Superadditivity: Why Two Agents Beat One

Kodak had a metol developer that worked — D-23, 7.5 g of metol and nothing but sulfite — and a hydroquinone developer that worked, D-9, 11 g of hydroquinone in a caustic bath. It nevertheless built its standard film developer out of 2 g of metol and 5 g of hydroquinone, which is less agent in total than either single-agent formula and a great deal more developer than the arithmetic promises.

That is superadditivity, and it is the reason nearly every developer you can buy contains two developing agents rather than one. This page is about what the pairing buys, what the standard explanation for it is, and why this course marks that explanation unverified while teaching everything around it as established.

The definition, and the measurement nobody in this corpus made

Superadditivity has a precise operational definition. Take a developer containing agent A at some concentration, measure the rate at which it builds density. Take the same developer with agent B instead, at some concentration, and measure again. Now make one containing both, at those same two concentrations, and measure a third time. If the third rate is greater than the sum of the first two, the pair is superadditive. If it merely equals the sum, the agents are simply doing their separate jobs in the same tank.

rate(A + B) > rate(A) + rate(B)

The test, in one line

Three things about that definition are worth pausing on. It is a statement about rates, so it cannot be tested with a single development time. It requires the concentrations to be held constant across all three baths, which means the mixture contains as much A as the A-only bath did, not half as much. And it is a claim about a pair, so it says nothing until both members have been measured alone.

No source in this course’s corpus reports that measurement. Not Kodak’s primers, not a manufacturer’s data sheet, not the 1907 monograph, not the patents read for this part. The literature that would supply rate curves — Mees and James’s Theory of the Photographic Process and Haist’s Modern Photographic Processing — is in the course’s bibliography as cite-only and has not been read. This page therefore has a phenomenon with excellent circumstantial evidence and no published rate curve behind it, and it says so before it says anything else.

!What the corpus does establish, and what it does not

Established, from Kodak’s 1928 primer. Hydroquinone “has so low a potential that it is rarely used alone but is generally used with Elon”. An Elon image “comes up very quickly and gains density slowly”, while a hydroquinone image “comes up very slowly but gains density steadily and rapidly”. The two agents therefore behave differently in a way that is visible in a tray, and a manufacturer chose to combine them.

Established, from the formulary. Every general-purpose film developer, X-ray developer and paper developer in Kodak Limited’s 1949 handbook carries two agents. The single-agent formulas are the special cases: D-23 and DK-15 for fine grain and the tropics, D-9 for maximum energy.

Established, from Ilford and Kodak today. MICROPHEN and BROMOPHEN are phenidone–hydroquinone developers; MULTIGRADE and PQ UNIVERSAL are dimezone-s–hydroquinone developers; XTOL is an ascorbate developer that Kodak states contains no hydroquinone. Four current products, four pairings, no single-agent formula among them.

Not established here. Any rate measurement showing that a pair exceeds the sum of its parts, and any mechanism explaining why it should.

What practice assumed before it knew, and why the assumption is the interesting part

Wall’s 1924 formulary contains the additive assumption in its purest form, in the arithmetic of the Watkins factor — the number by which you multiply the time an image first appears to get the total development time. For a combined developer with the reducing agents in equal ratios, Wall instructs, the factor is the mean of the two; in unequal ratios, weight each factor by its number of parts and divide by the total.

That is the additivity hypothesis written as a working rule and used by photographers for decades. The factors themselves come from Wall’s 1912 dictionary: metol 30, metol–hydroquinone 14, glycin with sodium carbonate 8, hydroquinone with minimum bromide 5.

Set the observed 14 against the mean of 30 and 5, which is 17.5, and it is tempting to call the difference a measurement of superadditivity. It is not, and the reason is worth understanding. The Watkins factor is a ratio of two times, not a rate: it says how late in its own development an image appears, which is a statement about the shape of the density curve rather than about how fast anything happens. A pair could be twice as fast as the sum of its parts and have exactly the mean factor. What the numbers do show is that the mixed developer behaves like neither parent — closer to hydroquinone’s late-appearing character than a simple average — and that is a qualitative observation, offered as one.

