Experiment: Process Control with Control Strips

Purpose

To stop guessing. Every darkroom conversation about whether the developer is tired, whether the new bottle is different, whether it was the temperature — all of it is guesswork until somebody puts a known exposure through the process on a regular schedule and plots the answer. That is what a control strip is, and now that you have an instrument that can make one, this is the session that turns the instrument into a habit.

The hypothesis, in two parts. First, that three numbers read off one small strip — fog, a shadow density and a contrast figure — plotted against the date, will separate a process that has genuinely moved from one that is merely noisy, and will do it before the change is visible on a picture. Second, and more sharply, that a 2 °C temperature error and a 20 per cent development-time error will move the chart by nearly the same amount, because ILFORD’s own time-temperature chart makes 1 °C worth about 9 per cent of development time. If that is right, the reason temperature is the dangerous variable is not that it is powerful. It is that being 2 °C out is easy and being 20 per cent out on a timer is not.

The control. Three, and they are not interchangeable. An unexposed strip in every tank, which fixes base plus fog for that development. A reference strip from the archived pre-exposed batch, which is the same known exposure carried unchanged from tank to tank and is what makes two runs comparable at all. And the instrument certificate from the calibration experiment, which is the only thing that lets you say a movement on the chart is bigger than the instrument that made the strips.

The one variable that changes is nothing at all during the monitoring runs — the calendar is the only variable, which is what makes the chart a measurement of drift — and then exactly one thing per run during the three deliberate perturbations at the end: temperature, time, or the age of the developer.

Learning objectives

By the end of this session you should be able to:

• pre-expose, label and store a batch of control strips, and say what storing them does to the method;
• choose three steps to track and justify each choice against what it is meant to detect;
• set aims and limits from your own data rather than copying a manufacturer’s, and say why the manufacturer’s cannot be copied;
• read a control chart for outliers, level shifts and trends, and resist the temptation to react to noise;
• name, for each of four chemical causes, the direction it moves each of the three plotted numbers, so that a pattern on the chart becomes a diagnosis;
• predict the effect of a 2 °C error and a 20 per cent time error before you run them, and account for the difference between the prediction and the result;
• separate instrument drift from process drift using the certificate and a freshly exposed strip;
• say plainly what a control chart cannot detect.

Prerequisites

The calibration experiment, with its certificate written. Without a repeatability figure you cannot set a limit, and a chart with arbitrary limits is decoration.

Part VIII on development kinetics, which owns why development is a rate process and why temperature moves it as much as it does. This page reads the chart; that page explains the chemistry underneath it.

Part VIII on restrainers and antifoggants and on sulfite and preservation, for the two mechanisms this page will watch happen: bromide as a restrainer, and sulfite as the thing that runs out.

Part XI on fixer capacity and exhaustion, because two of the faults a chart will show you are on the fixing side of the process rather than the developing side.

Safety classification

Level A, on the same reading of the rubric as the calibration experiment: working dilutions only, nothing heated above 50 °C, no mains construction, and waste that is spent developer and silver-bearing fixer collected for recovery.

The one thing that had to be considered. One arm of this experiment deliberately uses a developer kept past the manufacturer’s stated life in a half-filled bottle, and an aged developer contains more oxidation products than a fresh one. The course has found no manufacturer statement that an oxidised developer is more hazardous than a fresh one, and it will not invent one, so the instruction is the conservative one: handle the aged bath exactly as you handle a fresh bath, with the same gloves and the same eye protection, and do not decant it into anything unlabelled. What has certainly changed is its photographic activity, which is the whole point of the arm.

What is not a hazard here, and why. Nothing is weighed, so there is no dust and no need for a particulate mask. Nothing is concentrated: every solution is diluted from a stock made in an earlier part, and the most aggressive thing in the room is a working-strength fixer. The instrument’s emitter runs inside a closed light-tight box for every exposure. And the fixer, though it is the waste stream that matters environmentally, is not a splash hazard beyond what the Level A gloves and glasses already cover.

Hazards

The processing hazards are the ones on Part XIII’s step-wedge lab and are cited rather than repeated: developer as a skin sensitiser and irritant at working strength, fixer as an irritant with a silver-bearing stream, stop bath as a dilute acid.

Cross-contamination, which is this page’s own hazard and is a hazard to the data before it is a hazard to you. ILFORD’s advice to a beginner is to use a different-coloured beaker for each solution and warns that a trace of fixer can contaminate the developer. On this page that stops being housekeeping: a control chart is a record of one thing changing at a time, and a contaminated developer is several things changing at once in a way the chart will faithfully record and you will not be able to interpret.

Weeks of small exposures. The point of the method is that it runs for a month. That means many short handling episodes rather than one long one, and the cumulative skin contact is the reason the gloves go on every time rather than most times.

A month of stored chemistry. Part-used bottles of developer sitting on a shelf are the subject of one arm of this experiment and are also a storage question: labelled, dated, capped, and kept where they cannot be mistaken for something else, per the container SOP.

Required PPE

Nitrile gloves and eye protection whenever solutions are handled, as the Level A controls require — and on this page, for the aged developer as much as for the fresh, because the course has no evidence on which to relax anything for it.

Nothing else is specified. No respiratory protection, because nothing is weighed or sprayed; no splash goggles beyond the Level A pair, because nothing exceeds a working dilution. The glove page is the authority on what the published permeation data support.

Ventilation

General ventilation with a through draught, the standard on Part XIII’s lab. Nothing on this page produces a vapour that changes that. The practical wrinkle is the same one the calibration page has: the loading is done in darkness, and a light-tight space is often an airless one, so ventilate between loadings rather than during them.

