Experiment: Agitation, Stand and Semi-Stand Development

Everyone who develops film has an opinion about agitation, and almost nobody has a number. This session produces four numbers per regime and one more that nothing earlier in the course has measured at all: how even the development was, read off the film at twenty stated positions rather than judged by holding the negative up to a window. It also does something the enthusiasm around stand development rarely does — it says, before any film is spent, exactly what the published sources contain and what they do not.

Purpose

To develop matched strips under three agitation regimes at one dilution, time and temperature; to read contrast index, the relative speed point, maximum density and a compensation ratio off each; to measure each strip’s evenness as a density profile along its length; and to repeat the comparison on continuous-tone roll film, where the failure modes live.

Hypothesis. That reducing agitation lowers the upper-scale gradient more than the lower-scale one — so the compensation ratio defined below falls — while leaving the relative speed point inside the resolution of the apparatus, and that it raises the standard deviation of the density profile above the control’s floor by a factor the instrument can report. It fails if the compensation ratio does not move by more than the resolution sentence allows; if the speed point moves further than the gradient change accounts for; or if the least-agitated cell is as even as the control, which would mean either that this vessel is too small for the effect or that the effect is smaller than the practice claims.

Second hypothesis, the floor arm. That the same regime at a dilution putting the developing agent per unit film area below the manufacturer’s stated minimum gives a curve that looks compensating and is in fact starved. It fails if that cell’s upper scale is not depressed relative to the same regime run above the floor.

The control. The manufacturer’s specified agitation scheme at the manufacturer’s published time: ILFORD’s intermittent scheme, four inversions in the first 10 seconds and four again in the first 10 seconds of each further minute, at 16 minutes, 20 °C, Kodak D-76 diluted 1+3 — ILFORD’s published time for FP4 Plus at EI 125 in that developer at that dilution. It is run in both vessels this session uses, which is what ties the two arms together.

The one variable. The agitation regime: same emulsion batch, lamp, wedge, developer bottle, dilution, volume per unit area, vessel, temperature, time, fixer, wash and reader. Where a second variable has to move — in the floor arm and in the long-time cell — it is named in the cell’s own name, and the analysis never attributes those cells to agitation alone.

!What this page is not

It is not Part IX’s agitation arm run again: that session ranked three scripts by eye and archived the strips, while this one measures them, adds a defined evenness figure, adds a cell designed to cross the developing-agent floor, and adds roll film in a tank, because a strip in a cylinder cannot produce a sprocket-hole streak.

It is also not a test of sharpness. The measurement that would settle the acutance claim is a microdensitometer trace across a knife edge, and the Preparation says why the Part XV instrument cannot make one.

Learning objectives

By the end of the session and the reading that follows it you will be able to:

• state the transport mechanism agitation controls, and derive from it why the same local bromide accumulation is an adjacency effect at one scale and bromide drag at another;
• define compensation as a ratio of two average gradients before you have any data, and compute the smallest difference in it your densitometer can report;
• design a stand arm against the developing-agent floor rather than against a dilution ratio, and convert any volume, dilution and film area into millilitres of stock per 80 square inches;
• measure evenness as a density profile, with the control cell’s own profile as the floor, instead of describing it;
• say which claims about stand development your apparatus can settle, which it cannot, and which have no Tier 1 or Tier 2 evidence behind them at all.

Prerequisites

Designing an experiment that yields a number owns the method, and nothing here re-derives it. The development time and temperature series must come first, because its contrast-index-against-time curve turns a contrast difference into an equivalent time — the only fair currency for regimes that develop at different rates — and four curve families, measured carries the uncertainty budget this page inherits. The chemistry is in development kinetics, acutance, adjacency effects and compensating development and diffusion and osmosis; the two constructions in gamma, contrast index and average gradient and film speed and exposure index.

The hard prerequisites are the two certificates: the sensitometer’s and the densitometer’s. Without a repeatability figure for the densitometer there is no evenness measurement at all, because the quantity is a spread of readings and you could not say how much of it was the instrument.

Safety classification

Level A. Standard home-darkroom controls, on a session that opens no jar of powder.

Every solution is diluted from something already made: the D-76 stock from Part VIII’s mixing lab, the fixer and wetting agent from bought concentrates. Under the course’s rubric that means gloves, eye protection whenever a solution is poured or agitated, dedicated labelled vessels and the ordinary ventilation of a wet bench — the controls Part IX’s test-negative lab set for the same operation.

One thing here is not ordinary. Two cells stand unattended for an hour in open cylinders, in a room dark for part of it, which is how alkaline solution gets knocked over or topped up by somebody else. Label both cylinders on the vessel before filling, stand them in a tray that holds their whole contents, and put them where nothing else has to reach across them.

What is not a hazard here, and why. No powder is weighed, so the inhalation route that makes metol and hydroquinone a Level B proposition in the mixing lab is not open in this session: both are present, but in solution at a working dilution, so the control is contact rather than airborne. Nothing is heated. No concentrated alkali is handled, because D-76’s alkali is borax already in solution. A fixed and washed strip presents no chemical hazard of its own: its silver is metallic and held in hardened gelatin, and the soluble silver–thiosulfate complexes leave in the wash. None of those absences is a statement about the substances in general — each is about this operation, at these concentrations and quantities, which is what a hazard assessment is.

Hazards

Hazard	Where it arises	Control
Skin sensitisation from metol and hydroquinone	Eleven cylinders and three tanks poured, agitated and emptied over three hours	Single-use nitrile gloves throughout, changed if one is splashed inside. The glove page has the compatibility argument
Splashes to the eye, and working in darkness beside open vessels	Lifting strips once a minute with the light off	Eye protection from the first pour to the last emptied vessel; cylinders filled with headroom; the bench laid out in the light and not moved; one hand wet and one dry
Alkaline solution standing unattended for an hour	The two floor-arm cells	Labelled on the vessel before filling, in a tray that holds the whole volume, outside every working reach, never topped up
Spent fixer, acidic and silver-bearing	Emptying the fixing tray after fourteen pieces of film	Its own labelled container to the silver stream, never combined with developer waste
Loss of an exposure batch that cannot be replaced	Eleven strips and one roll exposed in single sittings	The tin stays closed except during loading; strips notched by touch before any light goes on

Required PPE

Single-use nitrile gloves for every wet operation, donned and removed per the PPE SOP, because metol and hydroquinone are skin sensitisers and a sensitisation, once acquired, does not go away. Eye protection from the first pour to the last emptied vessel. And clean dry hands for the film, which protects the film rather than you: this page measures density differences of a few hundredths, and a fingerprint on a uniform strip is a difference of exactly that size in exactly the place you will read.

Respiratory protection is not among the controls, for a specific rather than a reassuring reason: nothing here is a powder, nothing is heated, and no step produces a vapour or an aerosol. The particulate control belongs to the mixing lab.

Ventilation

Ventilation is not the control this page relies on, because nothing here evaporates or aerosolises at 20 °C: the developer is a dilute aqueous solution of an aminophenol salt, a phenol, a sulfite and a borate, and the fixer a thiosulfate at working strength. The room needs ordinary comfort ventilation for three hours, part of it in the dark, which Part XVI’s darkroom build provides through a light-tight vent. If the fixer is an acidic rapid type and the room is small, ventilate as the fixer mixing SOP sets out.

Materials

Item	Quantity	Note
Test strips from one exposure batch	11	135 mm, one emulsion batch: seven wedge, four uniform
A 36-exposure roll of the same film, exposed on an even-toned subject	1	Cut into three 490 mm lengths in the changing bag
Opaque batch tin	1	Opened only in darkness or inside the bag
Transmission step wedge	1	The Stouffer T2115 or equivalent, the same one throughout
Card label slips	14	The eight-field slip from the test-negative lab. A strip without one is not data
Measuring cylinders, 100 mL, dedicated to developer	7	One per cylinder cell, standing in the bath
One-reel 35 mm tanks with spirals	3, or 1 used three times	Three puts all the roll lengths in one half hour
Water bath, and an overflow tray for the standing cells	1 each	The bath deep enough for seven cylinders and a tank; the tray big enough for two cylinders’ contents
Film clips and a drying line	14 clips	Still, dust-free air. Do not squeegee anything here
Negative sleeves	8 pockets	Strips in pairs, roll lengths singly
Curve-plotting worksheet, lab notebook sheets and the formula version record	7 + 1 set + 1	One plotting sheet per wedge strip; the version record names the bottle

Chemicals

Nothing is weighed. Every solution is diluted from a stock or a concentrate, per the SOP for mixing from a stock.

