Wet-Plate Collodion: The Process That Made the Nineteenth Century Look Like That

The AIC’s conservation record for a collodion negative lists the photographer’s fingerprints in one corner of the glass plate among the object’s identification characteristics — not among its faults. Beside it, on the same list, is the absence of binder at all four corners, where the plate touched its holder. Both marks say the same thing: a hand held this object while it was wet, and there was no time to put it down.

That is the whole of the wet plate in two blemishes. Every other photographic material in this course can be made in one place and used in another. This one could not. The film had to be poured, sensitised, exposed, developed and fixed in a single unbroken run, because it stopped working when it dried — and everything about the process, from the portable darkroom to the fire risk to the particular look of the pictures, follows from that.

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

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

No procedure appears here. No quantity, no concentration, no ratio, no bath, no dilution, no development time, no coating instruction, and no sequence detailed enough to follow. Where a step has to be named for the chemistry to make sense it is named in the past tense, as something people did.

Two kinds of number are exceptions, and both are the kinds the policy allows. Historical exposure times appear because the material’s speed is one of the things this page has to explain, and a camera exposure is a fact about a material rather than an instruction. And one hazard calculation is worked below from a single millilitre of solvent, because the argument for the classification cannot be made without it; it is a conversion factor printed as evidence about a hazard, not a direction.

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

Where this page sits

Five pages carry parts of the collodion family, and they divide it by what question is being answered. Getting that straight first saves the same argument being made five times.

Who owns which question about the collodion family

1. 1The process atlas: what am I looking at?Three separate rows for three objects made from one chemistry — the wet-plate negative, the ambrotype and the tintype — each with identification, image, permanence and hazards in the atlas fields a conservator wants filled
2. 2The chemical encyclopaedia: what is this substance?Nitrocellulose as a polymer family and a fire load; diethyl ether as a solvent and an ignition hazard; cadmium bromide as a halide source at a carcinogen limit; silver nitrate at bath strength. Full hazard records, with sources and dates
3. 3Part XXVI, the cyanide lesson: what did the fixer do to people?Why cyanide fixed a silver iodide plate that thiosulfate could barely touch, what an acid stop bath does to it, and the occupational record. That argument is made once, there
4. 4This page: what is happening, and what did it cost?The mechanism at the level of the film and the particle; why the plate had to stay wet; a developer that plates rather than reduces; three objects out of one chemistry; the arithmetic of the solvent; and what present-day practice actually looks like
5. 5Part V and Part IV: what replaced it, and why that was different?Gelatin as a binder that can be precipitated, ripened, chemically sensitised, coated, dried, sold and kept — every one of which the collodion film cannot do

So identification is the atlas’s, with the ambrotype and the tintype as rows of their own; the substances are the encyclopaedia’s; cyanide as a working practice is its own lesson in this part; and collodion as a printing-out paper binder, which is a later and quite different use of the same polymer, is the collodion POP entry’s.

A solution, not an emulsion

Every silver-halide material this course teaches you to make is an emulsion: crystals precipitated in a beaker, ripened, washed, and then coated onto something. Part V is a whole part about that sequence. Collodion is not made that way, and the difference is the root of everything else.

Collodion is nitrocellulose dissolved in ether and ethanol. That is the entire definition; the AIC gives it in one line. It reached photography from surgery rather than from a laboratory: the DNB’s entry on Archer records that Schönbein discovered gun-cotton in 1846 and that in 1847 Maynard, of Boston, prepared collodion — an ethereal solution of gun-cotton — for surgical purposes. It was a wound dressing before it was a binder. Its solvent had a career in the same rooms, which PubChem’s ChEBI summary of diethyl ether records in two words: inhalation anaesthetic. Both halves of this material reached photography from surgery, and that is worth holding on to when the hazards are counted later: neither was chosen for photographic reasons in the first place.

Both solvents are needed and neither is a diluent. Archer’s own 1854 manual records the fact that confused his contemporaries: gun-cotton is not soluble in pure ether, that is, ether free from alcohol. The two also do different physical work, because ether boils at about 34 °C and ethanol at 78 °C. Ether is the one that leaves, and its leaving is what turns a poured liquid into a film; the alcohol is what stops it leaving too fast. A plate that sets before it is covered is unworkable and one that never sets runs off the glass.

The halide is not added; it is made, inside the film. Soluble iodides and bromides are dissolved in the collodion — Towler is explicit that only salts soluble in ether and alcohol can be used, so that the insoluble silver halide is produced in and on the film — and the halide itself appears only when the coated plate is lowered into a bath of silver nitrate:

KI + Ag⁺ → AgI + K⁺

Double decomposition inside a poured film: the sensitive substance forms where it will be used

A finished wet-plate negative, in section

Layer order and materials from the AIC's collodion negative record and IPI's negatives booklet; only the glass thickness is a published figure, and the organic layers are drawn at readable rather than true relative thickness because no source read here gives a collodion layer thickness. The image is not in the binder in the way a gelatin image is in its emulsion — it is on top of it. 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.

Why the choice of cation mattered, when in an ordinary precipitation it would not. In a beaker, the potassium of potassium iodide is a spectator: it washes away and the crystal does not remember it. Here the crystal grows inside a setting polymer film, so what the halide was dissolved in and what came with it decide how the film behaves. Towler sets out the trade with unusual clarity. A cadmium iodide glutinises a collodion; an alkaline iodide liquefies it. A collodion iodised with an alkali salt reaches its maximum sensitiveness quickly and loses it just as quickly; one iodised with a cadmium salt ripens slowly and then holds its sensitiveness far longer. So photographers combined the two, with the cadmium salt in excess — and Towler adds, with the honesty this course tries to match, that the best working proportions had not been satisfactorily determined.

Why the plate had to stay wet

“Wet plate” is not a nickname. IPI derives the term directly: the plate had to be exposed and processed while still damp with the sensitising chemistry to achieve optimum exposure speed. But “it stops working when it dries” is three different failures at three different moments, and they have three different mechanisms. Separating them is the most useful thing on this page, because only one of the three is the one usually quoted.