Materials

Item	Quantity	Note
Film, one emulsion batch	about 23 strips of 135 mm: twenty for the pre-exposed batch and three unexposed controls for the first three runs	About two 36-exposure cassettes, or the equivalent from a bulk tin — from one emulsion batch, because a batch change is a chart discontinuity and the crossover procedure below is what handles it
The 21-step wedge	1	The same physical piece, serial recorded
Opaque card	a strip 15 × 40 mm	Masks the tail of every strip, so each carries its own base-plus-fog patch
A freezer bag or a small airtight tin	1	For the pre-exposed batch
Adhesive labels and a pencil	—	Batch code, strip number and exposure date on every strip’s sleeve
Negative filing sheets	4 to 6	The processed strips are the archive and are re-read later

Chemicals

Nothing is weighed. Everything is diluted from a stock or a concentrate per the stock SOP.

Chemical	Quantity	Form
D-76 or ID-11 stock, from Part VIII’s lab — metol, hydroquinone, sodium sulfite, borax	about 2 L over the whole month	Solution. Two bottles: one full and tightly capped, one deliberately half-filled, both dated
Potassium bromide	none weighed	Present only as the bromide the film itself releases into the developer. It is on this list because it is the agent of one of the four signatures, not because you add any
Stop bath, citric acid type	about 2 L at 1+19	Solution
Rapid fixer, ammonium thiosulfate type, or a sodium thiosulfate fixer	about 2 L at 1+4	Fixed for twice the clearing time; ILFORD’s rule is that the bath is discarded when the clearing time in used fixer exceeds twice that in fresh
Wetting agent	about 2 L at 1+200	Final rinse

Equipment

The sensitometer, with its certificate, and the wedge. The wet bench of Part XIII’s lab: tank, graduates, a thermometer reading to 0.1 °C checked by the balance and thermometer SOP, and a water bath — which on this page stops being a nicety, because Kodak ask for the developer to hold to ±0.3 °C and a tank standing in room air will not.

A reading method for density, from Part XIII’s three, used consistently for the whole month. Changing reading method halfway through a control chart puts a step in the chart that has nothing to do with the process. Part XV’s densitometer will replace all three and its arrival is itself a crossover event.

Squared paper or a spreadsheet, ruled up before the first run rather than after the fourth.

Estimated cost

Cost band £. One cassette of film makes the whole batch of control strips, and each strip after that rides in a tank you were filling anyway. That is the economic argument for the method and it is a strong one: process control costs about one strip and five minutes per film you were already developing.

Estimated consumables cost

Costed for the set-up session — pre-exposing the batch and the first three runs that establish the aim — because that is what one run of this page is.

Film is counted at twelve 135 mm strips to a 36-exposure cassette, as the enclosure build sets out, and the batch here is large enough that the figure is worth checking against your own bulk loader before you cut.

Consumed	This session	Sourced price	Cost this session
Film, one emulsion batch	23 strips of 135 mm — twenty pre-exposed plus three controls — about two 36-exposure cassettes	£6.37–£11.40 per one 35 mm roll, 36 exposures	£12.23–£21.89
D-76 or ID-11 stock, from Part VIII	900 mL over three tanks at 1+1, discarded after each	Costed in Part VIII’s mixing lab	—
Rapid fixer concentrate	180 mL, to make 900 mL at 1+4	£21.05–£25.98 per 1 L of ammonium thiosulfate concentrate, diluted 1+4 for film	£3.79–£4.68
Stop bath concentrate	45 mL, to make 900 mL at 1+19	£10.66–£12.18 per 500 mL of citric acid concentrate, diluted 1+19	£0.96–£1.10
Wetting agent	4.5 mL, for 900 mL at 1+200	£28.70 per 1 L of concentrate, diluted 1+200	£0.13
Negative filing sheets and labels	6 pockets, 20 labels	None. A named price gap: sleeving that passes the Photographic Activity Test	—
Freezer bag or small tin	1	None. The planner carries no line for it	—

The priced rows come to £17.11 to £27.80 for the set-up session, at the ranges dated 5 September 2026 in the laboratory planner. That is a floor and not a total: two of the seven rows carry no dated price and are counted as nothing here.

Each monitoring run thereafter consumes two strips — the control strip and the unexposed strip beside it, about £1.06 to £1.90 of film — and no extra chemistry at all, because both go into a tank that was already being filled for a film. That is the number worth remembering, and it is why laboratories do this and darkrooms usually do not.

Waste streams

The three every processing page produces, kept apart, labelled per the container SOP and routed per the general waste SOP: alkaline spent developer carrying the developing agents nearly unconsumed; spent fixer and its first rinse, which are silver-bearing and go to the silver stream; and the rest of the wash water.

The aged-developer arm produces one extra bottle of spent developer at the end of the month. It goes into the same stream as the rest.

Alternative route

The loading needs darkness. A changing bag is enough, as it is everywhere else in this part, and on this page it is more than enough because the strips are short and there is one exposure per session rather than fourteen.

The route for a reader with no sensitometer at all is worth stating, because the method is more valuable than the instrument. Pre-expose the batch under an enlarger by Part XIII’s method, in one sitting, with the head height, aperture and time recorded — and then never move the enlarger’s setting again for that batch. That gives you strips that are identical to each other, which is all a control chart needs. What it does not give you is strips identical to the next batch, so the crossover procedure becomes compulsory rather than advisable, and the exposure carries the reciprocity error the design lesson described, which is a constant offset and therefore does not disturb a chart of changes.

What has no alternative is the discipline. A control strip processed when you remember is not a control strip; it is a souvenir.

Preparation

Pre-exposing the batch

Do it in one sitting, from one cassette, with the instrument warmed up and its exposure logged.

1. Cut about twenty strips, each long enough to carry the whole wedge, from one emulsion batch.
2. Expose every one of them at the same setting, through the same wedge, in the same orientation, with the tail masked so each carries its own base-plus-fog patch.
3. Number them as you go and record the number, the exposure log line and the date.
4. Seal them, cold, dark and dry.

Kodak’s instruction for their own strips is one line and it is the important one: keep control strips frozen to maintain consistency. That is a manufacturer telling you that the pre-exposed latent image is a perishable thing, which is exactly what Part IV’s latent-image page explains and what Part IX named as its one unavoidable compromise.