Chemical	Quantity	Form
D-76 stock from Part VIII’s lab — metol, hydroquinone, sodium sulfite, borax	400 mL	Diluted 1+3 or 1+19 immediately before use and discarded after one cell, as Kodak’s J-78 sheet directs
Plain water rinse at the cell’s temperature	3 L, two changes per piece of film	ILFORD note that a water rinse may replace a stop bath but increases the risk of processing marks and stains
Rapid fixer, ammonium thiosulfate type	750 mL at 1+4	Twice the clearing time, per the clearing-time SOP
Wetting agent	750 mL at 1+200	ILFOTOL at 5 mL per litre; ILFORD warn that too little or too much leads to uneven drying, which here would be read as a result

The silver bromide in the emulsion is a reagent rather than a substance you handle, and the bromide it releases is the subject of the experiment: made inside the layer, leaving by diffusion, at a rate agitation decides. p-Aminophenol appears only in Further experiments.

Equipment

The sensitometer from Part XIV with its certificate and uniformity map, used per the exposure-session SOP and, for the uniform strips, with the wedge lifted off and a short exposure — a mode the build supports and the calibration page did not test. The densitometer from Part XV with its certificate, 2.0 mm aperture and file logging, checked per its SOP. A camera, a changing bag, one thermometer checked per the bench-check SOP, a stopclock, a millimetre rule for the profile positions, and a spreadsheet. None of it is consumed, which is why none of it appears in the table below.

Estimated cost

££ on the planner’s bands, almost all of it capital bought earlier: the two instruments, the wedge, the cylinders, the tanks, the camera and the thermometer. This session spends two cassettes of film, about 400 mL of D-76 stock and about 150 mL of fixer concentrate.

Estimated consumables cost

One run, at eleven strips, one roll and fourteen pieces of film through the fixer:

Consumed	This session	Sourced price	Cost this session
35 mm film for the strips	one cassette, cut into eleven 135 mm strips	£6.37–£11.40 per 36-exposure roll	£6.37–£11.40
35 mm film for the continuous-tone arm	one cassette, exposed whole	£6.37–£11.40 per 36-exposure roll	£6.37–£11.40
D-76 stock, from Part VIII	about 400 mL, diluted and discarded cell by cell	Costed in Part VIII’s mixing lab	—
Rapid fixer concentrate	150 mL, to make 750 mL at 1+4	£21.05–£25.98 per 1 L, diluted 1+4 for film	£3.16–£3.90
Wetting agent	3.75 mL, for 750 mL at 1+200	£28.70 per 1 L, diluted 1+200	£0.11
Water for rinses and washes	about 12 L	Metered supply; the planner prices no water	—
Negative sleeves and card label slips	8 pockets, 14 slips	None. A named price gap: sleeving that passes the Photographic Activity Test	—
Squared paper for the plots and profiles	8 to 10 sheets	None. The planner carries no stationery line	—

The priced rows come to £16.01 to £26.81, at the retail ranges read on 5 September 2026 and recorded in the laboratory planner. That is a floor rather than a total: four of the eight rows carry no dated price, so they count as nothing here and are certainly not free, and a priced entry is a dated range to plan against rather than a quotation. Both film rows price a cassette because that is what the price file carries a figure for; a bulk tin is far cheaper per strip and the file has no bulk-tin price, so the cheaper route is the one this table cannot cost.

Waste streams

Three, kept apart, labelled per the container SOP and routed per the general waste SOP.

1. Spent developer, about 1.75 L of D-76 at 1+3 and 1+19, alkaline, carrying metol, hydroquinone, sulfite, borate and the bromide released by fourteen pieces of film. Kodak’s instruction is that a diluted solution is discarded after its batch, so all of it is waste.
2. Spent fixer and the first rinse after it, silver-bearing, to the silver stream.
3. Rinse and wash water: the two rinses after each development go with stream 1, the first change after fixing with stream 2, the rest with the general stream.

Keep streams 1 and 2 apart: an alkaline sulfite solution poured into an acidic silver-bearing one spoils the recovery of the silver and wastes both containers.

Alternative route

Darkness is needed for two operations: taking strips from the tin and notching them, and the cylinder cells themselves, because strips are lifted from open vessels. The camera roll, the tank cells, fixing, washing, drying, reading and all the arithmetic happen in room light or inside a changing bag.

Route one: the all-tank version, which removes the darkroom entirely. Load every piece of film onto spirals in a changing bag and run every cell in a closed tank. Two things improve — the agitation becomes inversion, the form every manufacturer publishes, so the control stops being a translation, and a closed tank holds temperature better. Three get worse: one cell per tank, so eleven strips become five or six sequential runs across two evenings; the volumes rise, because a one-reel tank needs about 290 mL where a cylinder needs 80, which moves every cell on the agent-floor plot below, so recompute it rather than copying these numbers; and the developer temperature can only be read going in and coming out, so log both and treat the difference as the term the cylinder would have measured.

Route two: ILFORD ORTHO Plus, whose notching and cylinders run under a deep red safelight — though the control’s published time must then be ORTHO Plus’s own. Route three: a windowless room at night, proved rather than assumed with the blackout leak test.

If none is available, the cylinder arm cannot be run — but route one is a complete substitute needing only a changing bag, a tap and patience, so this page has no dead end, only a slower route with a different vessel. The report must name which vessel produced the numbers, because the vessel is part of the result.

Preparation

About 75 minutes the evening before, plus two exposure sessions. The 180 minutes here is the wet run; the film cannot be read until bone dry, so the densitometer session is a third sitting of about 90 minutes. Every definition below is fixed before a number exists, because a decision made afterwards is a preference.

What the sources actually say about stand development

Read this before designing anything, because it decides what the experiment can be.

No manufacturer of a general-purpose film developer in the course’s corpus publishes a stand or minimal-agitation procedure. ILFORD publish an inversion scheme and warn that very short times with some films may lead to uneven processing. Kodak publish one and add that agitation “should always consist of irregular or random movements that will not cause solution currents to flow over the film constantly in any one direction; these currents increase film density along their paths, causing nonuniformity”. Bergger, selling a developer on its definition, require agitation every 15 seconds and list “denser image edges” against inadequate shaking.

Two sheets in the corpus do, and both are for the same single developer. Photographers’ Formulary and Bostick and Sullivan both print Sandy King’s recommendations for Pyrocat-HD: minimal agitation at 10 seconds every three minutes with times about 50 per cent longer, and semi-stand at 1 part A to 1 part B to 200 or 400 parts water, agitated for one minute at the start and 30 seconds at the halfway point, for 40 to 60 minutes. Part VIII’s rule stands: a scheme published for one developer is not evidence about another.

No Tier 1 or Tier 2 source in the corpus publishes measured sensitometric data for stand development, and this is the finding that matters most. King is the practice’s leading specialist advocate and publishes contrast-index charts for a dozen films — stating that they “are based on rotary processing in BTZS type tubes”, which is continuous agitation. Even the advocate’s numbers are not stand numbers. The nearest thing to data is Ed Buffaloe’s account on the Formulary sheet of two rolls of Delta 3200: one printed with “extremely high acutance”, and the other was ruined, because “the bromides released by the intense development in the heavily exposed areas diffused out and caused uneven development in surrounding areas”. An honest report of two rolls is not a measurement, and the sheet does not present it as one.

One genuine measurement touches the practice, and it is a century old and about dilution rather than agitation. Wall reports Wratten and Wainwright’s photometry on rodinal: a plate needing 3 minutes at 1:20 needed 42 minutes at 1:200 in air-free distilled water, 46 in ordinary distilled water and 52 in tap water, against the 30 proportionality predicts. Time does not scale with dilution, the error runs 40 to 70 per cent towards more time, and its size depends on dissolved air — so a stand arm designed from arithmetic comes out under-developed by an amount that depends on your tap.