Before the silver bath: the halide forms in the wrong place. Archer gives the window as five to twenty seconds between draining the plate and immersing it, and states the failure exactly: if the drying is prolonged, the sensibility and evenness of the coating are injured and the iodide of silver is liable to be thrown out from the film on to its surface. He then explains what that costs, in a passage about over-iodised collodion that applies equally here — the image is formed on a layer of iodide of silver outside the collodion, so when the fixer clears the iodide away, the picture goes with it. A plate that dried for a few extra seconds does not give a faint picture. It gives a picture that vanishes in the fixing bath.

Between the bath and the developer: the film shuts. This is the mechanism everybody quotes, and the AIC states it plainly: once the alcohol and ether have dried, the collodion becomes impervious to the processing solution. A developer cannot reach a halide it cannot get to. The AIC puts the working window at under ten minutes from exposure, which is why a photographer working away from a studio had to take the darkroom with him — around 250 kilograms of equipment and materials, by the same record, to make one negative.

And underneath both: the free silver nitrate is part of the sensitive system. A plate lifted out of the bath is still wet with it, and Hardwich, writing about the preserved and dry variants, records the relationship as a straight trade: in proportion as the quantity of free nitrate is large, both sensitiveness and intensity are increased, while the keeping qualities of the plates are diminished. He also names the reason a plate cannot simply be left to dry with the bath liquid on it — the free nitrate concentrates as the water evaporates and eats away the iodide of silver, leaving transparent spots.

⚗How much speed the wetness was actually worth, and why the number is a range rather than a figure

The dry-collodion processes of the late 1850s and 1860s are the nearest thing the period offers to an experiment on the question, and they are not very near. Their common move was to wash the free silver nitrate out, lay on a preservative — tannin, honey, sugar, glycerol, dextrin, or albumen in Taupenot’s 1856 version — and let the plate dry.

Hardwich reports the result as a range, not a number: from four to forty times in favour of the wet plate. He also says why the range is that wide, and both reasons are about experimental design rather than about chemistry. The most sensitive wet collodions were rarely the ones used in the comparisons; and the workers disagreed about what counted as a fully exposed dry plate at all.

His own working comparison is the more useful figure, because it is one person’s practice rather than a survey: twenty seconds for an ordinary wet collodion against two and a half minutes for a tannin-preserved dry plate — about seven and a half times. He adds the constraint that makes the whole subject difficult: he had made dry plates considerably faster than that, but never without losing intensity.

The AIC’s account of the same experiments is blunter and covers the far end of the range: these early attempts often reduced sensitivity, and development sometimes took up to twelve hours.

What the course concludes, labelled as its own reading. The three failures above are separable in principle and were never separated in any measurement this course has read. Hardwich’s four-to-forty is a comparison of whole processes — different halide, different preservative, different developer, different worker — and not a measurement of what water does to a film. It establishes that the loss was large and that nobody could pin it down, and this page will claim nothing more precise than that.

The developer that plates rather than reduces

Here is the part of the process that matters most to the rest of the course, and it is also the part its inventor got wrong.

Part IV divides development in two. In chemical development, the image silver comes out of the crystal the latent image sits on, reduced in place. In physical development, the image silver comes out of the solution, plated onto the latent-image speck from outside. A wet collodion plate develops the second way, and it does so for a reason that is almost accidental: the plate is still wet with the silver nitrate bath, so the developer meets a film that already contains free silver ions.

Ag⁺ + Fe²⁺ → Ag + Fe³⁺

One electron from iron(II), one atom of silver out of solution: the step underneath the wet plate

IPI states the mechanism in exactly those terms — development, while the binder is still moist, with a solution containing a reducing agent and free silver ions — and records the two agents in order: pyrogallic acid in acid solution in the early years, iron(II) sulfate as the commonest developer by 1860. The iron developer’s own behaviour, including the fact that the acid rather than the iron is the throttle, belongs to its chemical page and is not restated here.

⁇Archer's own account of his own process is not the modern one, and his own troubleshooting contradicts it

This is worth setting out carefully, because it is a clean example of Rule 7 — established chemistry and historical evidence kept apart — on a page where the historical source is the inventor himself.

What Archer wrote. Development, in his 1854 account, is “nothing more than giving an increased density and depth to an exceedingly faint and delicate impression”, and it is “in fact, a continuation of the action of light, which action is the commencement of the reduction of the iodide of silver contained in the collodion to the metallic state, carried on … by the reducing agent employed.” Read as mechanism, that is a description of chemical development: the silver comes out of the iodide.

What Archer did. Two paragraphs later he tells the reader what to do when a developer has spent itself before the shadows have appeared: pour it off, and pour on a fresh supply with two or three drops of nitrate of silver solution added to it. If the metal were coming out of the halide, adding soluble silver to the developer would do nothing for the shadows. It works because the metal is coming out of solution, and the bath he is topping up is the supply.

And what he noticed without explaining. He records that the silver bath acquires, from constant use, a large quantity of alcohol thrown out of the film — enough to stop the developer flowing evenly — and gives the remedy of boiling the bath to drive the alcohol off. That is direct evidence that the film and the bath exchange material in both directions, which is the physical picture the modern account needs and the chemical one does not.

The course’s position. The modern statement is IPI’s and it is the one to hold. Archer’s is reported as what he believed, not as what happened, and his own two observations are printed beside it because they are better evidence than his theory. Nothing here is a criticism of Archer: chemical and physical development were not distinguished as categories until long afterwards, and a man who can see that adding silver to his developer rescues his shadows has seen the important thing.

The consequences of physical development are visible in the object. A physically developed image is plated onto the exposed specks rather than grown out of them, so it lies on and just under the surface of the binder rather than through its depth — which is IPI’s identification statement, and also why the layer is so easily abraded. It is why a collodion negative reads, by reflected light, as a dull milky-tan or brownish grey. And it is why the developer had to be poured across the plate in one motion: an unevenly poured developer leaves a record of its own arrival, which the Graphics Atlas marks up on its example negative under the name developer sweeps.

What the plate could see

The single most consequential fact about this material is not in its chemistry of development at all. It is that it could not see most of the world.