!What latent-image keeping does to this method, said plainly

A control chart works by holding the exposure constant and watching the process. A pre-exposed batch does not hold the exposure constant: strip 1 is developed a day after exposure and strip 20 perhaps five weeks after, and latent-image keeping works on every one of them in between.

The consequence is specific and it is the worst possible one. If the latent image fades slowly, it shows on the chart as a gradual downward trend in speed — which is precisely the signature of a developer being exhausted or under-replenished. The method’s most useful diagnosis and its own worst artefact look identical.

Four things address it, and none of them is a fix.

Freeze the batch, on Kodak’s own instruction for their strips.

Keep the batch short. Twenty strips at two a week is ten weeks; twenty at four a week is five. Prefer the shorter run and make a new batch.

Run the crossover. Kodak’s procedure for changing to a new batch is three paired runs — one strip from each batch in each of three separate runs — averaging each set of three, taking the difference, and applying that difference to the aims. Do exactly that, and a trend that disappears at the crossover was your strips, not your developer. That single observation is what turns the confound into a diagnosis.

And Part IX’s structural mitigation stands unchanged: the shared reference strip in every session, from a batch whose exposure date is written down, so that drift is measured rather than assumed.

The honest statement for the certificate and the notebook: this method detects changes faster than it detects slow trends, and a slow downward trend in speed is ambiguous until a crossover resolves it.

Choosing the three numbers

Kodak’s own strips carry five steps — D-min, a toe density, a low density, a high density and D-max — and they compute contrast index, a speed value and D-min from them. Your strip has twenty-one steps, so you may choose, and the choice is worth making deliberately.

What you plot	Read it from	What it is sensitive to
Fog, D-min	The masked tail of the strip	Film age and storage, safelight and light leaks, developer age, contamination, over-development
Speed, as D(shadow step) − D-min	A step near the foot of your own curve — the one nearest the criterion point from Part XIII	The activity of the developer: bromide build-up, dilution, exhaustion, and latent-image keeping
Contrast, as (D(high) − D(shadow)) ÷ Δlog H between them	A step high on the straight line, and the same shadow step	Degree of development: time, temperature, agitation, and the sulfite and agents running out

✕Do not copy Kodak's aims, their limits or their formula

Kodak print an aim of contrast index 0.58 for printing with a diffusion enlarger and 0.43 for a condenser, a speed-value aim of 355, and a D-min aim of 0.06, with action limits of +0.07 to +0.20 and −0.07 to −0.12 on contrast index, +15 to +22 and −9 to −17 on speed, and +0.02 to +0.03 on D-min with no lower limit at all.

None of those numbers transfers to your strip. They belong to a specific pre-exposed product on T-MAX 400, read at five specific steps, put through Kodak’s own arithmetic — their speed value is an arbitrary number relating only to their control strip, and they say so themselves. Your strip is a different film, a different exposure and twenty-one steps. What transfers is the shape of the scheme: an aim, a warning band, an action band, and the discipline of plotting the deviation from aim rather than the raw number.

One further note, recorded because this course reports what its sources actually say. The mirrored copy of Z-133E prints its short contrast-index formula with different terms in different places — as (HD − TD) ÷ 2.26 + 0.10 twice, once where the procedure is first given and again on its own Worksheet 1, and as (HD − LD) ÷ 2.26 + 0.10 in three others. The course cannot tell which is the erratum from the document alone, so it uses neither and defines contrast from its own wedge, as Part XIII’s contrast-index convention already does. Kodak’s construction is fine; the discrepancy in this copy is a warning against copying a formula out of a scan without checking it against itself.

Procedure

Part 1 — Establish the aim (the first three runs)

Process one control strip in each of your next three ordinary film runs, at your standard condition: developer at your dilution, 20.0 °C held in the water bath, your written agitation script, and an unexposed strip alongside. Read the three numbers on each.

The aim is the mean of those three. Write it down and do not adjust it again except at a crossover.

🧮Setting your own limits from your own certificate

This is where the calibration experiment pays for itself. You have a measured session-to-session reproducibility, in log H, and a measured reading floor, in density. Convert the first into density at the step you are plotting, using the local gradient:

Δ D = Δ log H × G

and add the reading floor as a bound, on Part II’s rule. Call the result s: the ordinary noise of your own system, in the units of the thing you are plotting.

Then the course’s convention, adopted because it is arithmetic anybody can check:

• Action limit at 2s. Beyond this, investigate before the next run.
• Control limit at 3s. Beyond this, stop and find the cause before you process anything you care about.

Two sanity checks on the result. If your limits come out so wide that nothing could ever exceed them, your s is dominated by your reading method and the honest response is a better one, not a narrower limit. If they come out so tight that half the points fall outside, either the process really is out of control or the reproducibility figure on your certificate was optimistic — and the certificate is the thing to re-examine first, because it was measured over three days and the chart runs over five weeks.

Part 2 — Monitor (one strip per run, for four to six weeks)

One control strip in every tank, every time, with an unexposed strip alongside. Read the three numbers, compute the deviation from aim, plot it, and write one line in the notebook saying what else was true that day: which bottle, how old, what temperature the bath actually held, whether the agitation was the usual script, how many films had already been through that developer.

The chart, ruled up: three rows, one date axis, four annotations

1. 1Speed row — D(shadow step) − D-min, as a deviation from aim
2. 2Contrast row — (D(high) − D(shadow)) ÷ Δ log H, as a deviation from aim
3. 3Fog row — D-min, deviation from aim; no lower limit, because less fog is not a fault
4. 4Event rules — fresh mix, batch crossover, new fixer, a warm day — every one written on the chart, not in a separate note

Ruled up before the first run. A chart drawn after the fourth point is a chart whose aim was chosen to fit the data.

Read the chart the way Kodak read theirs, because the rules are about statistics rather than about photography and they transfer intact.