⁇Whether stand development is a technique or a folklore

Sandy King, who devised Pyrocat-HD, publishes minimal and stand procedures and claims three results — maximum apparent sharpness through adjacency effects, a compensating effect and increased emulsion speed — while calling stand development “fraught with dangers”, naming high-speed films as poor candidates and advising two or more back-up shots of any important scene. Wall in 1924 records the method as already old, allows it fine grain and uniform results “if the developer be frequently agitated”, and calls it uneconomical. ILFORD, Kodak and Bergger publish nothing for it, and the last two put unevenness squarely against inadequate agitation.

The course therefore treats the practice as an untested hypothesis rather than a procedure, writes its claims down before the session so they can fail, and reports the result as one film in one developer on one evening. Nothing here corrects King, who is describing a different developer, and nothing here vindicates the practice either.

The mechanism the whole experiment turns on

Development consumes developing agent at the film surface and gives back bromide.

AgBr(s) + e⁻ → Ag(s) + Br⁻

Chemical development, per formula unit: the silver stays where it was and the bromide goes into the bath

C₆H₆O₂ + 2 AgBr + 2 OH⁻ → C₆H₄O₂ + 2 Ag + 2 Br⁻ + 2 H₂O

The same thing with the electrons accounted for: two silver ions reduced per molecule of hydroquinone, two bromide ions released and two hydroxide ions spent

Both happen inside the gelatin, so everything the bath does has to cross the layer of solution sitting on the emulsion — the diffusion boundary layer. Transport across it is diffusion, and diffusion obeys Fick’s first law.

J = −D (dc/dx)

Fick's first law, the whole of what agitation controls

J is the flux in moles per square metre per second; D the diffusion coefficient, between about 0.6 and 2 × 10⁻⁹ m² s⁻¹ for small ions at room temperature; dc/dx the concentration gradient across the layer. Agitation changes none of the chemistry in the two equations above: it changes x, and therefore the gradient and the flux. Kodak’s primer of 1928 put it in one sentence, that the speed of development depends chiefly on the rate at which the developer diffuses into the film, and Kodak’s modern process control document says the same thing operationally: agitation maintains uniform solution activity by removing exhausted solution from the emulsion surface and replacing it with fresh.

The boundary layer at the emulsion surface under three regimes

1. 1Bulk developer — the composition on the label: unchanged, and never what the emulsion sees
2. 2The boundary layer — stagnant solution; its thickness is the only thing agitation sets
3. 3Agent profile (solid) — falls towards the surface, and falls furthest where density is high
4. 4Bromide profile (dashed) — rises towards the surface, and rises furthest where density is high
5. 5The loaded layer sinking — denser than bulk, so it flows downwards over whatever is below

The two chemical equations above are the same in all three panels. What changes from left to right is a distance, and every effect this experiment measures follows from it.

One mechanism, three scales, three names

Everything else on this page falls out of that picture by changing only the distance the bromide has to travel before it reaches solution that is being replaced.

• Micrometres — the adjacency effect. Where a heavily exposed region abuts a lightly exposed one, the inhibiting by-products diffuse sideways into the thin side while fresh agent diffuses the other way. Development runs faster at the edge of the dense area, giving a local density maximum (the border effect), and slower at the edge of the thin area, giving a local minimum (the fringe effect). That is the mechanism as the peer-reviewed literature states it, measured with a microdensitometer on traces whose distance axis is in micrometres.
• Millimetres to centimetres — bromide drag. The loaded layer over a dense area holds dissolved bromide and oxidised agent, is denser than the bulk, and sinks, retarding development in whatever it flows over: streaks below dense areas, running with gravity rather than with the picture. The troubleshooting entry has the diagnostic form.
• The whole frame — compensation and local exhaustion. A highlight consumes its local reservoir and slows; a shadow never does. The upper scale flattens and the lower does not — a shoulder manufactured in the tank rather than coated into the film.

One mechanism, two names, decided by scale

1. 1The knife edge — a dense region above a thin one: the geometry both effects need
2. 2Border and fringe — a density maximum on the dense side and a minimum on the thin side, tens of micrometres wide
3. 3The drag streak — the same chemistry carried by gravity, millimetres to centimetres wide
4. 4What the instrument can see — the 2 mm aperture averages the top inset away and measures the bottom one directly

The wanted effect and the unwanted one are not two phenomena. They are one phenomenon read at two magnifications, and only one of the two magnifications is available to this course's densitometer.

The three regimes, written out as scripts

A regime described as “occasional” is not data. Write these out, tape them to the bench, and record what you actually did against what you meant to do.

In a cylinder — a 135 mm strip standing in 80 mL in a 100 mL measuring cylinder, lifted by its top edge:

• G-C, continuous. Gently up and down for the whole 16 minutes without pause, about one cycle per second. ILFORD’s instruction is that continuous agitation reduces a spiral-tank time by up to 15 per cent, so at equal time this cell is expected to over-develop, by a published amount rather than a guessed one.
• G-I, intermittent — the control. Gently up and down for the first 30 seconds; then lift, drain for two seconds and re-immerse once at the start of each subsequent minute. This is Part IX’s script, and it is the course’s translation of ILFORD’s four inversions per minute into a vessel that cannot be inverted. Call it what it is in the report.
• G-S, semi-stand. Gently up and down for the first 30 seconds, then nothing except a single 10-second agitation at 8:00 — the halfway interval King’s published semi-stand scheme uses, transplanted as a shape and not as a specification, because it was published for a different developer.

In a tank — a 490 mm length of roll film on a spiral in 290 mL. R-H, the most agitation any maker publishes: Kodak’s small-tank procedure of 5 to 7 inversion cycles in 5 seconds at 30-second intervals. R-I, the control in its published form: ILFORD’s four inversions during the first 10 seconds and four again during the first 10 seconds of each further minute, tapping the tank afterwards to dislodge bubbles. R-S, semi-stand: ILFORD’s opening four inversions, then nothing until a single four-inversion sequence at 8:00.

✕Which maker's agitation scheme belongs to a borrowed time?

The control time is 16 minutes because ILFORD publish it for FP4 Plus in Kodak D-76 at 1+3, and the two makers publish different schemes: four inversions a minute against five to seven inversion cycles every 30 seconds. A published time is only meaningful with the agitation it was established under, so the control uses ILFORD’s scheme, because the time is ILFORD’s. If you take your time from Kodak’s own D-76 table instead, take Kodak’s agitation with it, and expect a different control.

!What no datasheet says about an hour in the tank

The design needs to know whether an hour of development harms the gelatin, and no film datasheet in the course’s corpus states a maximum development time, or any limit on how long a film may stand in a developer. What the sheets do bound is temperature: ILFORD give 20 to 24 °C as the usable range for these developers and ask that all process solutions be within 5 °C of the developer, and Bergger warn of emulsion damage above 27 to 28 °C for PMK. So the session holds 20 °C, fixes in the same non-hardening rapid fixer used elsewhere in the course, handles wet film by its edges, and records any physical damage as an observation rather than assuming there will be none. “No damage seen at 60 minutes at 20 °C on this film” is a small, real result; assuming it in advance is not.

The developing-agent floor

This part of the design has nothing to do with agitation, and it is the reason half the arguments about stand development are unresolvable.

Kodak state it plainly for XTOL, and the reasoning is general even though the number is not: the minimum amount of diluted developer needed to cover the film may not contain enough active ingredients to develop the film fully in the recommended time, so they recommend at least 100 mL of full-strength developer for each 135-36 or 120 roll, which they give as 80 square inches or 516 cm². Their D-76 sheet says the same in a different currency: one roll in 473 mL of 1+1, which is 236 mL of stock, or in 237 mL with the time increased by 10 per cent, which is 118 mL. Convert everything into one quantity before comparing anything:

V_stock = (V_bath ÷ (1 + n)) × (80 ÷ A)

Millilitres of stock per 80 square inches of film

V_bath is the working-solution volume in millilitres, n the parts of water per part of stock, and A the film area in square inches. A 135 × 35 mm strip is 47.25 cm², or 7.32 in² — a little over a ninth of a roll.