The standard explanation, and the honest label on it

The account given throughout the photographic literature is regeneration. The two agents are not peers; they have different jobs.

The regeneration cycle, with the species this course cannot name marked as such

1. 1Primary agent, adsorbed at the speck — metol or Phenidone, present in small quantity; hands electrons to silver at the surface. THE ADSORPTION IS THE STANDARD ACCOUNT, NOT A SOURCED FACT
2. 2Oxidised primary agent — drawn dashed because no source this course holds names metol's oxidation product; the quinone-imine of the wider literature is not asserted here
3. 3Hydroquinone regenerates it, in solution — the secondary agent supplies the electrons and is oxidised to quinone. NOT ESTABLISHED BY ANY SOURCE THIS COURSE HAS READ
4. 4Sulfite meets the quinone, two ways — back to hydroquinone, or on to the monosulfonate. Kodak states both and reconciles neither; this half IS sourced
5. 5Silver, and the halide leaving — one bromide ion into the bath for every silver atom made — the part of the picture nothing disputes

Drawn to show an argument, not a measurement. Two of the five stations carry the course's honest warning: the adsorption of the primary agent and its regeneration by the secondary are the account given throughout the photographic literature, and no Tier-1 source in this course's corpus states either. They are drawn because a reader must be able to see what is being claimed in order to judge it.

In that account, the primary agent is the one that touches the crystal. It is present in small quantity, it adsorbs at the development centre, and it hands its electrons to silver at the surface. The secondary agent never develops anything; it sits in solution in bulk and does one job, which is to reduce the oxidised primary agent back to its working form. A little of the primary agent therefore goes round the cycle many times, and the developer’s total capacity is set by the large reservoir of the secondary rather than by the small quantity of the primary.

If that is right, four things follow that are all observed. The primary agent can be used at a small fraction of the secondary’s concentration. The pair develops faster than either alone at those concentrations. The developer’s capacity tracks the secondary agent. And the character of the image — metol’s quick appearance, hydroquinone’s steady climb to density — is a blend rather than an average.

⁇Why this page marks the mechanism rather than teaching it

The disagreement is not between sources. It is between the literature and this course’s corpus.

The regeneration account is, as far as this course can tell, the consensus of photographic chemistry. It appears in every secondary description of superadditivity the course has encountered, and it is attributed to work published in the 1950s.

No Tier-1 source held by this course states it. Kodak’s 1928 primer predates it and offers nothing resembling it. The Kendall patent claims a substitution ratio and no mechanism. The manufacturers’ current sheets name the agent pairs and stop. The 1907 monograph is thirty years too early. The two books that would settle it are in the bibliography as cite-only, meaning the course records them as the literature to read and has not read them.

What the course does with that. It teaches the phenomenon as established, because four manufacturers’ product ranges are evidence enough that pairing works. It draws the mechanism, labelled, because a reader cannot evaluate an account they have not been shown. And it refuses to write the regeneration reaction as a balanced equation, for a specific reason that is worth more than the equation would be: the course cannot name metol’s oxidation product. The classical agents page establishes what it is not — not deposited in a coloured form with the silver — and stops. An equation with an unnamed species on both sides is not chemistry, it is a diagram with an equals sign in it.

The related species the literature also leans on, the semiquinone, is marked the same way on the classical agents page and for the same reason.

The one regeneration this course can write down

There is a regeneration reaction in the corpus, and it is worth putting beside the unverified one because it shows that the idea is not exotic. Kodak’s 1928 primer states it plainly: if sulfite is added to quinone, the quinone oxidises the sulfite to sulfate and is itself reduced again to hydroquinone.

C₆H₄O₂ + SO₃²⁻ + H₂O → C₆H₆O₂ + SO₄²⁻

A developing agent restored by a cheaper reductant — the same relationship the regeneration account proposes for the agent pair, and the only one this course can source

The sulfite page uses that reaction to explain preservation. It does a second job here. A solution-phase reductant restoring an oxidised developing agent is a documented reaction in this exact chemistry, so proposing that hydroquinone does for oxidised metol what sulfite does for quinone is not a leap. It is still a proposal.