• Random variation is normal and is what a process in control looks like: points scattered on both sides of aim, inside the limits, in no pattern. Kodak add the warning that matters most and is the hardest to follow: avoid over-controlling a process by reacting to random variation. Two points low is not a trend.
• An outlier is a point on or beyond the control limit. Kodak’s first instruction is not to change the process: it is to check the measurement — re-read the strip, then process another — before concluding anything.
• A level shift is four to eight consecutive points on one side of aim. Two or three may be noise; a fourth makes a shift likely and each further point raises the odds.
• A trend is four to eight points climbing or falling in a row.
• And one trend is not a fault at all. Kodak record that some developer systems are formulated so that a fresh solution gives a higher speed than a properly seasoned one, and name Developer D-76 with Replenisher D-76R among them: for those, a declining speed after a fresh mix is normal seasoning and stops when the developer is fully seasoned. Know which system you are running before you chase your own tail.

Part 3 — The three deliberate perturbations

Do these at the end, when the chart has a settled aim and at least eight ordinary points, and write the prediction down before each one.

Run	The one change	Predict before you process
P1	Developer at 22.0 °C instead of 20.0	Direction and rough size of the contrast movement, and whether speed moves too
P2	Development time 20 per cent longer, at 20.0 °C	The same two, and explicitly whether you expect P1 and P2 to land in the same place
P3	Developer from the half-filled bottle, kept past its stated life	Direction of all three numbers, and which of the three moves first

🧮Working the prediction for P1 and P2 before you run them

Part II read ILFORD’s published time-temperature chart and got the number this prediction turns on: a 1 °C error is worth about 9 per cent of development time. So

P1, at +2 °C, is worth about +18 per cent of development. P2, at +20 per cent of time, is worth +20 per cent of development.

They should land in nearly the same place, and that equivalence is the finding, not the size. It is also the argument for the water bath: being 2 °C out is an ordinary morning’s mistake, and being 20 per cent out on a timer is not something you would do by accident.

How big is “18 per cent of development” on the chart? The course will not print a density figure it has not measured. What it can give you is a sourced bracket: Kodak adjust development time in 10 per cent increments when they are moving contrast index towards aim, and they accept a strip when it is within ±0.02 of the aim. So a manufacturer treats 10 per cent of time as a step worth taking and 0.02 of contrast index as the width of “close enough” — which puts 18 to 20 per cent comfortably outside that window and therefore visible on any chart whose limits were honestly set. Your own number is the measurement. Write the prediction as a direction and a bracket, and let the strip settle the size.

Part 4 — Separating the instrument from the process

Once, in the middle of the run, do this: expose a fresh strip from the same cassette on the day of a monitoring run, and process it alongside the pre-exposed one from the batch.

Three outcomes and three readings.

They agree. The instrument has not moved and the batch has not faded appreciably since it was exposed. Everything on the chart is process.

The fresh strip is denser. The batch has faded, or the instrument has got brighter. The certificate settles it: if the movement is larger than the instrument’s reproducibility figure, the instrument alone cannot account for it, and latent-image keeping is the remaining explanation.

The fresh strip is thinner. The instrument has dimmed — check the warm-up rule was followed and the monitor log — or something in the box has changed. Re-run the enclosure page’s uniformity map before you touch the chart.

Expected observations

Within the first three runs, the three numbers should already differ from each other by less than you expected and by more than nothing. That is what “in control” feels like.

Over four to six weeks the chart should look boring for most of its length, with two or three interruptions that line up with events you annotated. A boring chart is the result. The month in which nothing happened is what makes the month in which something does interpretable.

The perturbation runs should be unmistakable — points outside the action band, not marginal — and P1 and P2 should land close together. If they do not, the discrepancy is worth more than the agreement would have been: check that the bath really held 22.0 °C for the whole development rather than starting there, because a tank cooling towards room temperature spends most of its time somewhere else.

What is happening chemically

Four causes, four signatures. This is the section that turns a chart into a diagnosis, and every mechanism in it is owned by an earlier page and cited rather than re-derived here.

Bromide accumulation. Every film developed releases halide into the developer. ILFORD say it in one sentence — as each film is processed it releases halides and other by-products that act as a restrainer on the development of subsequent films — and then quantify their own remedy: increase the development time by 10 per cent for each successive film through a litre of stock ID-11, up to ten films. Part VIII owns the mechanism: bromide suppresses development, and it suppresses it more where there is least silver to develop.

Signature: speed down, fog down, contrast down last and least. Fog falling as speed falls is the fingerprint — a restrainer restrains the unexposed grains hardest of all, which is why bromide is in developers on purpose.

Aerial oxidation. A developer in a part-filled bottle is a developer in contact with air, and the manufacturers price it exactly. Kodak: D-76 stock keeps six months in a full tightly closed bottle and two months half-filled, and they explain why in one clause — partially filled bottles allow some oxidation. ILFORD are stricter: ID-11 stock keeps six months in a full capped container and one month in a half full one, four months in a deep tank with a floating lid and one month without. Part VIII owns sulfite’s role and Part VIII’s aerial-oxidation experiment is where the course measured it.

Signature: contrast down, speed down with it, fog roughly unchanged or slightly down. Kodak’s own troubleshooting chart carries the first two of those against the cause “developer or replenisher too old or oxidised”: contrast index and speed both trending down. The fog half of the sentence is the course’s inference from the mechanism rather than a manufacturer’s figure, and it is the weakest line in this section — which is why the confirming test below is the bottle’s date and headspace rather than the fog row.

Temperature error. Development is a rate process, so a warmer bath develops further in the same time. ILFORD’s chart makes 1 °C worth about 9 per cent of time, and Kodak state flatly that variations greater than ±0.3 °C in the developer will affect process control and image quality.

Signature: contrast moves first and furthest, speed follows, fog moves only for large errors. The reason contrast leads is that contrast is the degree of development, which is what temperature changes, whereas speed is set mostly by where the toe sits and moves more slowly with development.

Contamination and carry-over. ILFORD warn a beginner that a trace of fixer can contaminate the developer, and Kodak list solution contamination among the causes of an out-of-control process along with its routes: unclean mixing equipment, dry chemicals going airborne during mixing and settling in an adjacent solution, and splashing.