Developing agent per unit film area, against dilution, for the two vessels this session uses

• One 135 mm strip in 80 mL — this session's cylinder cells
• One whole 36-exposure roll in a 290 mL tank

Show the numbers behind this plot

Two falling curves against dilution on a linear horizontal axis running from zero to forty parts of water per part of stock, and a vertical axis of millilitres of stock developer per eighty square inches of film running from zero to two hundred and fifty. The upper curve, marked with dots, is the cylinder used in this session: one 135 by 35 millimetre strip in eighty millilitres of working solution, giving 218 millilitres of stock per eighty square inches at a one-plus-three dilution, 87 at one-plus-nine, 44 at one-plus-nineteen and 22 at one-plus-thirty-nine. The lower curve, marked with squares, is a one-reel tank holding a whole 36-exposure roll in two hundred and ninety millilitres, giving 145 at one-plus-one, 73 at one-plus-three, 29 at one-plus-nine, 15 at one-plus-nineteen and 7 at one-plus-thirty-nine. Three horizontal guide lines cross both curves: at 100 millilitres, which is the minimum Kodak recommend per roll for XTOL; at 118 millilitres, which is Kodak's own D-76 figure for one roll in a 237 millilitre tank with the time increased by ten per cent; and at 236 millilitres, Kodak's D-76 figure for one roll in 473 millilitres. The two cells of this session's fair test sit on the upper curve at one-plus-three, well above every guide, and the two floor cells sit on the upper curve at one-plus-nineteen, below all three. The drawing's point is that the lower curve, which is the ordinary way somebody stand-develops a roll of film at home, is already below the hundred-millilitre guide at one-plus-three and far below it at every dilution a stand procedure would actually use.

Series	Parts of water per part of D-76 stock (the n in 1+n)	mL of stock developer per 80 in² of film
One 135 mm strip in 80 mL — this session's cylinder cells	3.00	218.40
One 135 mm strip in 80 mL — this session's cylinder cells	9.00	87.40
One 135 mm strip in 80 mL — this session's cylinder cells	19.00	43.70
One 135 mm strip in 80 mL — this session's cylinder cells	39.00	21.80
One whole 36-exposure roll in a 290 mL tank	1.00	145.00
One whole 36-exposure roll in a 290 mL tank	3.00	72.50
One whole 36-exposure roll in a 290 mL tank	9.00	29.00
One whole 36-exposure roll in a 290 mL tank	19.00	14.50
One whole 36-exposure roll in a 290 mL tank	39.00	7.25

Arithmetic on three published volume statements and on the geometry of 35 mm film — not a measurement of development, and not a claim that Kodak's XTOL figure transfers numerically to D-76. What transfers is the reasoning: below some quantity of agent per unit area, the bath cannot finish the job whatever the time, and no manufacturer publishes that quantity for most developers. The strip cells were sized at 80 mL so that the fair test would sit at 218 mL per 80 in², and the roll-film lengths were cut at 490 mm so that the tank cells would sit at 214 — close enough that the two arms differ in vessel and not in nourishment.

Three things follow, and together they justify the floor arm. The fair test is deliberately far above the floor: both arms sit near 215 mL of stock per 80 in², close to Kodak’s generous D-76 figure of 236, so no regime is short of agent and the comparison is about transport rather than supply. The floor arm is deliberately below it: two cells at 1+19 in the same 80 mL sit at 44 mL per 80 in², under half Kodak’s XTOL minimum, so if a curve compresses at the top there, exhaustion is a live explanation.

And the ordinary home stand procedure is below the floor before it starts. A whole roll in a 290 mL tank is at 73 mL per 80 in² at 1+3 and 29 at 1+9, and a stand procedure would dilute further still. That is the trap: the worker sees a negative thin in the highlights, calls it compensation, and it is starvation. The two do not respond the same way — compensation is undone by more agitation, starvation only by more agent.

Compensation, defined before there is any data

“Compensating” is a word this course will not use qualitatively on a page with a densitometer on the bench. The definition below is the course’s own construction, stated here because no page before this one needed a number for it.

Pure Silver compensation ratio (R). Plot diffuse density against log exposure. Take base-plus-fog D₀ from the strip’s own masked patch. Locate the anchor m, the point where the curve reaches D₀ + 0.10 — the same point the course’s speed criterion uses. Then

G_low = [ D(log H_m + 1.00) − D(log H_m) ] ÷ 1.00

The lower-scale gradient

G_high = [ D(log H_m + 2.00) − D(log H_m + 1.00) ] ÷ 1.00

The upper-scale gradient

R = G_high ÷ G_low

And the ratio

Four notes. The intervals are 1.00 log H each, three and a third stops, and the pair spans 2.00 — the same span as the contrast-index chord, so a curve that supports one supports the other. On a T2115 that is thirteen and a third steps against a usable scale of about sixteen. Interpolate linearly between the two steps straddling each exposure, and say so.

R is compared between cells, not against unity. A negative whose toe is still rising through the first interval usually gives R above 1, because the second interval is straight line where the first is part toe. Compensation shows as R falling relative to the control, and a cell whose R drops below 1 has an upper scale genuinely flatter than its lower.

R is nearly immune to the error everyone worries about: both gradients are differences of densities, so a constant offset in the whole scale cancels in each, and again in the ratio. And R is not a measure of sharpness and must never be reported as one — it describes the shape of the characteristic curve, which is a whole-frame quantity, while the edge effects live three orders of magnitude below the aperture.

🧮The smallest difference in R this apparatus can report

Take u(D) as your densitometer’s combined uncertainty for one reading; a home-built head is unlikely to beat the ±0.01 D repeatability a commercial metal-cased instrument publishes for itself, and your own figure is on your own certificate.

Each gradient is a difference of two densities over a fixed 1.00 log H, so u(G) = u(D) × √2 = 0.0141 at u(D) = 0.01. For a typical pair — G_low = 0.50, G_high = 0.60 — the relative uncertainties are 2.8 and 2.4 per cent and the ratio carries √(2.8² + 2.4²) = 3.7 per cent, so at R = 1.20, u(R) = 0.044. A difference between two cells carries √2 of that, giving 0.062 and a 2σ band of 0.12.

Write this in the notebook before the session: at u(D) = 0.01, a difference in R smaller than about 0.12 is not a result. The band scales with u(D), so an instrument at 0.02 cannot report a difference below about 0.25 — which is most of the effect anyone claims.

Evenness, defined before there is any data

Nothing earlier in the course has measured this, and it is the quantity that makes the experiment worth running.

The evenness measurement. On a uniformly exposed strip, read density at 20 positions at 5 mm pitch along the strip’s length, starting 20 mm from the head, with the aperture centred across the strip’s width. Report the standard deviation s of the twenty readings, the range, and the profile itself. Then subtract the control cell’s profile position by position and report the standard deviation of the difference profile, which is the figure the cells are compared on.

Why each choice. Twenty positions at 5 mm pitch, because the aperture is 2.0 mm so the readings do not overlap, the run covers 95 mm of a 135 mm strip, and twenty is enough for s to mean something. A standard deviation rather than a range, because a range grows with the number of readings and is decided by one outlier, which on film is as likely to be a dust speck as a fault; quote the range too, but compare on s.

The difference profile is the headline, because every strip carries the same systematic pattern before development starts. The sensitometer’s field is a shallow dome, computed on the enclosure page as falling 0.053 log H at the ends of the 127 mm field — at a gradient of 0.6, 0.032 D of built-in curvature, the same size as the effect being looked for. Subtracting the control’s profile removes the dome, the coating variation and any common systematic in one operation, with no modelling, at a cost of √2 in noise. The control’s own s is then the floor: it contains the source’s non-uniformity, the coating, the densitometer’s replaced-sample repeatability and your hand on the rule, and no cell can be reported as more even than that.

!What the 2 mm aperture can and cannot resolve

The aperture is a low-pass filter, and its cutoff must be known before anything is attributed to chemistry.

Resolved: anything of roughly 4 mm or more — a drag streak below a dense step, the centimetre-scale patches manufacturers call mottle, the bubble artefacts King describes at a quarter to three-eighths of an inch across. Two samples per feature is the practical limit, so read a suspected streak again at 2 mm pitch across it rather than trusting the 5 mm run.

Not resolved: structure finer than about 4 mm turns into scatter. Fine mottle returns a profile with no shape and a large s; a drag streak returns one with a clear depression and a large s — same number, different diagnosis, and only the plotted profile separates them. Always plot the profile; never report s alone.