What an undyed collodion film responds to, against the rest of the visible spectrum

• Silver iodide alone (320–490 nm) — Abney, 1885: iodide of silver in its pure state is sensitive only to the ultra-violet, the violet and the blue rays
• Iodide with a bromide added (320–520 nm) — Towler, 1864: the photographed solar spectrum is much broader on a bromide film — violet, indigo, blue and partially green
• Green, yellow, orange, red: no record at all (520–720 nm) — A face in shadow, a red brick wall and green foliage are all, to this plate, closer to black than the sky is

Both edges are this course's rendering of verbal statements by Abney and Towler, following the same convention as Part IV's spectrum diagrams; neither is a measured curve, and this course has read no calibrated wedge spectrogram for a wet collodion plate. The 490 nanometre edge is the figure Part IV already uses for pure silver iodide. What is not in doubt is the shape: the response stops somewhere in the blue-green, and everything to the right of it is invisible to the plate. The coloured strip approximates where the visible spectrum falls and is a reading aid only; the wavelengths in the labels carry the information. The bands and curves are drawn to show the relationship, not measured.

The AIC’s ambrotype record states the working consequence for a photographer: the plate is only sensitive to blue and ultraviolet light, so blue sources are helpful and sunlight is effective. Towler, a century earlier, is doing better than that — he has photographed the solar spectrum on both kinds of film and reports the difference, and he refuses the conclusion his contemporaries drew from it. Bromides had been called accelerators, he writes, and on this ground alone the deduction would be false; high authority held them, from careful experiments, to be retarders of the actinic action. The true deduction is a greater capacity for colours.

That is an unusually clean piece of nineteenth-century reasoning, and it is worth naming what he has done: he has separated sensitivity, which is how much light a material needs, from spectral range, which is which light it can use at all. Part IV spends a lesson on the same distinction and reaches the same place from the physics.

📷Why the collodion look is a fact about a band gap before it is an aesthetic

Everything people mean by “the collodion look” in a portrait follows from the diagram above, and the reasoning below is the course’s own inference from a sourced spectral range, not a claim any source in this corpus makes about faces.

A material that records only blue and ultraviolet is not making a monochrome translation of what an eye sees. It is making a picture of one narrow property of a surface: how much short-wavelength light it sends back.

• Skies print empty. A blue sky is bright in exactly the band the plate is most sensitive to, so it overexposes into the clear glass of the negative long before the land beneath it has recorded anything. Every landscape from this era with a blank white sky is showing you a spectral response, not a photographer’s choice.
• Reddened or freckled skin goes dark. Skin that reddens, and the melanin concentrations in freckles, absorb in the blue. To a plate that can see nothing else, they are close to black — which is why the process is unkind to weathered faces in a way panchromatic film is not.
• A blue iris goes pale and a brown one goes dark, for the same reason and with more visible effect, because the two are side by side in the same picture.
• And the exposure has to be long even in sun, because most of what is arriving is unusable. IPI gives the historical camera exposure as twenty seconds to three minutes, depending on the light, the temperature and the humidity.

The honest limit of this reasoning: it explains tendencies, not a particular plate. Image colour, contrast and the appearance of skin also depend on the developer, the fixer and the varnish, and IPI publishes a table showing that the developer and the fixer each shift the image colour on their own.

Three objects out of one chemistry

The same plate, developed with different intentions and finished in different ways, produced three kinds of object that a collection catalogues separately. Understanding that they are one material is worth more than the three entries separately.

One chemistry, three objects, distinguished only by what is behind the film and whether anyone printed from it

further from the negative the chemistry naturally makes

1. The collodion negativeDeveloped fully, on clear glass, and printed from by contact — usually onto albumen paper. This is the object that made photography reproducible: a support with no fibres to scatter the printing light, and no limit on the number of printsa matrix
2. The ambrotypeThe same plate exposed and developed to a deliberately weak image, then backed with black — varnish, cloth, metal, paper, or a dark ruby glass used as the support itself. The pale scattering silver reads as highlight against a ground that returns almost nothing, and the clear glass reads as shadowa unique image
3. The tintype, properly the ferrotypeThe same weak image on thin sheet iron under a polished black japan. Towler names both halves of the name: melainotype from the black background, ferrotype from the iron. The dark ground is the support, so no backing and no glass are needed — though early tintypes were cased like ambrotypes all the samea unique image, and postable

Chemically these are one process; the atlas gives them three rows because a conservator meets them as three objects. Nothing is reversed in an ambrotype or a tintype: the polarity is supplied by the ground, which is why taking the backing off an ambrotype turns it back into a negative.

Nothing about an ambrotype is positive, and the AIC’s definition says so in its first line: it is an under-exposed wet collodion negative on glass that appears positive because of a dark backing or a dark support. Archer, whose own book already discusses positives and negatives as two intentions for the same plate, gives the backing advice from experience: cotton velvet behind the plate in the case, in preference to black varnish, which is apt to penetrate and reduce the picture several tones, or to paper, which is seldom of sufficient depth.

And they are laterally reversed, because a direct positive is the surface that faced the lens. Towler treats this as a working constraint rather than a curiosity: every part of such a picture is inverted, so a sportsman raises his gun in the wrong hand, and the application of collodion positives is therefore limited to portraiture. He also records the way round it — the alabastrine variant, in which the plate is turned over and the image is viewed through the glass, in its natural position. The tintype could not do that at all, because its support is opaque; Towler says so explicitly.

The tintype’s advantage was never the picture. It was the object: unbreakable, light, and flat enough that Towler describes cutting the corners off with shears and folding them into a mat for mailing in a letter. Its one process-specific hazard is also an object property — iron conducts heat far better than glass, so a careless flame during varnishing raises the japanned film into blisters.

Fixing, and the choice that stays visible in the object

Fixing a wet plate was the hardest version of the standard problem, because the plate is largely silver iodide — much the least soluble of the three halides. IPI records both answers as genuinely coexisting: sodium thiosulfate was the primary fixer of the 1850s, and potassium cyanide, introduced for ambrotypes, gave the wet-plate photographer a faster second agent that was also, in IPI’s word, very poisonous. Both continued in use for negatives throughout the era.

