

Answer: Partial sulphide toning of the silver nanoparticles, which changes their colour rather than their quantity - and taken too far, converts them entirely to silver sulphide and fades the print

Ware attributes the colour shift from yellowish-red to a rich mahogany-brown to the nanoparticles acquiring a coat of silver sulphide perhaps only a few atoms thick, which changes the surface plasmon resonance profoundly, and cites energy-dispersive X-ray evidence of sulphur in the image. He also gives the failure mode from the same mechanism: overlong immersion transforms the particles completely into silver sulphide and badly fades the image. That is why every source in this part specifies a weak, short thiosulfate bath. Some genuine further development does occur in wet processing - Ware puts it at half to one stop in the high values - but it is not what the fixer is doing.

Check my answersClear