MQ against PQ, in the ratios the formulas actually use

The molar arithmetic below uses the molar masses from the encyclopaedia: metol as the hemisulfate 344.39, of which each formula unit supplies two developing molecules; hydroquinone 110.11; phenidone 162.19.

Developer	Primary agent	Secondary	Molar ratio, secondary to primary
DK-50	metol 2.5 g/L	hydroquinone 2.5 g/L	1.6 : 1
D-76	metol 2.0 g/L	hydroquinone 5.0 g/L	3.9 : 1
D-72	metol 3.1 g/L	hydroquinone 12.0 g/L	6.1 : 1
D-19b	metol 2.2 g/L	hydroquinone 8.8 g/L	6.3 : 1
Kendall’s Example II	phenidone 0.3 g/L	hydroquinone 8.0 g/L	39 : 1

Read down the middle column and the film developers, the paper developer and the X-ray developer are all built the same way: a nearly constant small quantity of the primary agent — between 2.0 and 3.1 g per litre across every one of them — and a secondary agent varied from 2.5 to 12 g/L to set the energy and the contrast. The primary agent is a catalyst-shaped ingredient and the secondary is a fuel-shaped one. That is the formulary’s own statement of the regeneration account, made without stating it.

Then the pyrazolidone arrives and the small quantity gets much smaller. John David Kendall’s 1941 patent for Ilford declares among its objects “to provide a substitute for metol”, and claims that replacing the metol in a metol–hydroquinone developer with about one fifth to one sixth of its weight of 1-phenyl-3-pyrazolidone gives a developer of similar development characteristics. On D-76’s 2 g of metol that is 0.33 to 0.40 g of phenidone; in molar terms, because phenidone is the heavier molecule, it is about an eighth to a seventh of the metol on a mole-for-mole basis. The figure usually quoted in conversation is “a tenth”, which is the right order and not the sourced number.

🧮The comparison the patent prints, and what it is not

Kendall’s Example I uses the agent alone: 3 g in 500 cm³, which is 6 g/L, with 1 g/L of potassium bromide. His Example II uses it as a metol substitute: 0.15 g in 500 cm³, which is 0.3 g/L, against 8 g/L of hydroquinone and twice the bromide.

Twenty times the concentration when it works alone as when it works alongside hydroquinone. That number is the observation superadditivity has to explain, and the classical agents page states the limit on it that this page repeats: the patent does not comment on the difference, the two examples were not run as a comparison, and the bromide differs between them. It is two examples, not a study. What it is evidence of is that a formulating chemist in 1941, working for a company that had to sell the result, thought a twentyfold reduction was available from the pairing.

Ascorbate and Phenidone, and the pair that has no quinol in it

Kodak’s XTOL is the modern member of the family and it changes one half of the pair. Kodak’s own technical sheets describe it as an ascorbic acid developer giving very high image quality at full emulsion speed, with stable performance across a range of temperatures, dilutions and agitation methods — and state that it contains no hydroquinone. The working tank solution is specified at pH 8.20 ± 0.05 with a specific gravity of 1.085 ± 0.003 at 25 °C.

What Kodak does not publish is the second agent, and this course does not name it. Ilford’s parallel products disclose a class — dimezone-s — and nothing further, and the classical agents page records that the course researched neither substance and states nothing about either. The honest description of XTOL’s chemistry is therefore: an ascorbate as the bulk reductant, paired with an undisclosed agent, at a pH more than a unit and a half below a paper developer’s.

Two consequences follow from the ascorbate half alone, and both are established on the alternative agents page. An ascorbate developer has no sulfonation route: there is no quinone for sulfite to add to, so the tell-tale that Kodak warns about for MQ developers — a colourless solution that is nevertheless exhausted — does not apply in the same way. And ascorbate developers are sensitive to trace iron in a way that quinol developers are not.