Why cyanide worked so much better on this material is arithmetic, and the cyanide lesson does it — the formation constants, the combined constant across the three halides, and the finding that thiosulfate is thermodynamically barely able to touch silver iodide while cyanide handles it easily. That page also carries the occupational record and the reaction with acid that makes the salt lethal in a room. None of it is restated here. What belongs to this page is the part that shows in the photograph.

The fixer chose the image colour, and a conservator can still read the choice. IPI publishes it as a process variation with a visible signature: fixing in hypo gives a darker, brownish-grey image, fixing in cyanide a lighter, milky tan. Towler explains the mechanism as the period understood it — cyanide is not only a solvent of the silver salts but also a reducing agent, and it produces in the ambrotype and the melainotype a whiteness in the silver film which hyposulphite cannot give. So it was regarded as the fixer peculiarly adapted to collodion positives viewed by reflected light, while hyposulphite was the proper fixer for negatives, where whiteness does not matter.

A person holding an ambrotype with milky-white highlights is therefore looking at the residue of a decision somebody made about how white they wanted the face to be — at a cost this course would not now accept, and, on Towler’s own account, at a cost to the picture too, since a cyanide bath strong enough to work quickly dissolves the fine parts of the image along with the halide.

🕰Archer's own route to a whiter positive was a mercury bleach, and it is the same chemistry as an intensifier

Archer’s manual has a section on turning a feebly developed picture into a white positive, and then into a dense negative, and the reagents are worth naming even though no quantity from it appears here.

The whitening agent is corrosive sublimate — mercury(II) chloride in acid — and Archer’s own warning about it is mechanical rather than toxicological: it is a very corrosive compound, and a weak collodion will not survive it. The bleached plate is then blackened again, either with hyposulphite of soda or, faster and with more care needed, with potassium cyanide — which, he notes, can overrun and bleach the picture before water can be got to it.

That is a bleach-and-redevelop cycle, and it is the ancestor of the mercury intensifiers that survived into gelatin practice. The chemistry of that family, and why toning and intensification stop being different operations when mercury is the reagent, belongs to Part XX and to the heavy-metal treatments. What this page adds is the date: the inventor’s own manual, in 1854, already puts mercury and cyanide into the ordinary finishing of a collodion picture.

The varnish is not a finish; it is the reason the object still exists

Every source read here agrees, and the agreement is unusually flat. Archer: a collodion picture, positive or negative, requires varnishing to protect it from injury. IPI: unvarnished collodion negatives are extremely fragile, easily abraded and easily oxidised, and the coat also prevents the silver image from oxidising. The AIC: an unvarnished collodion layer is very fragile and must be treated with care.

That follows directly from where the image is. Physically developed silver sits on and just under the surface of a layer only microns thick. There is nothing above it. A varnish is not a gloss; it is the only thing between the picture and a thumb.

And varnishing has its own two-part problem. The resins used — sandarac and shellac in IPI’s account, shellac or copal or a home-brew of many components in the AIC’s, white lac in Archer’s — are carried in alcohol, and alcohol dissolves collodion. So an aqueous pre-coat goes on first, of gum arabic, dextrin, albumen or gelatin, precisely so that the varnish’s solvent cannot reach the binder. A good deal of the practice of this process is about protecting the material from its own solvents.

The second part is heat. Archer’s instruction is that the plate be gently warmed before the varnish is poured and warmed again afterwards to dry it. A modern supplier’s sheet still describes warming the plate over an open alcohol flame, still warns that the varnish is highly flammable, and still tells the reader what to do if the plate catches fire. A step that applies a naked flame to a solvent-borne resin lying on a nitrated-cellulose film is not an incidental detail of this process, and it is one of the reasons the classification comes out where it does.

Why it won, and why it lost

It won on law as much as on chemistry. Archer published, and did not patent. The DNB’s entry is careful about what he did and did not originate — he “does not seem to have been the first to suggest this application of collodion, but there appears no doubt whatever that he was the first to carry it into effect” — and equally careful about the consequence, recording that he did not patent the invention, “possibly because he did not realise its value”, though he did patent a development of no practical value four years later. He died in May 1857; a subscription was started for his widow, and when she died the following year the amount — over £600 — was devoted to his children, to whom the Crown also granted a pension, on the explicit ground that their father had reaped no benefit from an invention which had been a source of large profits to others.

Set that beside a daguerreotype, which was patented in England, and a calotype, whose patent Talbot went to court over and lost — in December 1854, against a photographer the DNB names Sylvester Laroche, whose collodion development he held to infringe it — and the commercial outcome stops being mysterious. Two years before that action Talbot had already, at the request of the presidents of the Royal Society and the Royal Academy, thrown his discoveries open except for portraiture. Both dates are after collodion had arrived, free, and Part I tells that story from Talbot’s end. The first genuinely good process that nobody had to pay to use displaced both of them within a decade. It was sharper than a calotype, because glass has no fibres; it was far cheaper than a daguerreotype; and unlike either of the two 1839 processes taken as a pair, it gave you both a matrix you could print from many times and, if you wanted one, a unique object made while the sitter waited.

⁇Who invented it, and what the sources will and will not support

Established. Collodion existed as a surgical preparation before it was photographic. Archer applied it to photography, published in The Chemist in March 1851, published a book-length account in 1854, and did not patent the process.

Disputed, and this course does not adjudicate it. The AIC records a claim for Gustave Le Gray between 1850 and 1851, notes that his method was received as “theoretical at best”, and that he abandoned coated glass to pursue paper. The DNB, writing in 1885 with living witnesses available, separates the two questions cleanly: first to suggest the application, uncertain; first to carry it into effect, Archer without doubt. That distinction is the honest form of most invention claims in this subject, and Part I makes the same point about 1839.

Not established. The course has read no account resolving the priority question, and repeats neither “Archer invented photography on glass” nor “Le Gray got there first” as a settled fact.

It lost to a material that could be manufactured. The AIC dates the replacement to the 1880s and gives the reason in one clause: the gelatin silver process was equally sensitive but had long lasting, portable plates. IPI’s date band for collodion glass negatives is 1851 to about 1885.