Out of reach entirely: the adjacency edge structure. The literature’s edge traces are made with a microdensitometer on a distance axis in micrometres, and Higgins and Jones’s threshold for the smallest visible density gradient is 0.005 D per micrometre — a 2 mm aperture averages over two thousand of them. This experiment cannot measure acutance, and any sentence in your report claiming it did is wrong.

Two profiles with the same standard deviation and different diagnoses

• A control-like profile: small scatter, no shape
• A resolved feature: a broad trough, as a drag streak reads
• Unresolved fine structure: scatter with no shape

These three traces are drawn to show the shapes the measurement can take and what each one means. None is a measurement of any film, and none is attached to a regime: which regime produces which shape, if any, is exactly what your session decides and what this page refuses to tell you in advance. The curve is drawn to show the shape, not measured from a real material. Your own materials will differ, and measuring them is what the sensitometry part of the course is for.

Where an even exposure actually comes from

The uniform strips need a source flat enough that its own structure does not swamp what is being measured. Three candidates are on the course’s bench, and only one of them qualifies.

The sensitometer with the wedge lifted off, which is the choice. Same box, lamp, fences and firmware, and a field whose non-uniformity is already a number in your certificate. The one change is the exposure: the box is dimensioned to expose through up to 3.05 D of wedge, so without it the film sees up to a thousand times more light. The design exposure is 500 ms; the uniform strips will want something near 5 to 20 ms, and the firmware reports measured_us for every flash, so what was delivered is logged rather than assumed. Bracket it first — four scrap strips at 4, 8, 16 and 32 ms developed together in the control regime — and take the one landing about 0.9 above base, where the gradient is high and a small development difference shows most.

The enlarger, which is the fallback if your timer cannot deliver a repeatable few milliseconds: the head with no negative in the carrier, stopped down, strips on the baseboard in darkness. Its uniformity is mapped in density units on the material by the commissioning page, whose acceptance criterion is opposite corners within 0.04 D; the cost is a second instrument with a second calibration.

The pinhole camera, which is disqualified, because illumination on its film plane falls off as cos⁴ of the field angle and Part VI computes 3.7 stops of corner falloff for the course’s reference camera. A source whose own field varies by four stops cannot measure evenness to a few hundredths of a density.

The camera roll is not a measurement instrument either. It is exposed through a lens with its own falloff, on a real subject, and it is here because mottle and drag show themselves in continuous tone and on perforated film. It yields photographs of defects and a difference profile, not an absolute evenness figure — and the difference profile is legitimate because the lens’s falloff is identical in every frame and cancels when one regime’s frame is subtracted from another’s.

The run order and the table drawn before the data

Eight cells, in an order drawn from a hat and written down beforehand, so that a drift in the bath, the lamp’s age or your own attention cannot line up with the variable. Two constraints bind the draw: the continuous cell occupies both hands for a quarter of an hour and takes a batch to itself, and the floor cells occupy an hour, so they start first with everything else run underneath them.

Cell	Vessel	Dilution	Volume	Time	Regime	Strips
G-C	100 mL cylinder	1+3	80 mL	16:00	Continuous	wedge, uniform
G-I	100 mL cylinder	1+3	80 mL	16:00	Intermittent (control)	wedge, uniform, wedge replicate
G-S	100 mL cylinder	1+3	80 mL	16:00	Single agitation at 8:00	wedge, uniform
F-16	100 mL cylinder	1+19	80 mL	16:00	Single agitation at 8:00	wedge, uniform
F-60	100 mL cylinder	1+19	80 mL	60:00	Single agitation at 30:00	wedge, uniform
R-H	one-reel tank	1+3	290 mL	16:00	Kodak’s 30-second scheme	490 mm roll length
R-I	one-reel tank	1+3	290 mL	16:00	ILFORD’s per-minute scheme	490 mm roll length
R-S	one-reel tank	1+3	290 mL	16:00	Single sequence at 8:00	490 mm roll length

Read the table for what it admits. G-C, G-I and G-S differ in one thing. F-16 differs from G-S in one, the dilution, and therefore in agent per unit area. F-60 differs from F-16 in one, the time — and so from G-S in two, which is why it is never quoted as an agitation result. R-H, R-I and R-S differ in one and share the control condition, so the arms are tied together at G-I and R-I and nowhere else.

Draw the empty data table now, a row per piece of film and a column per quantity, with a dash in every cell: a table drawn afterwards has its interesting rows already chosen.

Procedure

After the exposure batches: preflight, 30 minutes; allocating the film in darkness, 15; the floor cells started, 5; the cylinder arm, 45; the tank arm, 55 nested inside the floor cells; wash, dry and file, 30; and the reading session on another evening, 90.

Stage 0 — The two exposure batches

1. The strips, in one sitting, from one lamp at one distance, per the exposure-session SOP, logging the minutes since switch-on against each: seven wedge strips plus two spares through the wedge, then, with the wedge and cover glass lifted off and the mask still in place, four uniform strips plus two spares at the single bracketed exposure chosen in Preparation, logging measured_us for each. Every strip carries the masked patch the test-negative lab specified, and the uniform strips are notched on the opposite edge.
2. The continuous-tone roll. Expose a whole cassette on one even-toned subject at one meter setting in one session — a matt wall lit obliquely, or an overcast sky well away from the sun — filling the frame every time, aiming for the same mid-grey as the uniform strips. Do not bracket.
3. Store everything in the closed tin and develop as soon as the calendar allows: ILFORD’s instruction is to process exposed film as soon as practical, and the course has no figure for what months of keeping do to a latent image.

Stage 1 — Preflight

4. Check the thermometer per the bench-check SOP and write the offset down. Bring the bath to 20 °C and give it twenty minutes to settle with the empty vessels in it. Kodak state that a developer temperature varying by more than 0.3 °C affects process control and image quality; ILFORD ask for all solutions within 1 °C. If your bath cannot hold that, write the figure it can hold and carry it through the analysis.
5. Lay the bench out in the light in the order it will be used and find the light switch by touch. Make up the fixer at 1+4 and the wetting agent at 1+200, bring the rinse water to 20 °C, and measure the clearing time on a scrap per the clearing-time SOP; fixing time is twice it.
6. Label the two floor cylinders on the vessel and set them in the overflow tray, in their final place, before anything is poured. They stand there for an hour.

Stage 2 — Allocate the film, in darkness

7. With the light off, nip each strip along its top edge to its cell — one nip for G-C through five for F-60 — with the uniform strips nipped on the lower edge so they cannot be confused by touch. Close the tin before the light goes on.
8. In the changing bag, cut the continuous-tone roll into three 490 mm lengths, notch each and load each onto its own spiral. The length is not arbitrary: at 290 mL of 1+3 it puts the tank cells at the same agent per unit area as the cylinder cells.

Stage 3 — The floor cells, started first

9. Dilute 160 mL at 1+19 — 8 mL of stock into 152 mL of water — split it between the two F-16 cylinders and start the clock; then a second 160 mL for F-60, a minute later. Both get 30 seconds of gentle movement and nothing until their single agitation.
10. Record the volume, the dilution and the millilitres of stock as a number, not a ratio. Four millilitres of stock per strip is the point of these cells, and a ratio hides it.

Stage 4 — The cylinder arm

One evening, eight cells: an example draw

F-60 · 1+19 · single agitation at 30:00

develop 60:00rinsefix

F-16 · 1+19 · single agitation at 8:00

develop 16:00rinsefix

G-S · 1+3 · single agitation at 8:00

develop 16:00rinsefix

G-I · 1+3 · intermittent · control

develop 16:00rinsefix

G-C · 1+3 · continuous · both hands

develop 16:00rinsefix

R-I · tank · ILFORD scheme · control

develop 16:00rinsefix

R-S · tank · single sequence at 8:00

develop 16:00rinsefix

R-H · tank · Kodak 30-second scheme

develop 16:00rinsefix

0 min20406080

The floor cells are started first because they are the longest, and everything else runs underneath them; the continuous cell takes a batch to itself because it occupies both hands for a quarter of an hour. The order shown is one draw from the hat and is not a prescription — yours will be different, and the difference is the point.