The limits: sulfite, and the pH window

Two limits on the pairing are visible in the corpus, and they are of different kinds.

Sulfite competes for the product, and Kodak gives two incompatible answers. In a regeneration cycle the hydroquinone pool is the reservoir, and what happens to the quinone decides how long it lasts. Route one — the reaction printed above — returns the hydroquinone and spends sulfite. Route two, from the same book’s chapter on developer life, converts the quinone to hydroquinone mono- and disodium sulfonates, which are colourless and are not hydroquinone. On route one the reservoir is topped up; on route two it is drained while the bottle stays clear. Kodak states both and reconciles neither, and this course does not reconcile them either. Whether the sulfonate is itself a developing agent — which would change the answer entirely — is unsourced, as the sulfite page records.

The pH window is set by the secondary agent, and it is narrow. Hydroquinone’s first pK_a is 9.85 to 9.91, so at a paper developer’s 10.4 about three quarters of it is ionised, at ID-11’s 8.65 about six per cent, and at PERCEPTOL’s 7.75 well under one per cent. If the secondary agent is the reservoir, a bath that leaves it unionised has no reservoir worth the name. Metol, whose amine is neutral above about pH 8, does not care.

That is one reading of why the family of developers splits where it does: the MQ and PQ formulas cluster between pH 8.6 and 10.6, where hydroquinone can be reached; the low-pH developers either drop the hydroquinone entirely — as XTOL does at 8.20 — or accept being slow. It is a reading, offered as one: this course has not sourced a statement that the ionised hydroquinone is the species doing the regenerating, and the classical agents page records the same gap for development itself.

What the pairing buys, itemised honestly

Contrast and shadow speed together. This is established and it is the strongest claim on the page. Kodak’s primer gives metol a quick appearance and slow density gain, hydroquinone the reverse; the formulary pairs them at every energy from DK-50’s 1.6 : 1 to D-19b’s 6.3 : 1, and the ratio is the contrast control.

Tolerance of cold and of bromide. Also established, and it follows from the same primer’s ranking: a small temperature change affects hydroquinone greatly and metol very little, and a small quantity of bromide affects hydroquinone and does not affect metol nearly so much. A pair carries a component that keeps working when conditions turn against the other one.

Capacity out of proportion to the primary agent. This is the regeneration account’s own prediction and the course has no capacity measurement isolating it. What can be said is that ILFORD rates ID-11 and MICROPHEN identically at 10 films per litre reused, and PERCEPTOL at 4 — and that ILFORD publishes none of the three formulas, so the difference cannot be attributed to the agent pairing rather than to everything else that differs.

⚠The claim about PQ and sensitisation, and what the sheets actually support

It is often said that phenidone developers were adopted partly because they are kinder to the skin than metol ones. This course looked for that claim in ILFORD’s literature and did not find it. The powder developer sheet gives the same generic advice for all three products — gloves, eye protection and an apron or overall when handling and mixing — and makes no comparative statement about any agent. The same sheet also carries a blanket reassurance about photographic chemicals that this course does not repeat, because it is not tied to a concentration, a quantity or a source.

What the hazard record supports is narrower and still worth acting on. Metol carries a harmonised classification under Regulation (EC) No 1272/2008 with H317, may cause an allergic skin reaction, and HSE’s HSG262 makes the point that decides how to treat it: sensitisation may take weeks, months or years to appear, and once it has, tiny amounts provoke the reaction and preventing further exposure is the only remedy left. Phenidone’s classification rests on a far thinner notification base, as its encyclopaedia page records.

Those are not the same statement. Replacing a substance with a harmonised sensitisation classification by one whose hazard data are thin removes a known hazard and substitutes an incompletely characterised one. That may well be the right trade — it is the trade the whole industry made — but “less well documented” is not “safer”, and the course will not write it as though it were. Both agents are powders, both are weighed, and the control for both is the same: no dust, gloves, eye protection, and the classification rubric for the rest.

