The Level D policy

A course that refuses to teach the dangerous half of its own subject is not being careful; it is being incomplete, and it leaves a reader who meets a mercury-developed plate in a museum, or a nineteenth-century formula in a reprint, with no way of understanding what they are looking at. Level D is how this course teaches those processes without ever telling you to perform one.

The letter is assigned by the same rubric as the other three, on the safety classification page, and it answers the same question the other levels answer: what does a reader need to have, and be able to do, for this page to be reasonable? For Level D the answer is that no set of controls a domestic reader can assemble makes the procedure reasonable, so the page gives no procedure at all.

The rule this page exists to state

A Level D page carries no Procedure section, no Preparation section and no Build section.

That is not a style preference. npm run validate:labs reads it as a rule and fails the build on a Level D page that carries any of those three headings. The check is keyed off the page’s safetyLevel and deliberately not off its pageType, because every Level D entry in the course manifest is a lesson or an assignment rather than a lab: gating the rule on the practical page types left the course’s most safety-critical rule sitting in the repository, reported as passing, and never once executed.

☠What a Level D page may and may not contain

May, and generally must: the chemistry, written as reactions with mechanism; the historical workflow in outline, told as what people did rather than as what you do; the image characteristics and why the chemistry produces them; the hazards, with named sources and dates; the reason the process is classified here rather than at Level C; where a safer modern route to the same result is taught in this course; and, where institutions or conservators work with the process under controlled conditions, a pointer to their published accounts.

May not, at any length or in any tone: quantities to weigh out; a sequence of steps to follow; a parts list; a temperature and time to hold; a substitution that “makes it safer”; or a sentence of the form “if you must do this, then…”. A partial procedure is more dangerous than none, because it reads as permission and omits the controls.

The prohibition covers the tone as well as the sections. A Level D page describing a mercury box in the second person, with the steps in order, has given a procedure whatever its heading says.

Why the course teaches them at all

Four reasons, and each of them fails if the process is simply left out.

The chemistry is the same chemistry. Mercury amalgamating with the silver of an exposed daguerreotype plate is physical development; the cyanide fixer is complex formation with a different ligand from thiosulfate; the dichromated colloid is a light-driven crosslink. A reader who understands the safe processes but has never met the dangerous ones has a partial mechanism.

The history is not optional. These are not curiosities at the edge of the subject. The daguerreotype is one of the two processes photography began with, wet-plate collodion is why the second half of the nineteenth century looks the way it does, and cyanide was the ordinary fixer of whole decades of practice.

Identification and conservation depend on it. A print or plate in a collection has to be identified before it can be cared for, and its hazards belong to whoever handles it now.

Assessing a hazard is a skill the course teaches. Part XXVI’s assignment asks a reader to assess a historical process for hazard themselves. That is only possible if they have seen the assessment made, several times, on processes where the answer is unambiguous.

The families this covers, and why each is here

Every figure below is a workplace exposure limit from HSE’s EH40 or a notified GHS classification aggregated on PubChem from the ECHA C&L Inventory. A limit is quoted as a measure of how much control the substance is judged to need, not as a threshold you are invited to work to.

Mercury

Daguerreotype development over heated mercury, and the mercury(II) chloride intensifier that survived into gelatin practice. Mercury(II) chloride’s notified classification is signal word Danger with H300, fatal if swallowed, and H310, fatal in contact with skin, both in 100 per cent of 218 reports, alongside H314, H372 and H410; EH40 gives mercury and its divalent inorganic compounds a long-term limit of 0.02 mg/m³ measured as mercury. What puts the process out of reach is not the salt but the metal: mercury has a vapour pressure at room temperature, the vapour is invisible and odourless, and the historical procedure deliberately warms it.

Cyanide

Potassium cyanide as fixer, as silver solvent and in intensifiers. The notified classification carries H300, H310 and H330 — fatal by every route — each in 99.8 per cent of 575 reports. The decisive fact is not the salt’s own toxicity but its incompatibility: contact with any acid liberates hydrogen cyanide gas at once, and a darkroom is a room with an acid stop bath in it. EH40 gives hydrogen cyanide 0.9 ppm long-term and 4.5 ppm over fifteen minutes, with the Sk notation for skin absorption. The incompatibility matrix carries that pair with the others.

Uranium

Uranyl nitrate as toner and intensifier. The notified classification gives H300 and H330, fatal if swallowed and fatal if inhaled, in 100 per cent of reports, with H272 as an oxidiser — but the sample is 39 reports across 2 notifications, much the smallest behind any classification this course cites, so the unanimity is worth less than the same figure would be elsewhere. A second block in the same record, from the Japanese NITE-CMC scheme, adds suspected genetic defects and possible carcinogenicity. None of that covers the radiological hazard, which no GHS classification describes and which EH40 does not list at all, because it is regulated under an entirely separate regime everywhere the course has looked. A substance with two hazard regimes, only one of which the course can read, is a substance the course does not put in a reader’s hands.