11. Batch one: G-S and G-I. Dilute one 160 mL jug of 1+3 for G-S and one 240 mL jug for G-I, whose control cell has three strips in three cylinders, pouring each cylinder immediately before its cell begins. The semi-stand cell needs no hands, so start it a minute early and leave it.
12. Log the developer temperature at the start, middle and end of every cell, with the thermometer in the cylinder rather than the bath. Then lift each strip at its own mark, rinse for one minute in two changes of plain water at 20 °C, and fix for twice the clearing time.
13. Batch two: G-C. Fresh 160 mL of 1+3, two cylinders, hands on both for the full 16 minutes. Kodak’s instruction is that a diluted solution is discarded after its batch, and a bath that has already developed a strip is a different bath — the subject of the exhaustion experiment, which must not leak into this one.

Stage 5 — The tank arm, nested inside the floor cells

14. Fill each tank with 290 mL of freshly diluted 1+3 and start it on its written scheme. With one tank, run the three in sequence and record each clock time, because the bottle is an hour older by the third.
15. Tap the tank after every agitation sequence, as both makers instruct, to dislodge bubbles: a bubble on the emulsion during a stand cell stops all exchange of developer at that spot, and King records that the resulting circular artefact ruins the frame. Rinse and fix as for the strips, and do not open a tank early to look.

Stage 6 — Wash, dry, label and file

16. Wash everything for 5 to 10 minutes in running water within 5 °C of the process temperature, or use ILFORD’s spiral-tank sequence of five, ten and twenty inversions in three changes. Final-rinse in wetting agent at 1+200 and hang in still, dust-free air. Do not squeegee anything: ILFORD warn that too little or too much wetting agent leads to uneven drying, and here an uneven dry reads as a result.
17. Complete a card slip for every piece of film and record the developer stock’s formula version code, issued per the versioning SOP on the formula version record. A result tied to “the D-76 on the shelf” is a result nobody can check.
18. When everything is bone dry — not before — sleeve and file it. Do not read anything tonight: film that is not quite dry reads high.

Stage 7 — The reading session, on another evening

19. Run the densitometer’s warm-up and daily check per its SOP, recording the calibration step before the first piece of film and after every third.
20. Read all twenty-one steps and the masked patch of every wedge strip, in a separately drawn reading order, logging to a file. Read one strip twice, lifting it off the stage and replacing it between readings: that replaced-sample spread is this session’s Type A term.
21. Read the twenty profile positions on every uniform strip, marking the sleeve so the positions match across strips to better than a millimetre, then five transverse positions at the midpoint as a check for a cross-strip tilt.
22. Read a twelve-position profile across one frame of each roll length, in the same place in the frame on all three, from one perforated edge to the other — the direction gravity ran while the film was coiled.

Expected observations

What follows is what the design expects, and what would show the expectation wrong. It is not a prediction of your numbers: a page that supplies those has run the experiment for you.

In the trays. The three fair-test strips should look more alike than the two floor strips do, and the 1+19 cell at 16 minutes should come back visibly thin — not a spoiled strip but the cell doing its job. Whether an hour is enough at 1+19 is genuinely open, for the reason Wratten and Wainwright measured, and finding out is part of what the 60-minute cell is for.

Where to look for a defect. On the cylinder strips, drag runs along the strip, because the strip stood vertically. On the roll lengths gravity ran across the film, because it was coiled, so look for streaks running inwards from the perforated edges and for regular marks at the perforation pitch — measure that pitch on your own film rather than assuming it, because the course has not verified the standard figure. Kodak’s troubleshooting table places streaks of non-uniform density against excessive or uneven agitation and mottle against inadequate agitation, so both ends of the ladder are expected to fail, and differently; the surge-marks entry is explicit that the course has found no manufacturer publication naming the sprocket-hole mark specifically.

Five things that would falsify the design rather than the chemistry. The control cell’s two wedge strips disagreeing by more than the resolution sentence allows. The control’s uniform profile having a standard deviation as large as the differences you hoped to see. The 1+19 cells clearing the top of the wedge, which would mean the dilution did not cross the floor. Any cell’s temperature log leaving the stage 1 tolerance. And R-I disagreeing with G-I by more than the between-strip spread, which would say the two vessels are not running the same process and the arms cannot be tied together at all.

What is happening chemically

Bromide does two jobs, and the experiment has to keep them apart

The transport account is in the Preparation, above; what it leaves to be said is what the bromide does once the boundary layer has held onto it. Chemically it is a restrainer, acting through the common-ion effect at the crystal surface:

AgBr(s) ⇌ Ag⁺ + Br⁻

Added bromide pushes this equilibrium left and lowers the silver ion concentration at the crystal surface, which is how a restrainer restrains

That is a concentration effect and it acts wherever the bromide is. Well agitated, the released bromide disperses into 80 or 290 mL and the change is negligible; in a still bath it stays within a fraction of a millimetre of where it was made, and the local concentration can be large. The chemistry Part IX’s bromide arm studied by adding potassium bromide to a whole bath is here generated selectively by the image, in proportion to local density. A still bath restrains itself in exactly the places developing hardest — compensation in one sentence, and the reason Kodak can say that agitation affects the rate of development particularly in high-density areas.

The second job is physical, and nothing chemical happens during it: the loaded layer is denser than the bulk and sinks, arriving somewhere it was not made and restraining development there, giving a streak whose shape follows gravity and the film’s geometry rather than the picture.

Why the compensating shape and the starved shape look alike

Compensation flattens the upper scale because the highlights run their local reservoir down faster than diffusion refills it. Starvation flattens it because the whole bath never held enough agent to develop them out. The shapes are similar and the causes are not, and the discriminator is built into the design: compensation is a function of transport, starvation a function of supply. Two cells at one dilution and volume differing only in agitation isolate the first; two cells at one agitation and volume differing only in dilution isolate the second. That is the whole reason the floor arm exists, and it is why a claim about stand development made without a stated volume of stock per unit area is not checkable.

A third mechanism sits at the same end of the curve and must not be attributed to either: in a very dilute bath standing for an hour with a large surface open to the room, the agent is also consumed by air. Aerial oxidation is why Wall could already call stand development uneconomical in 1924. The floor cells stand longest and are the most exposed solution in the session, so oxidation is a real term in them that nothing here separates from starvation — say so rather than choosing between them.

Data to record

Session header, as the design lesson sets out: date, operator, room temperature, the developer’s formula version code, both instrument certificates by date, the thermometer’s offset, the wedge’s serial and whether it is calibrated, and the drawn run and reading orders.

Per cell, before any film enters it:

Field	Why
Vessel and its volume	The vessel is part of the result
Dilution as 1+n, and millilitres of stock	The ratio hides the quantity that matters
Film area, and mL of stock per 80 in²	The floor comparison is impossible without it
Agitation regime, as intervals and durations	“Occasional” is not data
Time, to the second, as run rather than as planned
Temperature at start, middle and end, in the developer	Decides whether the cell is comparable
Clock time the cell started	The bottle ages during the session

Per wedge strip: all twenty-one step densities, the masked patch, the replicate readings, and the derived base plus fog, contrast index, relative speed point, maximum density, G_low, G_high and R, each with its uncertainty.

Per uniform strip: twenty profile densities with their positions in millimetres, the five transverse readings, the mean, the range, s, and s of the difference profile against the control.

Per roll length: the twelve-position transverse profile of the nominated frame, and a photograph of every defect, taken on a light box with a rule in the frame before the film is sleeved and indexed to the notebook page. A defect described in words is an anecdote; one photographed with a scale beside it is evidence.

Analysis

1. The resolution sentence, restated for this page’s five quantities

Before any comparison, write down what your instrument can report, by the propagation method the curve-families page sets out. Five quantities, five different answers from the same readings:

Quantity	How its uncertainty is got	Rough size at u(D) = 0.01
Maximum density	Read directly	0.01; a difference between cells, 0.014
Contrast index	Perturb the density table and re-solve the construction a few thousand times	About the same as one reading
Relative speed point	The same perturbation; the multiplier is one over your own toe gradient	Three to four times worse than a density
Compensation ratio R	The arithmetic in Preparation	0.044, so a difference below about 0.12 is not a result
Evenness s	Sampling uncertainty of a standard deviation from n = 20	About 16 per cent of s; a ratio of two must exceed about 1.5

The last row is the one nobody expects. The relative standard error of a standard deviation from n readings is 1 ÷ √(2(n − 1)), which at n = 20 is 16 per cent; a ratio of two carries 23 per cent, so two cells’ evenness figures must differ by a factor of about 1.5 before the difference is a two-sigma result. Resolving a factor of 1.2 would need about forty positions on a strip 135 mm long.