Signature: unpredictable, which is itself the signature. And here the page states a limit rather than a rule. The course has found no manufacturer measurement of what a stated quantity of thiosulfate in a developer does to a characteristic curve, so it will not tell you how far fog rises per millilitre of carry-over. What it can tell you is what the sources do say: that contamination is a named cause, that it is prevented by separate utensils, and that a stain has its own diagnosis — Kodak note that a severe magenta stain after fixing points at a fixer near exhaustion or an inadequate fixing time, which is a fixing fault and not a developer fault at all. When a chart moves in a way none of the first three signatures explains, contamination is the hypothesis, and the test is a fresh set of solutions in clean vessels rather than a further reading of the chart.

The four signatures, drawn as they appear on a chart

• Bromide accumulation — a steady trend
• Aerial oxidation — flat, then accelerating
• A 2 °C error on one run — an outlier
• A new bottle — a level shift

Show the numbers behind this plot

Four traces over twelve runs, plotted against a vertical axis of deviation from aim expressed in units of the action limit, so that plus or minus one is the action limit and plus or minus one and a half the control limit. The first trace, bromide accumulation, falls steadily and almost linearly from zero at run one to about minus one and a quarter by run twelve, a classic trend. The second, aerial oxidation, holds near zero for the first five runs and then falls away increasingly steeply, reaching about minus one and a half by run twelve, because oxidation in a part-filled bottle accelerates once the sulfite reserve is depleted. The third, a temperature error on a single run, sits on the aim line throughout except for run seven, where it jumps to plus one and three quarters and returns to the line immediately afterwards: a single outlier. The fourth, a level shift, sits scattered about the aim line for the first six runs and then steps abruptly to about minus three quarters and stays there, inside the action limit but consistently on one side of aim, which is the pattern four to eight consecutive points on one side of the line is meant to catch. Two horizontal guide lines mark the action limits at plus and minus one.

Series	Runs, in order	Deviation from aim, in units of the action limit
Bromide accumulation — a steady trend	1.00	0.00
Bromide accumulation — a steady trend	2.00	-0.10
Bromide accumulation — a steady trend	3.00	-0.20
Bromide accumulation — a steady trend	4.00	-0.35
Bromide accumulation — a steady trend	5.00	-0.45
Bromide accumulation — a steady trend	6.00	-0.60
Bromide accumulation — a steady trend	7.00	-0.70
Bromide accumulation — a steady trend	8.00	-0.85
Bromide accumulation — a steady trend	9.00	-0.95
Bromide accumulation — a steady trend	10.00	-1.05
Bromide accumulation — a steady trend	11.00	-1.15
Bromide accumulation — a steady trend	12.00	-1.25
Aerial oxidation — flat, then accelerating	1.00	0.05
Aerial oxidation — flat, then accelerating	2.00	-0.02
Aerial oxidation — flat, then accelerating	3.00	0.03
Aerial oxidation — flat, then accelerating	4.00	-0.05
Aerial oxidation — flat, then accelerating	5.00	-0.10
Aerial oxidation — flat, then accelerating	6.00	-0.25
Aerial oxidation — flat, then accelerating	7.00	-0.45
Aerial oxidation — flat, then accelerating	8.00	-0.70
Aerial oxidation — flat, then accelerating	9.00	-0.95
Aerial oxidation — flat, then accelerating	10.00	-1.15
Aerial oxidation — flat, then accelerating	11.00	-1.35
Aerial oxidation — flat, then accelerating	12.00	-1.50
A 2 °C error on one run — an outlier	1.00	0.05
A 2 °C error on one run — an outlier	2.00	-0.05
A 2 °C error on one run — an outlier	3.00	0.10
A 2 °C error on one run — an outlier	4.00	0.00
A 2 °C error on one run — an outlier	5.00	-0.08
A 2 °C error on one run — an outlier	6.00	0.06
A 2 °C error on one run — an outlier	7.00	1.75
A 2 °C error on one run — an outlier	8.00	0.02
A 2 °C error on one run — an outlier	9.00	-0.06
A 2 °C error on one run — an outlier	10.00	0.08
A 2 °C error on one run — an outlier	11.00	-0.02
A 2 °C error on one run — an outlier	12.00	0.04
A new bottle — a level shift	1.00	0.08
A new bottle — a level shift	2.00	-0.06
A new bottle — a level shift	3.00	0.05
A new bottle — a level shift	4.00	-0.10
A new bottle — a level shift	5.00	0.09
A new bottle — a level shift	6.00	-0.04
A new bottle — a level shift	7.00	-0.72
A new bottle — a level shift	8.00	-0.78
A new bottle — a level shift	9.00	-0.70
A new bottle — a level shift	10.00	-0.80
A new bottle — a level shift	11.00	-0.74
A new bottle — a level shift	12.00	-0.76

Drawn to teach the four shapes, not measured from any process: the vertical scale is deliberately in units of your own action limit, because the limit is the only thing that makes a deviation large or small. What is real in the drawing is the shapes and the order in which they are distinguishable — an outlier in one run, a shift in four, a trend in six. 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.

Three strategies for holding the developer still, and what the chart says about each

The four signatures above are what happens when a developer is allowed to change. There are three standard ways of not allowing it, and each one produces a differently shaped chart, so knowing which you are running is a prerequisite for reading your own plot.

One-shot. Dilute from stock immediately before use and throw the working bath away afterwards. Kodak are explicit for D-76: the 1:1 dilution is mixed just before use, discarded after one batch, and neither reused nor replenished. ILFORD say the same for ID-11 at 1+1 and 1+3 and add the reason to prefer it — one-shot processing is recommended where image quality, reliability and consistency matter more than economy, and it eliminates or greatly reduces the oxidation, contamination and precipitate problems that come with reuse.