That is worth reading as a statement about binders rather than about speed, because the speed had been matched rather than beaten. A gelatin emulsion can be precipitated in a vessel, ripened, chemically sensitised, washed, coated, dried, boxed, sold, stored and exposed months later — and Part V exists because every one of those verbs is a separate piece of chemistry. A collodion film can do none of them, for the one reason this page keeps returning to: its permeability depends on solvents that are leaving.

The hazards, counted

Five families of hazard meet in one operation, which is unusual even among the Level D processes and is why this part gives collodion a lesson of its own. Four of the five have full records elsewhere in this course and are named here rather than restated.

Nitrocellulose is the fuel, and it is its own encyclopaedia entry. The one line to carry from the safety card: the photographic grade is stored only if damped, a spill is kept wet and not allowed to dry out, and a fire is fought with water in large amounts and with no foam and no carbon dioxide — because smothering does not work on a fuel that carries its own oxidant.

Cadmium bromide in the salting mixture is its own entry too, and it is the case that teaches how to read a hazard record. Its aggregated GHS headline is mild — signal word Warning, harmful by three routes, very toxic to aquatic life. The regulator disagrees: EH40 gives cadmium and its compounds 0.025 mg/m³ as cadmium with the Carc notation, NIOSH marks its cadmium entry as a carcinogen and states that the limits apply to all cadmium compounds, and the same PubChem record carries NITE-CMC blocks classifying the substance H340 and H350. It was weighed as a dry powder and then dissolved in a flammable solvent.

Silver nitrate at bath strength, in a standing dip tank rather than a brushed-on film, with nitric acid used to hold its pH. Cyanide, where the period fixing was followed, has its own lesson in this part.

The fifth family is the solvents, and it is the one that decides the classification. It has two members, and the smaller one is worth naming first, because it is the one that gets forgotten. Ethanol is not the mild half of the pair. Its safety card gives a flash point of 12 °C and explosive limits of 3.1 to 27.7 volume per cent in air, so an ordinary room is already above its flash point too. It is also the solvent of the varnish, which is why an open flame appears twice in this process rather than once.

The other member deserves the arithmetic.

🧮What a millilitre of ether does to a room, from NIOSH's figures and the ideal gas law

Diethyl ether’s hazard record is not in doubt, and its own page carries it in full. NIOSH gives a specific gravity of 0.71, a vapour pressure of 440 mmHg — more than half an atmosphere at room temperature — a flash point of −49 °F (−45 °C), an explosive range from 1.9 to 36.0 per cent in air, and the bracketed note that the substance is a gas above 94 °F (34 °C). It classes the liquid Class IA flammable, its most volatile category. CAMEO, through PubChem, adds the property that decides how a spill behaves: the vapours are heavier than air.

Two sums, both the course’s own, from those figures and the relative molecular mass of 74.12.

First, how much vapour a small spill makes. One millilitre of the liquid weighs 0.71 g, which is 0.0096 mol, and a mole of any gas occupies about 24.1 litres at 20 °C and ordinary pressure:

V = (0.71 g ÷ 74.12 g/mol) × 24.1 L/mol ≈ 0.23 L

Vapour from one millilitre of liquid ether at 20 °C

Second, how much air that vapour brings into the explosive range. The lower limit is 1.9 per cent by volume:

V(air) = 0.23 L ÷ 0.019 ≈ 12 L

Air brought to the lower explosive limit by one millilitre of liquid

About twelve litres of air per millilitre of liquid. A spill of a few millilitres — a drip from a poured plate, the last of a bottle, what runs off the corner while the plate drains — is enough to make a bucket-sized volume of air ignitable, and the vapour does not stay where it was spilled.

Why it does not disperse upwards. Ether’s relative molecular mass is 74.12. Nitrogen’s is 28.0 and oxygen’s 32.0, so a given volume of ether vapour weighs roughly two and a half times what the same volume of air weighs. It sinks, runs along a bench, crosses a floor and finds an ignition source that may be several metres away and below the person who spilled it.

And why the exposure limit is the wrong instrument here. EH40 sets diethyl ether at 100 ppm over eight hours and 200 ppm over fifteen minutes; OSHA’s permissible limit is 400 ppm; NIOSH publishes no recommended limit at all and refers the reader to its appendix of substances without one. Those are toxicological figures about breathing ether across a working day. The explosive range starts at 1.9 per cent, which is 19,000 ppm — nearly two hundred times EH40’s long-term limit. The two hazards do not share a scale, and the smaller number is not the governing one.

The assumptions, stated so they can be attacked. Ideal-gas behaviour at 20 °C and one atmosphere; complete evaporation; and perfect mixing into the air volume quoted, which a real room does not do — in the direction that makes it worse rather than better, because a heavier-than-air vapour pools rather than diluting. And the calculation says what a source is capable of, not what a particular room contained. What it establishes is the size of the gap a control would have to close.

The hazard has a specific shape, and it is not a dose. Gloves do not address it. A mask does not address it. Working carefully does not address it, because a vapour denser than air with a flash point of −45 °C finds its own way to an ignition source. What addresses it is removing every ignition source from a volume you cannot see into — which is what a flammable-solvent facility is, and what a domestic room, with a boiler pilot light, a light switch and a jumper that discharges static when it comes off, is not. ICSC 1560 names that last one explicitly for the nitrocellulose sitting in the same solution.

And the solvent does not stay in the collodion. Archer already knew this in 1854: the silver bath acquires, from constant use, a large quantity of alcohol thrown out of the film, and he gives the remedy as boiling the bath to drive it off. A present-day supplier’s sheet describes exactly the same maintenance operation, naming collodion, ether and alcohol as the contaminants and offering two routes — days of sunning in an open vessel, or boiling on a hotplate. The second of those is the deliberate heating of a solution known to contain ether, and a reader assembling this process at home would have to do one of the two.

What present-day practice actually looks like

This process is practised now, and practised well. Pretending otherwise would be dishonest and would also make the classification easier to dismiss than it deserves to be. The AIC’s record lists contemporary practitioners by name; the honest statement is that the people who do this safely learned it in person, from somebody who already knew.