Cadmium

Cadmium bromide in collodion salting mixtures and in some period emulsion formulas. This one is here for a reason worth naming, because its headline classification is mild: the ECHA aggregate is signal word Warning, GHS07 and GHS09, H302, H312, H332 and the two aquatic statements. Read alone, that looks like an ordinary irritant. Read beside EH40, it does not: cadmium and its compounds carry a long-term limit of 0.025 mg/m³ as cadmium with the Carc notation, and the same PubChem record carries NITE-CMC blocks classifying cadmium bromide H340, may cause genetic defects, and H350, may cause cancer. Where the aggregate and the regulator disagree this sharply, the course takes the stricter reading and says why.

Chromium(VI)

Dichromates and chromic acid, in the dichromated colloids and in the bleaches and intensifiers built on them. Potassium dichromate’s notified classification runs to twelve statements including H340, H350 and H360; EH40 sets chromium(VI) compounds at 0.01 mg/m³ as chromium, with Carc, Sen and a biological monitoring guidance value. This family has a page of its own, because the word “chromium” covers a second oxidation state the course does permit: see the chromium policy.

The dangerous solvents

Wet-plate collodion is nitrocellulose dissolved in ether and alcohol, poured by hand onto a plate in a small room, and the governing hazard is fire rather than toxicity. EH40’s entry for diethyl ether — 100 ppm long-term, 200 ppm over fifteen minutes — measures the wrong thing for that purpose, and the course has not found a full hazard record for diethyl ether in its own corpus, so it does not print a flash point or an explosive range it has not read. What it can state is that the process combines a volatile flammable solvent with an open pour, a period fixer that was frequently cyanide, and cadmium salts in the salting mixture, and that the case for Level D does not rest on the gap. The bitumen route to a pigment image has a second solvent problem with a limit that is easy to source: benzene sits in EH40 at 1 ppm, with both Carc and Sk.

Platinum, for a different reason

The course ruled on 4 September 2026 that students perform palladium and never handle platinum. Platinum is not acutely toxic in the way mercury and cyanide are. Potassium tetrachloroplatinate’s notified classification carries H301, but the statement that governs the decision is H334, may cause allergy or asthma symptoms or breathing difficulties if inhaled, which appears in 99.5 per cent of 212 reports. EH40 sets halogeno-platinum compounds — its own definition, coordination compounds in which the platinum atom is directly bonded to halide ions, which is exactly what a platinotype sensitiser is — at 0.002 mg/m³ as platinum, with the Sen notation. That is the tightest limit anywhere in this course — five times tighter than silver’s soluble compounds and five times tighter than chromium(VI), both of which sit at 0.01 mg/m³, and 2,500 times tighter than platinum metal’s own 5 mg/m³, which is the clearest possible statement that the hazard belongs to the compound and not to the element.

!Why an irreversible outcome lands in the same place as an acute one

Level D is not a ranking of how nasty a substance is. It is an answer to one question: can the reader assemble the controls that make this reasonable? For mercury and cyanide the answer is no because a failure can kill in the room. For platinum the answer is no for a different reason. HSE’s guidance on skin and respiratory exposure is explicit that sensitisation may take weeks, months or years to appear, that once a person is sensitised tiny amounts of the allergen trigger a reaction, and that the only remedy then is to prevent further exposure. A control failure whose consequence is permanent cannot be recovered from by tightening the control afterwards, and the course has no way to verify that a reader has the extraction a 0.002 mg/m³ limit implies. Two very different hazards, one classification, because the classification is about controls.

Panchromatic sensitising dyes, for a third reason

The same sitting permitted erythrosin at Level B, in one project, on stated conditions, and kept panchromatic sensitisation — pinacyanol chloride and its relatives — at Level D. The reason is neither acute toxicity nor irreversibility. It is that the corpus holds no hazard record for any of those dyes at all, and that the work demands total darkness while hot gelatin and silver nitrate are being handled, which is an accessibility barrier as well as a hazard-control problem. EH40’s own caution applies exactly here: the absence of a substance from the list does not indicate that it is safe. Where the course cannot read a classification, it does not assume one in either direction, and it does not hand the substance to a reader working blind.

What Level D is not

It is not a claim that nobody may do these things. Conservators, museum laboratories and teaching institutions work with several of them under engineered extraction, written procedures and supervision. Where such an account is published, a Level D page points to it rather than paraphrasing it into something that reads like instructions.

It is not a dare. A page that describes the chemistry precisely enough to be worth reading is a page a determined reader could try to reconstruct from. The course accepts that and does not respond by teaching the chemistry badly; it responds by never publishing the quantities, the sequence or the conditions, and by saying plainly on every such page what is missing and why.

It is not the same as Level C. Level C is a real level with real pages: a fume cupboard, controlled waste and a second competent person make the procedure acceptable, and the course will give the procedure to a reader who has those. Level D is the level at which the course gives no procedure to anyone, because the controls that would be needed are not ones a page can assume.

It is not a permanent verdict on a substance. A level is assigned to a procedure. Sodium hydroxide is a Level A ingredient in a developer and a Level B one as a concentrated solution; by the same logic a substance that appears in a Level D process may appear elsewhere, in a different quantity and a different operation, at a level the course does teach.

How to cite this page

A lesson that classifies itself Level D states the level, names which criteria of the rubric applied, and links here rather than restating the policy in its own words. A lesson that mentions one of these families in passing — a history page reaching the daguerreotype, a toning page naming the toners the course studies and does not use — links here for the reason and keeps its own argument about the photography.