2. The three fair-test cells

Plot the three curves on one field, tabulate the five quantities with their uncertainties, and answer three questions in this order, writing each answer down before moving on.

Did the contrast move, and by how much in units of time rather than contrast? ILFORD state that continuous agitation reduces a spiral-tank time by up to 15 per cent, so at equal time that cell is expected to over-develop by a published amount. Convert each cell’s contrast index into an equivalent development time on the curve you built in the time series and report the regimes as time differences. That is the fair currency: this regime at 16 minutes did what the control would have done at 14, or at 19, and a photographer can act on that.

Did the speed point move, and does the contrast change alone account for it? A change in gradient moves the point at D₀ + 0.10 even when sensitivity is untouched, so read the shift, convert it to stops and subtract the shift the contrast change predicts. A speed difference is only a speed difference if it survives that subtraction — which is how King’s claim of increased emulsion speed under minimal agitation gets tested rather than repeated.

Did R fall as agitation fell? Report ΔR against the control with its uncertainty and against the 0.12 band. If the ordering is right but the differences sit inside the band, the honest conclusion is that the effect on this film in this developer at this dilution is smaller than the apparatus can resolve — a result, and a more useful one than a rank order asserted without an error bar.

3. Evenness, and the two ways it fails

Plot all four uniform-strip profiles on one field with the control’s ±2u band on it, then the three difference profiles against the control on a second. Report per cell: the mean, the range, s, s of the difference profile, and one sentence saying whether the profile has a shape.

That sentence is the diagnosis and no number supplies it. A shaped difference profile — a trough, a ramp, a step — is a resolved feature, and the film should be inspected in transmitted light where the profile says it is. A shapeless one with a large s is unresolved structure below the aperture, which is mottle by another name. A shapeless one with a small s is a cell that developed evenly, which is a real finding.

Then check the transverse five. A consistent tilt across the width in every cell including the control belongs to the source or the vessel rather than the regime, and goes in the certificate, not the result.

4. The floor arm: separating agitation from starvation

Compare F-16 against G-S first: same regime, vessel, volume, time and temperature, differing only in dilution and so in stock per unit area, by a factor of five. Whatever separates them is a difference in supply and not in transport.

Then compare F-60 against F-16, which differ only in time, and say whether the extra 44 minutes brought the curve back up. If it did, the 16-minute cell was time-starved rather than agent-starved; if it did not — if the upper scale stayed depressed while the lower rose — the bath ran out of what it had, which is the floor doing what Kodak’s sheet says it does.

Finally, put F-60 beside G-S and refuse to draw an agitation conclusion from the pair, which differs in two things. Its use is in the verdict: it is the session’s closest cell to what a photographer actually does when stand-developing a roll, and its position on the agent-floor plot is the sentence to quote whenever somebody reports a compensating negative without saying how much developer was in the tank.

5. The roll-film arm

The tank cells produce three things: a tie to the cylinder arm through R-I against G-I, whose difference in contrast index and in R is the vessel term for the session; a difference profile across a frame for R-H and R-S against R-I, in which the lens’s falloff cancels because it is the same in every frame; and the defects, which are why the arm exists.

Score each defect as a diagnostic atlas does rather than as a forum post does — what you see, where it sits relative to the film’s geometry, its scale in millimetres — then look it up against bromide drag, surge marks, mottled development and uneven density across a strip, and say which entry it matches and which it does not.

6. The verdict, and the form it has to take

Four paragraphs, in this order, keeping your number and the published number visibly apart in every one.

What this session measured. Five quantities for three regimes, on one film in one developer at one dilution, time and temperature, in one vessel, by one worker, on one evening, with an instrument whose uncertainty is stated. Every number is the course’s own under the course’s own criteria.

What the sources contain, restated in your own words because it bounds the conclusion: no datasheet for a general-purpose developer publishes a reduced-agitation scheme, the two that do are for one developer nobody may transplant from, and no Tier 1 or Tier 2 source publishes measured sensitometric data for stand development at all.

What the data supports and what it does not. Name each hypothesis and say whether the difference exceeded the resolution; then name what the session could not test — sharpness, because the aperture is three orders of magnitude too coarse; reproducibility, because n is one per regime; generality, because it is one film and one bath.

The cost side. Stand development takes an hour instead of sixteen minutes; leaves a large volume of dilute developer standing and oxidising, which Wall called uneconomical in 1924; and is irreproducible enough that its own advocate advises two back-up shots of any important scene. It also produces a curve shape available a second way: ILFORD publish 8 minutes at stock and 16 minutes at 1+3 as giving the same average contrast on the same film — a fourfold dilution and a doubled time, with normal agitation and no evenness penalty. If what you want is compensation, the sourced route is dilution and time. If what you want is the edge effect, this apparatus cannot tell you whether you got it.

What cannot be concluded from this session

• That stand development is better or worse than intermittent agitation. The session measures five quantities, and one of the things photographers care about most is not among them.
• That any result here transfers to another film, developer, dilution, vessel, water supply or worker. Each of those is a variable this session held fixed.
• That an observed sharpness difference is real: it was not measured and cannot be, with this instrument.
• That the floor cells’ behaviour is starvation rather than oxidation. Both were present and the design cannot separate them.
• That an absent defect will stay absent. One roll per regime is one sample, and King’s own account is that the practice ruins some frames some of the time — a rate no session this size can estimate.

Troubleshooting

What you see	Likely cause	What to do
Streaks running down from the dense steps of the semi-stand strip	Bromide drag, the published failure mode of reduced agitation	A result, not a fault. Photograph it with a rule, read a profile through it at 2 mm pitch, and report it
The continuous strip is no denser than the control	The agitation was not actually continuous, or the effect is below your resolution	ILFORD’s figure is up to 15 per cent of development time, roughly one wedge step. If your resolution is one step, this comparison was always marginal and the report should say so
The control’s two wedge strips disagree by more than the budget allows	Session repeatability is worse than assumed	Nothing here can be reported to a tighter band than that spread. Widen the resolution sentence and re-read every conclusion against it
The 1+19 cells are so thin the toe cannot be found	The floor arm crossed further than intended	Read what you can, report those cells as not analysable for R, and repeat the arm at 1+9 rather than interpolating
A circular blotch a few millimetres across on a stand cell	An air bubble, which King names as the characteristic ruin of the method	Record it and consider a pre-soak next time — but ILFORD advise against a pre-rinse, so say which advice you followed
Regular marks near the perforated edges of a roll length	Solution driven through the perforations	Measure the pitch. If it matches the perforations it is a surge mark, from the over-agitated end of the ladder
A gradient along the length of every strip, control included	The sensitometer’s field, a computed 0.053 log H dome	This is why the difference profile exists. Subtract the control and carry on

Clean-up

Empty each developer cylinder into the developer waste container as its cell finishes, rather than leaving a standing cylinder for the end of a long session. Rinse cylinders, tanks and spirals in warm water and stand them to dry — dedicated developer glassware never meets fixer, per the general waste SOP. Wipe the bench, dry the thermometer and put the wedge back in its sleeve, because a fingerprint on a step wedge is a systematic error in every session that follows. Cap the stock bottle full and record on the version record how much was drawn.

Storage

Sleeve the strips in pairs and the roll lengths singly, each with its card slip, in the same sequence as the data file, and file them with the Part IX agitation strips: the two sets are the only between-session replicate the course has for this comparison. Keep the defect photographs with their notebook page. Store the D-76 stock per the storage rotation SOP, tightly capped and full; ILFORD’s instruction for their own diluted developers is that 1+1 and 1+3 solutions are not kept more than 24 hours, so every dilution here is made and discarded on the day.

Disposal considerations

Three streams, described as chemistry and general practice; what you may actually do with each is set by your local regulations and your water authority, and this page does not override them.