On the chart, one-shot removes both of the working-bath signatures at a stroke: no bromide can accumulate in a bath that is used once, and no bath sits in air long enough to oxidise. What remains is the stock ageing in its bottle, which is far slower, so a one-shot chart should be flat with no trend at all — which makes any trend on it a real signal rather than a bookkeeping artefact. That is why this course processes one-shot on every sensitometric page, and it is the single largest thing you can do to make a control chart interpretable.

Reuse with time compensation. ILFORD publish the arithmetic: ten 135/36 films per litre of stock ID-11, with the development time increased 10 per cent for each successive film, the used developer poured back and mixed with the unused part before the next film. Kodak give D-76 at full strength a useful capacity of 16 rolls per gallon — four per litre — with a 15 per cent time increase after every four rolls per gallon. ILFORD also say what the compensation is worth: it can only be an approximation to cover a range of circumstances, because a roll of night shots consumes almost no developing agent and a roll of beach scenes consumes a great deal.

On the chart, this is the case the method was made for. The compensation is a guess and the chart measures whether the guess was right: apply the 10 per cent, and if speed still trends downwards, the compensation is too small for your subjects — which is exactly what ILFORD’s caveat predicts and what no published table can tell you.

Replenishment. Top the bath up with a replenisher that restores what was consumed. Kodak state that replenishing D-76 with Replenisher D-76R extends the capacity to 120 8 × 10-inch sheets per gallon and that development times are not increased when you do. The route is not open for every developer: HARMAN discontinued the 2.5 litre ILFORD ID-11 Replenisher pack in September 2012, with ID-11 itself unaffected, so an ID-11 user in the United Kingdom has one-shot and compensated reuse and not this.

On the chart, a replenished system has its own normal shape and it looks alarming until you know it. Kodak record that some systems are formulated so a fresh solution gives a higher speed than a properly seasoned one, and name D-76 with D-76R among them: for those, a declining speed after a fresh mix is normal seasoning and stops of its own accord when the developer is fully seasoned. Chasing it with replenishment adjustments is over-controlling, which is the error Kodak warn about most.

And how the chart says a bath is finished. Not by a date and not by a film count, but by this: a point beyond the control limit, confirmed by a second strip, that returns to aim when the bath is replaced and nothing else is changed. That last clause is the whole of it. Replacing the developer and the fixer and the stop bath together on the same afternoon tells you that something was wrong; replacing one of them tells you which.

Carrying the method to paper, and to your own emulsion

The instrument, the wedge and the chart do not care what is being processed, so the same three numbers work anywhere the course processes anything — and the two places they will matter most are both ahead of you.

Paper. ILFORD note that Multigrade RC papers are roughly equivalent to a film ISO of 3 to 6, which is a factor of about a hundred slower than a 400 film, so the same instrument makes a paper control strip by running the lamp at a higher current or giving a longer exposure — the design lesson’s factor of 130 between film and paper is precisely this. Track maximum black rather than a shadow step, a mid-grey step, and the paper base. One chemical difference changes how the chart behaves: a print developer works on far more silver per unit area than a film developer does, so it exhausts on the work rather than on the air, and its chart trends down much faster and much more predictably.

Your own emulsion, later in the course. The moment you coat your own material, every make is a new batch, and the first question after every disappointing print is “was it the coating or the developer?”. A control strip from a reference batch, processed alongside every new make, answers it in one strip: if the reference is on aim, the process is fine and the make is the variable. That is the crossover idea of this page applied to the material instead of to the strips, and it is the reason the archive of control strips is worth keeping long after the chart has been drawn.

Data to record

In the lab notebook, one page per run, and the chart itself kept with it.

Per run. Date and time; run number; the control strip’s batch code and strip number, and its exposure date; the developer’s version code from the formula version record; its dilution; whether it is fresh, reused or replenished, and how many films have been through it; measured temperature at start, middle and end; the agitation script in words; the fixer’s age and its measured clearing time on a scrap.

Per strip. D-min from the masked tail; the shadow step’s density; the high step’s density; the three derived numbers and their deviations from aim; and the reading method used.

Per event. Anything that could plausibly have changed the process, written on the chart at the date it happened: a fresh mix, a new bottle, a batch crossover, a new box of fixer, a hot afternoon, a different tank, a different water supply.

The version code is the join between this chart and everything else you record. Part IX’s laboratory report owns the versioning scheme — stem, initials, sequence — and a control chart whose points cannot be tied back to a specific mixed batch of a specific formula is a chart that can show you a change and never let you find its cause.

Analysis

1. Is the process in control? Count the points inside the action band, the runs of consecutive points on one side of aim, and any monotone runs. Apply the three patterns in order: outliers first, then level shifts, then trends. Write a one-line verdict for each of the three plotted numbers.

2. Read the pattern across all three rows, not one at a time. This is where the diagnosis lives, and it is a table rather than a paragraph.

Speed	Contrast	Fog	Most likely cause	The confirming test
Down, trend	Down, later and less	Down	Bromide accumulation from reuse	Count the films through that bath. Mix fresh and re-run one strip: if it returns to aim, the developer was seasoned, not spoiled
Down, accelerating	Down, accelerating	Flat or slightly down	Aerial oxidation in a part-filled bottle	Check the bottle’s date and headspace against the maker’s stated life. Decant to a full smaller bottle and re-run
Follows contrast	Up, single outlier	Flat	Temperature high on that run	Look at the recorded start, middle and end temperatures. A bath that started warm and cooled shows as a smaller effect than the start temperature predicts
Flat	Flat	Up	Fog from somewhere that is not the developer: a light leak, a safelight, film age, or the strips themselves	Process a strip from a fresh exposure and one that has never been exposed at all. A rise on the never-exposed strip is film or safelight; a rise only on the batch strips is the batch
Down	Down	Up	Contamination is the hypothesis, because no single clean cause moves all three that way	Fresh solutions in clean vessels, new tank, one strip. This is the one case where the answer is to stop reading the chart
Step down, then steady	Step down, then steady	Step	A level shift at an event: new bottle, new batch of strips, new reading method	Look at your own annotations before you look at the chemistry. A crossover that was not run is the commonest cause

3. The perturbations. Tabulate prediction against result for P1, P2 and P3, and answer the specific question: did the +2 °C run and the +20 per cent run land in the same place? Compute the ratio of their deviations. If it is near 1, ILFORD’s chart has been confirmed on your own film and developer, which is a real result and belongs in the notebook as one.