Reading a current supplier’s kit instructions as a manufacturer document, which is how this course reads any maker’s sheet, tells you what present practice consists of — and it corrects two things this lesson’s own manifest entry assumed.

⁇Two corrections this page makes to its own brief, from the supplier's own document

The manifest entry for this lesson anticipated that modern practitioners had moved to “cadmium-free salting” and thiosulfate fixing. Half of that holds and half of it does not, and the page has been written to the evidence rather than to the brief. The manifest has been corrected to match.

Thiosulfate fixing: confirmed. Bostick & Sullivan ship an ammonium thiosulfate rapid fixer and describe it as “a safe and sane alternative to deadly cyanide-based fixers”. The comparison is sound; the adjective is not one this course applies to any photographic chemical, and Part XI gives thiosulfate’s own hazards and its waste route. Cyanide fixing is, on this evidence, gone from the commercial kit route.

Cadmium-free salting: not confirmed. The same kit’s bromo-iodiser has its composition printed on the contents page, and it is cadmium bromide, ammonium bromide and potassium iodide in a half-and-half alcohol and ether solution. The salt that made a cadmium collodion keep its sensitiveness is still what is being sold, a century and a half after Towler explained why, and the cadmium bromide entry already records the same finding. A reader who had been told the modern practice is cadmium-free would buy a carcinogen at a 0.025 mg/m³ limit believing otherwise, which is exactly the kind of error Rule 7 exists to prevent.

Three other things in that document are worth reading as evidence about the operation rather than as instructions.

The sheet recommends supervision. In its own words, a beginner should be under the supervision of an experienced wet-plate photographer. It also says, of itself, that it is a basic guide to the kit and “in no way should be considered a working guide to the wetplate process”. A manufacturer telling you that its instructions are not sufficient instructions is a finding about the process.

Its protective-equipment list ends with a gas mask, marked optional. Read that beside the arithmetic above. The hazard the sheet’s own warnings emphasise — flammable vapours, chemicals to be kept away from a water heater or any other flame or pilot light — is not one a respirator addresses at all, and the item on the list that would matter most, an extracted workspace with no ignition sources in it, is not equipment that can be put in a box.

And the flame is still there. The varnishing step still involves an alcohol lamp, an open flame and a warning about what to do if the plate ignites.

Running the assessment yourself

Part XX sets out the four questions this course applies when it puts something at Level D, and this part’s closing assignment will ask you to run them on a process you have not met. Here they are on this one.

1. What are the hazards, from a named source with a date? Diethyl ether: harmonised CLP classification, Danger, H224 extremely flammable liquid and vapour, with H302 and H336; NIOSH’s flash point, explosive range and peroxide note; EH40 at 100 ppm long-term. Nitrocellulose: ICSC 1560, a flammable solid with an explosion risk on friction or shock, to be stored only damped, with no health data at all — an absence the card itself flags. Cadmium: EH40 0.025 mg/m³ with Carc. Silver nitrate and cyanide: their own pages. Not one reputation among them.

2. What control addresses each one, and can a domestic reader assemble it? The governing hazard is ignition of a heavier-than-air vapour in a small dark room, and the control is engineered extraction with every ignition source removed from a volume the worker cannot see into. A domestic reader cannot assemble that, and — importantly — cannot verify that they have. The cadmium hazard needs a second and different control, extraction at the point where a powder is weighed, plus a waste contract.

3. Is any hazard one the sources cannot characterise? Yes, and it is worth naming because it is easy to miss under the fire risk. ICSC 1560 prints “See Notes” against routes of exposure, short-term effects and long-term effects alike, and its note reads that insufficient data are available on the effect of this substance on human health. You cannot specify a control against a hazard nobody has characterised, and EH40’s own caution — that absence from the list does not indicate that a substance is without risk — applies exactly here.

4. Is there a route to the same photograph that this course can teach? No, and this is where collodion resembles the daguerreotype rather than the toners. There is no version of a wet plate that is not a wet plate: the object’s whole character comes from a halide formed inside a solvent-cast film that has to stay permeable, and the solvent is the hazard. Remove the ether and there is no film; leave the ether and there is no control a page can assume. The course is not withholding a milder form of this process. There is not one.

The object in a collection

The likeliest way a reader of this course meets any of this is not in a workshop. It is a box of glass plates in an attic, or a small hinged case, and the useful skills are identification and restraint.

Identifying a wet plate is mostly identifying a handmade object. IPI’s list is the practical one: view the negative against a dark background and the image reads as a positive; then look for rough-cut edges, glass thicker than machine-made plates, a plate that is not quite square, an uneven glass surface, uneven edges to the binder and the varnish, and an uncoated corner where the plate was held. The AIC adds the fingerprint. A gelatin dry plate — machine-coated, on machine-made glass — has none of these, and the two are otherwise easy to confuse. The Graphics Atlas’s marked-up features on its own example negative are the same list read as a picture: glass side, image tone, developer sweeps, varnish lines, surface abrasions, silver image particles.

A tintype identifies itself. Iron under a magnet, no glass to look through. An ambrotype is identified by taking nothing apart at all: it is a transparency with something dark behind it, and against a light table the dark ground shows itself for what it is. The AIC records the ruby variant, made on dark glass that reads red in transmitted light, alongside the common sort with a separate backing, and adds one analysis note that settles a common question: a varnish fluoresces under ultraviolet, and an unvarnished plate shows white or red-yellow tarnish layers over the image where a varnished one does not.

!What the conservation sources actually advise, and the failure that looks like a different failure

The plate that looks faded may not be. Where an ambrotype’s applied backing has flaked or lifted, the image silver is intact and its ground has gone. On a silver gelatin print a pale image means lost silver; here it may mean lost black paper. That inverts the usual diagnostic instinct, and the AIC’s remedy is correspondingly gentle: a dark secondary support behind the plate, rather than anything done to the plate.

Do not put solvents on it. The AIC is unusually specific: varnishes, paints and the collodion layer are all sensitive to many organic solvents, even water can dissolve a weakened collodion layer, and ethanol and acetone are not recommended. An unvarnished collodion layer should have nothing but a soft brush near it.