Spent developer is alkaline and carries metol, hydroquinone, sulfite, borate and the bromide released by fourteen pieces of film — a reducing solution with an oxygen demand, containing a boron compound, which is the part most likely to attract a local restriction. Spent fixer and the first rinse after it carry dissolved silver–thiosulfate complexes and go to the silver stream, which exists both because silver is a regulated discharge and because it is recoverable. The remaining rinse and wash water is dilute enough to go with the general stream in most places, and “most places” is doing real work in that sentence. The disposal page has the framework; your authority has the rule.

Questions

Check your understanding

Question 1. Two cells develop the same film in the same D-76 at the same temperature for the same time, and one is agitated continuously while the other is agitated once at half time. Which of these does the agitation difference change?

• The thickness of the stagnant layer at the emulsion surface, and therefore the flux of agent inwards and bromide outwards
• The redox potential of the developing agents
• The solubility of silver bromide in the bath
• The pH of the developer at the emulsion surface, by an amount the course can calculate

Show the answer and why

Answer: The thickness of the stagnant layer at the emulsion surface, and therefore the flux of agent inwards and bromide outwards

Neither development equation contains a term for stirring. Agitation sets a distance — the thickness of the stagnant layer — and through Fick's first law that distance sets the concentration gradient and therefore the flux. Everything else follows: the agent arrives more slowly, the bromide leaves more slowly, and both effects are largest where the emulsion is working hardest, which is why Kodak say agitation affects the rate of development particularly in high-density areas. Option 4 is the tempting one: the local pH does change slightly, because development spends hydroxide, but the course has no figure for the size of it in a working bath and inventing one would be Rule 1 territory.

Question 2. A photographer reports a beautifully compensated negative from a whole 36-exposure roll stand-developed in 300 mL of D-76 at 1+9. What is the first question to ask before believing that agitation produced the shape?

• How many millilitres of stock developer was that per 80 square inches of film, and how does it compare with the manufacturer's stated minimum?
• What was the temperature of the bath?
• Was the film fixed for twice the clearing time?
• Was the developer mixed with distilled water?

Show the answer and why

Answer: How many millilitres of stock developer was that per 80 square inches of film, and how does it compare with the manufacturer's stated minimum?

300 mL at 1+9 is 30 mL of stock, and one 135-36 roll is the 80 square inches Kodak use as their reference area. So that negative was developed on 30 mL of stock per 80 in², against the 100 mL Kodak recommend as a minimum for XTOL and the 118 mL their own D-76 sheet implies for a 237 mL tank with the time increased. The bath was at roughly a quarter to a third of the published floor, so a depressed upper scale has an explanation that has nothing to do with agitation: there was not enough agent to develop the highlights out. Compensation and starvation give similar curve shapes and respond to completely different remedies, and no claim about stand development is checkable unless the volume of stock per unit area is stated.

Question 3. Two uniform strips both return a profile standard deviation of 0.026 D across twenty positions. One profile is a smooth broad trough and the other is shapeless point-to-point scatter. What do the two mean?

• The trough is a feature larger than the aperture, most likely a drag streak; the scatter is structure finer than about 4 mm, which the aperture cannot resolve and turns into noise
• They mean the same thing, since the standard deviation is the same
• The scatter is an instrument fault and the trough is real
• The trough is an instrument fault and the scatter is real

Show the answer and why

Answer: The trough is a feature larger than the aperture, most likely a drag streak; the scatter is structure finer than about 4 mm, which the aperture cannot resolve and turns into noise

A 2 mm aperture sampled at 5 mm pitch is a low-pass filter. Anything broader than about 4 mm is resolved and appears as a shape; anything finer is averaged inside the aperture and re-emerges as scatter between positions. Same statistic, different physics, and only the plotted profile separates them — which is why the page insists that s is never reported alone. It is also why fine mottle and a coarse streak are both real findings: the instrument is telling you the truth about both, just in different currencies.

Question 4. Sandy King publishes contrast-index charts for a dozen films in Pyrocat-HD and also publishes a stand-development procedure. Why can the charts not be used as measured data about stand development?

• Because he states the charts are based on rotary processing in BTZS tubes, which is continuous agitation
• Because contrast index is not a valid measure for a staining developer
• Because the charts are for sheet film and stand development is for roll film
• Because the charts have no error bars

Show the answer and why

Answer: Because he states the charts are based on rotary processing in BTZS tubes, which is continuous agitation

The charts carry their own condition and it is the opposite end of the agitation ladder from the procedure. That is not a criticism of King, who states the condition plainly on the same page; it is a warning about how easily two things published by one author get read as one body of evidence. It is also the single most useful fact in the source review: if the practice's leading specialist advocate publishes no measured curves for it, then the reader's own strips are not merely a contribution to the evidence, they are very close to all of it. Option 4 is a fair complaint about almost every published sensitometric chart, but it is not what disqualifies these ones.

Question 5. Your control cell gives R = 1.24 ± 0.044 and your semi-stand cell gives R = 1.17 ± 0.044, both at u(D) = 0.01. What may you report?

• That the difference of 0.07 is inside the 0.12 two-sigma band, so the effect on this film in this developer is smaller than the apparatus can resolve
• That reduced agitation lowers the compensation ratio by 0.07
• That reduced agitation has no effect on the compensation ratio
• That the measurement failed and the cells should be re-run

Show the answer and why

Answer: That the difference of 0.07 is inside the 0.12 two-sigma band, so the effect on this film in this developer is smaller than the apparatus can resolve

The difference of two quantities each carrying 0.044 carries 0.044√2 = 0.062, so two sigma is 0.12 and a gap of 0.07 does not clear it. Option 2 reports a number the instrument cannot support; option 3 makes the opposite error, since failing to resolve an effect is not evidence that it is absent. Option 4 mistakes a bounded result for a failure: "smaller than 0.12 on this film in this developer" is a genuine, quotable finding, and it also tells the next worker exactly what they would need — a better densitometer, or more replicates, or a larger difference between the regimes — to do better.

Question 6. Why does this page tie its cylinder arm and its tank arm together only through their two control cells, instead of comparing any cylinder cell directly with any tank cell?

• Because the vessel changes the agitation, the volume and the film geometry all at once, so a cylinder-to-tank comparison is a compound change with the controls as the only shared point
• Because the tank cells use a different developer
• Because the tank cells are developed for a different time
• Because a densitometer cannot read roll film

Show the answer and why

Answer: Because the vessel changes the agitation, the volume and the film geometry all at once, so a cylinder-to-tank comparison is a compound change with the controls as the only shared point

A strip standing in 80 mL in an open cylinder and a coiled length in 290 mL in a closed tank differ in the form of agitation (lifting against inversion), in the direction gravity runs relative to the image, in whether perforations are present, and in how much solution surface is open to the air. Holding the developing agent per unit area equal — which is why the roll lengths are cut at 490 mm — removes one of those differences and not the rest. The controls are run in both vessels precisely so that the remaining vessel term can be measured once and quoted, instead of being smuggled into every comparison.

Check my answersClear

Further experiments

The Rodinal-type stand run, done honestly. The practice is done overwhelmingly in a p-aminophenol developer at dilutions like 1+100, and the corpus contains no published time for that pairing at all — ILFORD’s FP4 Plus sheet stops at 1+50. Run it as a declared exploration rather than an experiment: one roll, one dilution, one time, the concentrate per 80 in² recorded, the result reported as a single observation. One safety point from the formulary entry is not optional: the concentrate is strongly alkaline in the bottle even though the working solution is dilute, so handle it with eye protection and add it to water rather than the other way about.

The volume arm this session did not run. Hold the dilution at 1+19 and the regime at semi-stand and change only the volume, 80 mL against 240 mL in a larger cylinder. That crosses the agent floor with one variable instead of two, and costs two strips.

The pre-soak question, where two published sources disagree. King strongly recommends a five-minute pre-soak before any minimal or stand development, to prevent the bubbles that ruin frames; ILFORD state that a pre-rinse is not recommended because it can lead to uneven processing. Run one pair of semi-stand cells with and without, and report which failure you got.

The between-session replicate. Read the archived Part IX agitation strips on today’s densitometer, alongside these: same developer, dilution and time, under three scripts differing only in the semi-stand cell’s timing. The spread between the two sessions is the closest thing this course has to a reproducibility figure.