4. What this chart cannot detect, said as plainly as what it can. A control chart is a measurement of one exposure through one process, and everything outside that is invisible to it.

• It cannot detect a change in your camera, your metering or your exposure. The strip never goes through a camera.
• It cannot detect a change in the wedge, because the wedge is on both sides of every comparison. A scratched or fingerprinted wedge shifts every point equally and the chart stays flat.
• It cannot separate a slow fade of the latent image from a slow exhaustion of the developer, which is the compromise stated at the top of this page, and only a crossover resolves it.
• It cannot tell you that your aim is right, only that you are still where you were. A process held perfectly in control at the wrong development level produces consistently wrong negatives, and Kodak make the same point from the other direction when they warn that correcting a long-standing out-of-control process may surprise the customers who had adapted to it.
• It cannot see anything after fixing. Washing, drying and permanence leave no mark on these three numbers. Part XII owns that and has its own tests.
• And it detects a change, never a cause. The table above is a set of hypotheses ranked by likelihood. The confirming test is what turns one of them into an answer.

Troubleshooting

What you see	Likely cause	What to do
Every point in the first three runs is wildly different	The aim has not settled because something else is unstable: temperature, agitation script, or the reading method	Do not set the aim yet. Fix the unstable thing first — the water bath is the usual answer — and start the three runs again
The chart is flat and the negatives are still wrong	The process is in control at the wrong level	The chart is doing its job. Change the development time deliberately, in Kodak’s 10 per cent increments, and re-establish the aim at the new level
A trend in speed that ends exactly at a batch crossover	The strips, not the developer. This is the latent-image confound announcing itself	Believe the crossover. Apply the batch difference to the aims, note it on the chart, and shorten the next batch
Points scatter far more than your limits allow, with no pattern	Limits set from an optimistic reproducibility figure, or a reading floor larger than you assumed	Re-measure the reading floor: read one dense step five times and take the spread. If it is comparable with your limits, the reading method is the problem, not the process
Fog climbing on the control strips but not on ordinary films	The pre-exposed batch is ageing, or its storage has failed	Check the freezer. Then process a never-exposed strip from the same cassette: fog on it too means the film, fog only on the exposed ones means the latent image and the batch
A magenta or pink stain on the strips after fixing	Kodak name this one: a fixer near exhaustion, or an inadequate fixing time	Re-fix a stained strip in fresh fixer. Then run ILFORD’s clearing-time test on a scrap and discard the bath if the clearing time exceeds twice that in fresh
The perturbation runs land inside the limits	The limits are too wide, or the perturbation was smaller than intended	Check the recorded bath temperature over the whole development, not at the start. Then re-examine the reproducibility figure the limits came from
The chart has gaps	The method has stopped, whatever the notebook says	Four points a month with no gaps beats twelve with three. Reduce the frequency to one you will actually keep

Clean-up

Solutions to their labelled containers. The half-filled bottle from arm P3 goes to waste at the end of the run and is not kept for a second month, because a bottle whose history is uncertain will contaminate the next chart rather than the last one. Rinse and dry the tank and reels completely. Wipe the wedge with a dry lens cloth and sleeve it. Close the laboratory to the SOP.

Storage

The unexposed batch stays frozen, sealed, labelled with its exposure date and its strip numbers, on Kodak’s own instruction for their strips. Let a strip come to room temperature in its bag before it is opened, or condensation will do to it what a month of ageing could not.

The developed strips are the archive. Sleeved, labelled with run number, date, batch code, developer version and the three readings, and kept in run order. Part XXVII will re-read these on a densitometer, and this chart will be re-drawn from measured densities rather than from your reading method’s floor.

The chart itself is a record and gets a version and a date like everything else. Keep it with the notebook and photograph it at each month’s end, because a chart that exists on one sheet of squared paper is a chart that will be lost.

Disposal considerations

The three streams, unchanged from every other processing page: alkaline spent developer carrying its agents nearly unconsumed; silver-bearing spent fixer and its first rinse, to the silver stream; and rinse water. The chemistry and the general practice are on the disposal page.

Local regulation governs what may lawfully be done with any of it, and it differs by jurisdiction — check your local regulations. The course describes chemistry and general practice and gives no jurisdiction-specific instruction as though it were universal.

Questions

1. Your chart shows six consecutive points below aim in speed, all inside the action limit, with contrast flat and fog flat. Name the pattern, say whether Kodak’s rules call it a fault, and give the two hypotheses that fit and the one test that separates them.
2. A 1 °C error is worth about 9 per cent of development time. Your process is 11 minutes at 20 °C. Compute the time you would have to give at 22 °C to develop to the same degree, and then say why that arithmetic does not make a water bath unnecessary.
3. You mix a fresh litre of D-76 stock and the speed row trends downwards over the next five runs before flattening. Give the reading that says this is normal, name the source, and say what you would do differently if you were running an ILFORD ID-11 stock instead.
4. Fog is rising by 0.01 a run over four runs while speed and contrast are flat. Work through the table in the Analysis section and give the order in which you would test the causes, with the reason each test comes where it does.
5. Explain why a scratched wedge is invisible to a control chart but fatal to an absolute speed measurement, and say which page in this part is responsible for catching it.
6. A friend keeps a control chart with limits set at ±0.05 in density “because that seemed sensible”. Explain what is wrong with that as a method, and set out the three measurements you would need in order to set the limits properly.

Further experiments

Run the latent-image fade series the method needs. Expose thirty strips in one sitting, and develop one on day 0, 3, 7, 14, 28 and 56 in the same developer at the same condition, with an unexposed control each time. Plot the three numbers against days since exposure. That is the missing calibration of this whole method: it converts an acknowledged confound into a correction you can apply, and no manufacturer publishes it for the film you are using.