Recap

• Superadditivity is a claim about rates: the pair exceeds the sum of the two agents measured separately at the same concentrations. No source in this course’s corpus reports that measurement.
• The phenomenon is nevertheless well evidenced by what manufacturers build: every general-purpose developer in the formulary and every current product named in this part carries two agents.
• The standard explanation is regeneration — a small quantity of primary agent working at the crystal surface, restored again and again by a bulk secondary agent in solution. It is drawn here and marked unverified, and the course will not write it as an equation because it cannot name metol’s oxidation product.
• One regeneration in this chemistry is sourced: sulfite reducing quinone back to hydroquinone. It makes the proposal plausible and does not make it established.
• The ratios say the same thing the account does: the primary agent stays near 2 to 3 g/L across every formula while the secondary varies fivefold to set the energy.
• Phenidone moved the primary agent’s concentration down by roughly a factor of five to six by weight against metol, on Kendall’s published substitution ratio.
• The limits are the secondary agent’s: sulfite decides what becomes of the quinone by two routes Kodak never reconciles, and hydroquinone’s pK_a of 9.88 puts a floor under the pH at which a quinol pair is worth having.

Check your understanding

Question 1. A reader claims that D-76 proves superadditivity, because 2 g of metol plus 5 g of hydroquinone develops faster than 7 g of either alone. What is wrong with the argument as a test?

• Nothing: that is exactly the definition
• The definition requires the same concentration of each agent in the mixture as in the single-agent baths — 2 g of metol alone and 5 g of hydroquinone alone — and it requires rates rather than a single development time
• D-76 also contains borax, which invalidates any comparison
• Hydroquinone cannot develop at all without metol

Show the answer and why

Answer: The definition requires the same concentration of each agent in the mixture as in the single-agent baths — 2 g of metol alone and 5 g of hydroquinone alone — and it requires rates rather than a single development time

The test is rate(A+B) > rate(A) + rate(B) with the concentrations held fixed, so the comparison baths are 2 g of metol alone and 5 g of hydroquinone alone, not 7 g of either. Substituting 7 g changes the very quantity being tested. It also has to be a rate: development time to a stated density is one point on a curve, and superadditivity is a statement about the whole growth. This course reports no source that made that measurement, and says so rather than borrowing the conclusion.

Question 2. Across DK-50, D-76, D-72 and D-19b the metol stays between 2.0 and 3.1 g per litre while the hydroquinone runs from 2.5 to 12 g per litre. What does that pattern suggest, and what is it not?

• That metol is expensive and hydroquinone is cheap
• That the primary agent behaves like a catalyst whose quantity barely varies while the secondary behaves like a fuel whose quantity sets the energy — which is what the regeneration account predicts, and is consistent with it rather than proof of it
• That hydroquinone is the only agent that develops
• That the two agents are interchangeable at any ratio

Show the answer and why

Answer: That the primary agent behaves like a catalyst whose quantity barely varies while the secondary behaves like a fuel whose quantity sets the energy — which is what the regeneration account predicts, and is consistent with it rather than proof of it

It is the formulary making the regeneration account visible without stating it. If a little primary agent is reused many times while the secondary is consumed supplying electrons, then the primary should be nearly constant and the secondary should be the knob a formulator turns — which is exactly the pattern from a general-purpose developer at 1.6 to 1 up to an X-ray developer at 6.3 to 1. Consistency is not proof: the same pattern would follow from hydroquinone simply being the contrast agent, which Kodak’s primer also says. The course reports both readings and elevates neither.

Question 3. Why does this page refuse to print a balanced equation for hydroquinone regenerating oxidised metol?

• Because the reaction is too fast to write down
• Because no source the course holds names metol’s oxidation product, so one species in the equation would be invented
• Because it happens on a surface and surface reactions cannot be balanced
• Because sulfite interferes with it

Show the answer and why

Answer: Because no source the course holds names metol’s oxidation product, so one species in the equation would be invented

Rule 1 of this course forbids inventing chemistry, and an equation is a set of specific claims about specific species. Kodak’s primer establishes what metol’s oxidation product is not — it is not deposited in a coloured form with the silver, which is why a sulfite-rich pyro image comes out almost as blue as an Elon one — and nothing else. The quinone-imine of the wider literature is not asserted, and the semiquinone the same literature leans on is marked contested on the classical agents page for the same reason. What can be written is the reaction the course can source, sulfite reducing quinone back to hydroquinone, which shows the shape of the idea without inventing a participant.