Store glass plates dry, cool and vertical. IPI gives below 18 °C and 30 to 40 per cent relative humidity, with under 40 per cent critical because the real enemy is the glass rather than the image: excess alkali in a period glass leaches out, softens the binder and varnish from below and appears as moist droplets in damp storage or crystals in dry. Below 30 per cent is not recommended either, because it dehydrates the glass surface. The AIC asks for individual envelopes in custom boxes, stored vertically where possible.

A cased ambrotype is a composite object. The AIC asks for a non-fluctuating environment around 68 °F and 45 per cent relative humidity, and warns specifically against freezing, because condensation during thawing may be catastrophic. The Graphics Atlas’s cased-object note describes the package — photograph, brass mat, cover glass, preserver, case — and records a detached lid as a common form of deterioration for cased objects.

And one failure is built in and cannot be repaired. Where the glass itself was badly made, the AIC records that the collodion and varnish go soft or crack, and that there is no easy fix because the problem is inherent. The Graphics Atlas records a second self-inflicted one: ambrotypes sealed under Cutting’s 1854 Canada balsam patent are known to deteriorate because of the sealing, and the example it publishes shows it as a yellowish-green cast. A protective measure that turned out to be the damage is worth remembering the next time one is proposed.

If you intend to pursue it, where to go

This course will not teach you this process, and it will not pretend that means nobody should. What follows is not a way in and not a shortened version of one: it is three things to do before any of the chemistry, and each of them is about who teaches you and what you have read, not about a plate.

Learn it in person, from somebody who already does it. The supplier’s own document says so, and it is the only recommendation here that comes from the people who sell the materials.

Read the supplier’s instructions as what they are — a manufacturer document about a specific product, with the hazards its maker has chosen to state — and read them alongside the safety data for each substance separately, because a kit sheet is not a hazard assessment and does not claim to be.

Write the risk assessment before you buy anything, not after. The four questions above are the shape of one. If the answer to the second question is that you cannot assemble the control, the answer has arrived before the parcel has, which is the whole point of doing it in that order.

Recap

• Collodion is a solution, not an emulsion. Nitrocellulose dissolved in ether and ethanol, salted with soluble iodides and bromides, poured onto glass; the silver halide forms inside the setting film when the plate is dipped in silver nitrate. Nothing is precipitated, ripened or chemically sensitised, because there is no emulsion to do it to.
• “Wet plate” describes three separate failures, not one. Dry it before the silver bath and the halide forms on top of the film, so the fixer takes the picture with it; dry it before development and the binder becomes impervious; and drying costs the free silver nitrate that Hardwich shows is itself part of the sensitivity.
• Development is physical, and the inventor’s own book proves it against his own explanation. Archer described a reduction of the iodide and then told his readers to add silver nitrate to a spent developer to rescue the shadows — which only works if the image metal is coming out of solution.
• The look is a spectral fact. Undyed silver iodide responds only to ultraviolet, violet and blue; a bromide addition reaches partway into the green. Towler had already separated range from speed in 1864 and refused to call bromides accelerators.
• One chemistry, three objects. A negative to print from; an ambrotype, which is a thin negative read against black; and a tintype, which is the same thing with the black built into the support. Nothing is reversed in either positive — the ground supplies the polarity.
• The varnish is structural. The image lies on the surface of a layer microns thick, so an unvarnished plate can be wiped away; and the varnish needs an aqueous pre-coat, because its own solvent dissolves the binder it is protecting.
• The hazard is ignition, and it is arithmetic. One millilitre of ether makes about 230 millilitres of vapour, enough to bring twelve litres of air into the explosive range; the vapour is about two and a half times as dense as air and travels; and the explosive limit sits nearly two hundred times above the exposure limit, so the toxicological figure is measuring the wrong thing.
• Present practice is real, and it still uses cadmium. A current kit’s bromo-iodiser is cadmium bromide with ammonium bromide and potassium iodide; its fixer is ammonium thiosulfate rather than cyanide; and its own sheet asks for supervision by an experienced worker and warns that the varnishing flame can set the plate alight.
• If you own one, do very little. Dark background to identify it, magnet for a tintype, no solvents, no water, individual envelopes, cool and dry, and a black card behind an ambrotype whose backing has failed rather than anything done to the plate itself.

Check your understanding

Question 1. A wet collodion plate is allowed to dry for a minute too long between draining and the silver bath, and the picture develops normally but disappears in the fixer. What has happened?

• The collodion has become impervious to the fixing solution, so the halide cannot be removed
• The silver iodide has been thrown out of the film onto its surface, so the image forms on a layer of halide sitting outside the collodion; when the fixer clears that halide away, the image goes with it
• The free silver nitrate has evaporated, so no latent image formed at all
• The bromide has oxidised, which makes the image soluble in thiosulfate

Show the answer and why

Answer: The silver iodide has been thrown out of the film onto its surface, so the image forms on a layer of halide sitting outside the collodion; when the fixer clears that halide away, the image goes with it

Archer gives both the window and the failure in his 1854 manual: five to twenty seconds may elapse between draining and immersion, and if the drying is prolonged the sensibility and evenness of the coating are injured and the iodide of silver is liable to be thrown out from the film onto its surface. He then explains the consequence in a passage about over-iodised collodion that applies identically here — the image is formed on a layer of iodide of silver outside the collodion, so when the iodide is cleared off, the picture goes with it. The imperviousness of a dried film is a real and separate failure, but it happens later in the sequence, between sensitising and development, and it prevents development rather than fixing.

Question 2. What makes wet collodion development physical development rather than chemical development, and what evidence in Archer's own book supports it?