Question 4. Kodak states that XTOL contains no hydroquinone and gives its working solution as pH 8.20. Using hydroquinone’s pKa of about 9.88, why is that combination coherent?

• Because ascorbate needs a lower pH than hydroquinone
• Because at pH 8.2 only about two per cent of any hydroquinone present would be ionised, so a quinol reservoir would be largely idle; an ascorbate that works at that pH is the way to build a low-pH developer at full speed
• Because hydroquinone precipitates below pH 9
• Because ascorbate and hydroquinone react with each other

Show the answer and why

Answer: Because at pH 8.2 only about two per cent of any hydroquinone present would be ionised, so a quinol reservoir would be largely idle; an ascorbate that works at that pH is the way to build a low-pH developer at full speed

The Henderson-Hasselbalch fraction with pKa 9.88 puts hydroquinone at about 2 per cent ionised at pH 8.2, 6 per cent at ID-11’s 8.65 and about three quarters at a paper developer’s 10.4. If the ionised form is what does the work — the obvious reading, and one the course has not sourced — then a bath at 8.2 gets very little out of a quinol. Kodak’s answer is to drop the quinol and use an ascorbate, which is why the sheet advertises full emulsion speed and very high image quality at a pH where an MQ developer would be sluggish. What the course does not know is the identity of XTOL’s second agent, and it does not guess.

Question 5. A photographer switches from an MQ to a PQ developer and describes it as "the safer option". What is the accurate version of that statement?

• It is correct: phenidone is non-hazardous
• Metol carries a harmonised classification as a skin sensitiser, while phenidone’s classification rests on a thin notification base — so the switch replaces a known hazard with a less well characterised one, which is not the same as replacing it with a smaller one
• It is wrong: phenidone is the more hazardous of the two
• It depends only on the concentration used

Show the answer and why

Answer: Metol carries a harmonised classification as a skin sensitiser, while phenidone’s classification rests on a thin notification base — so the switch replaces a known hazard with a less well characterised one, which is not the same as replacing it with a smaller one

Metol’s H317 is harmonised under the CLP Regulation, and HSG262 explains why sensitisation deserves particular respect: it may appear after weeks, months or years, and afterwards tiny exposures provoke it and avoidance is the only remedy. Phenidone’s entry rests on far fewer notifications, which means thinner evidence rather than demonstrated safety. The course looked for a comparative claim in ILFORD’s own literature and found none — the sheet gives identical generic advice for all three of its powder developers. Both are powders and the control is identical for both: weigh without dust, gloves and eye protection, and keep the jar closed.

Question 6. Wall’s 1924 rule says the Watkins factor of a combined developer with equal parts is the mean of the two factors. Metol is 30, hydroquinone with minimum bromide is 5, and metol–hydroquinone is listed at 14 rather than 17.5. Why is that not a measurement of superadditivity?

• Because the factors were measured on different plates
• Because the Watkins factor is a ratio of two times and describes when the image appears relative to completion, not how fast development runs — so it cannot test a claim about rates
• Because 14 and 17.5 are within experimental error
• Because superadditivity applies only to phenidone developers

Show the answer and why

Answer: Because the Watkins factor is a ratio of two times and describes when the image appears relative to completion, not how fast development runs — so it cannot test a claim about rates

The factor is total development time divided by the time the image first appears. A developer twice as fast overall, with the appearance and completion times both halved, has exactly the same factor. So the discrepancy tells you the mixed developer has a different curve shape from the weighted average of its parents — closer to hydroquinone’s late-appearing character — which is a real and interesting observation, but it is not the rate comparison the definition requires. It is a good example of a number that looks like evidence for a claim it cannot actually address.

Check my answersClear