• The plate is physically agitated, which distinguishes it from a still developer
• The plate comes out of the silver bath still wet with it, so the film contains free silver ions that iron(II) reduces onto the exposed specks; Archer instructs the reader to add nitrate of silver to a spent developer to bring out the shadows, which only helps if the image metal comes from solution
• Mercury vapour condenses on the plate, which is a physical rather than a chemical change
• The image is deposited by evaporation of the solvent rather than by any reaction

Show the answer and why

Answer: The plate comes out of the silver bath still wet with it, so the film contains free silver ions that iron(II) reduces onto the exposed specks; Archer instructs the reader to add nitrate of silver to a spent developer to bring out the shadows, which only helps if the image metal comes from solution

IPI states the mechanism directly: development, while the binder is still moist, with a solution containing a reducing agent and free silver ions. The step is one electron from iron(II) for one atom of silver out of solution. Archer himself described development as a continuation of the action of light reducing the iodide of silver, which is an account of chemical development, but two paragraphs later he tells the reader to pour off a decomposed developer and pour on a fresh supply with two or three drops of nitrate of silver added, to bring out the faint parts. That instruction is evidence against his own explanation and for the modern one. The categories were not distinguished until long afterwards, so this is a case of a good observer without the vocabulary rather than a careless one.

Question 3. Towler reports that the photographed solar spectrum is much broader on a bromide film than on an iodide one, and then explicitly refuses to call bromides accelerators. What distinction is he making?

• He is distinguishing between the negative and the positive process
• He is distinguishing spectral range — which light a material can use at all — from sensitivity, which is how much light it needs; high authority held bromides to be retarders of the actinic action, so the broader spectrum is a capacity for colours rather than extra speed
• He is distinguishing between cadmium and alkaline bromides
• He is distinguishing between reflected and transmitted light

Show the answer and why

Answer: He is distinguishing spectral range — which light a material can use at all — from sensitivity, which is how much light it needs; high authority held bromides to be retarders of the actinic action, so the broader spectrum is a capacity for colours rather than extra speed

It is the same distinction Part IV builds a lesson on, reached from photographic practice rather than from physics. A bromide film records violet, indigo, blue and partially the green where an iodide film represents only part of the blue, so a scene with foliage in it yields more; that is a wider range. Whether the material needs more or less light within the range it does respond to is a separate question, and Towler reports that careful experiments made bromides retarders rather than accelerators. Getting these two confused is the commonest error in reading period claims about speed, and Towler names it as a false deduction in 1864.

Question 4. One millilitre of diethyl ether evaporates in a small room. Using NIOSH's figures, what does the arithmetic establish, and what does it not?

• That anyone in the room receives a dose above the occupational exposure limit
• That the millilitre makes about 230 millilitres of vapour, enough to bring roughly twelve litres of air to the lower explosive limit — an upper bound on what the source can do, not a statement about a particular room, which will be worse in one respect because the vapour is denser than air and pools rather than dispersing
• That EH40's 100 ppm limit is the governing figure for the hazard
• That the room becomes explosive throughout, since ether mixes readily with air

Show the answer and why

Answer: That the millilitre makes about 230 millilitres of vapour, enough to bring roughly twelve litres of air to the lower explosive limit — an upper bound on what the source can do, not a statement about a particular room, which will be worse in one respect because the vapour is denser than air and pools rather than dispersing

From a specific gravity of 0.71 and a relative molecular mass of 74.12, one millilitre is 0.0096 mol, which is about 0.23 litres of vapour at 20 degrees Celsius; divided by the lower explosive limit of 1.9 per cent by volume, that is about twelve litres of air brought into range. The calculation says what the source is capable of and not what anybody breathed or what ignited. The exposure limit is measuring a different hazard on a different scale: 1.9 per cent is 19,000 ppm, nearly two hundred times EH40's long-term figure of 100 ppm, so a room can be far past the point where anyone should be in it and still be nowhere near a fire, or the reverse near the floor. And the vapour is roughly two and a half times as dense as air, so it runs along a bench or a floor to an ignition source rather than dispersing upwards.

Question 5. A currently sold wet-plate collodion kit is examined as a manufacturer document. What does its stated composition establish about modern practice?

• That modern practice has moved to cadmium-free salting and thiosulfate fixing
• That the fixing has moved to ammonium thiosulfate, but the bromo-iodiser is still cadmium bromide with ammonium bromide and potassium iodide in alcohol and ether, so the cadmium has not gone
• That modern kits no longer contain ether
• That the process is now performed without any flame or heating step

Show the answer and why

Answer: That the fixing has moved to ammonium thiosulfate, but the bromo-iodiser is still cadmium bromide with ammonium bromide and potassium iodide in alcohol and ether, so the cadmium has not gone

This lesson's manifest entry anticipated cadmium-free salting, and the supplier's own contents page does not support it: the iodiser's stated composition is cadmium bromide, ammonium bromide and potassium iodide in a half-and-half alcohol and ether solution. The fixing half of the expectation does hold — an ammonium thiosulfate rapid fixer is supplied and offered as an alternative to cyanide fixers. The manifest was corrected rather than the page written to it. The distinction matters practically as well as editorially: a reader told that modern practice is cadmium-free would be buying, unwarned, a substance EH40 lists at 0.025 milligrams per cubic metre with a carcinogen notation. The varnishing step in the same document still uses an open alcohol flame.

Question 6. An ambrotype in a family collection appears to have faded almost to nothing. What is the most likely explanation, and what does a conservator do about it?

• The silver image has been lost to oxidation, and the plate should be redeveloped
• The applied dark backing has flaked or lifted, so the image silver is intact and its ground has gone; the response is to place a dark secondary support behind the plate rather than to treat the plate
• The varnish has yellowed and should be removed with ethanol
• The glass has become opaque and the plate should be rehoused in a dry environment

Show the answer and why

Answer: The applied dark backing has flaked or lifted, so the image silver is intact and its ground has gone; the response is to place a dark secondary support behind the plate rather than to treat the plate

An ambrotype is an under-exposed collodion negative that reads as a positive only because something dark sits behind it, so the characteristic loss is mechanical and inverts the usual instinct: on a silver gelatin print a pale image means lost silver, and here it may mean lost backing. The AIC's remedy is a dark secondary support such as black matboard or a toned board, with consolidation and inpainting where a conservator judges them appropriate. The wrong moves are the tempting ones: the AIC records that varnishes, paints and the collodion layer are all sensitive to many organic solvents, that ethanol and acetone are specifically not recommended, and that even water can dissolve a weakened collodion layer.

Check my answersClear