Measure the bromide effect directly. Instead of waiting for a bath to season, run ILFORD’s own prediction as an experiment: process one control strip in fresh stock, then push five films through the same litre and process another, then five more and a third. ILFORD’s remedy implies about +10 per cent of time per film; your chart says what the untreated deviation actually is, and the two together let you check whether their compensation is right for your film.

Chart a paper process. ILFORD note that Multigrade RC papers are roughly equivalent to a film ISO of 3 to 6, so the same wedge and the same instrument make a paper control strip with a much longer exposure. Track maximum black, a mid-grey step and the paper base. The chemistry is different in one important way — a paper developer exhausts on the silver it reduces far faster than a film developer does — and the chart will show you that difference rather than being told it.

Carry the method into the emulsion laboratory. When you coat your own emulsion, every make is a new batch and the question “was it the coating or the developer?” arrives immediately. A control strip from a reference batch, processed alongside every new make, answers it in one strip — and it is the same idea as the crossover, applied to the material rather than to the strips.

Check your understanding

Question 1. Your speed row trends steadily downwards over eight runs. At run nine you change to a new batch of control strips, run the crossover properly, and the trend disappears. What have you learned?

• The developer was exhausting, and the fresh strips masked it
• The trend was latent-image fade in the old batch, not a change in the process
• The crossover was done wrongly, since a correct one changes nothing
• Nothing, because a batch change makes the two halves of the chart incomparable

Show the answer and why

Answer: The trend was latent-image fade in the old batch, not a change in the process

This is the observation that turns the method's worst confound into a diagnosis. A fading latent image and an exhausting developer produce the same downward trend in speed, which is why the page states the ambiguity at the top. A batch change replaces the strips and not the developer — so a trend that survives the crossover belongs to the process, and one that vanishes belonged to the strips. Option 4 has the crossover backwards: its whole purpose is to make the two halves comparable, by measuring the difference between the batches over three paired runs and applying it to the aims.

Question 2. Your process is 11 minutes at 20.0 °C, and the bath is actually at 22.0 °C. Taking 1 °C as worth about 9 per cent of development time, what development time at 22 °C would give the same degree of development, and what does that tell you about the +20 per cent time perturbation?

• About 9 min 15 s, and it means the temperature error is much larger than the time error
• About 9 min 20 s, and it means the two perturbations should land in nearly the same place
• About 13 min, and it means the temperature error works in the opposite direction
• About 10 min 45 s, and the two are unrelated

Show the answer and why

Answer: About 9 min 20 s, and it means the two perturbations should land in nearly the same place

Two degrees is worth about 18 per cent of development, so at 22 °C you need roughly 11 ÷ 1.18 = 9.3 minutes, about 9 min 20 s, to reach the same degree of development. Run at the full 11 minutes instead and you have over-developed by about 18 per cent — which is within a couple of points of the 20 per cent that P2 gives deliberately, so the two perturbations should land close together. That equivalence is the finding. It is also the argument for the water bath, because being 2 °C out is an ordinary mistake and being 20 per cent out on a timer is not.

Question 3. Which pattern across the three plotted numbers points at bromide accumulation from a reused developer, rather than at aerial oxidation?

• Speed down, contrast down, fog UP
• Speed down, contrast down later and less, fog DOWN
• Speed flat, contrast down sharply, fog flat
• Speed up, contrast up, fog up

Show the answer and why

Answer: Speed down, contrast down later and less, fog DOWN

Bromide is a restrainer, and a restrainer acts hardest on the grains with the least exposure — which is exactly what fog is. So fog falling alongside speed is the fingerprint, and it is the thing that distinguishes a seasoned developer from a spoiled one. ILFORD state the mechanism plainly: each film released halides and other by-products that act as a restrainer on the development of subsequent films, and their remedy is 10 per cent more time per successive film. Aerial oxidation is the other pattern in the list of causes: contrast and speed both trending down, accelerating, with fog roughly unchanged, which is the pattern Kodak's own troubleshooting chart carries against a developer too old or oxidised.

Question 4. Why can Kodak's published aims and action limits not be copied onto your own chart?

• Because they are copyright and the course may not reproduce them
• Because they describe a specific pre-exposed product on T-MAX 400 read at five specific steps through Kodak's own arithmetic, and their speed value is an arbitrary number relating only to that strip
• Because Kodak measured them with a densitometer and you are reading by eye
• Because they were published in 2003 and the products have changed

Show the answer and why

Answer: Because they describe a specific pre-exposed product on T-MAX 400 read at five specific steps through Kodak's own arithmetic, and their speed value is an arbitrary number relating only to that strip

Kodak say it themselves: the speed value is an arbitrary number that relates only to the control strip and is not an ISO or ASA speed, and the aims are printed on their own plotting form for their own strips. Your strip is a different film given a different exposure and read at steps you chose from twenty-one. What does transfer is the structure — an aim, a warning band, an action band, deviations plotted rather than raw numbers — and the statistical rules for reading the chart, which are about runs of points and not about photography. Option 3 is a real limitation of your method and a separate one: it changes the size of your limits, not whose limits you may use.

Question 5. A chart has been flat and inside its limits for six weeks, and the negatives coming out of the same process are consistently flat and hard to print. What has the chart told you?

• That the chart is broken and its limits are too wide
• That the process is in control at the wrong development level — which is exactly what a control chart can say and what it cannot
• That the fault is in exposure rather than processing, since the chart is flat
• That the control strips have faded uniformly

Show the answer and why

Answer: That the process is in control at the wrong development level — which is exactly what a control chart can say and what it cannot

A control chart measures repeatability, not correctness. Flat means you are still where you were; it says nothing about whether where you were is where you should be. The remedy is deliberate: change the development time in increments — Kodak use 10 per cent — until the negatives print as you want them to, then re-establish the aim at the new level and start plotting again. Option 3 is a genuine possibility in the real world and it is not something this chart can tell you, because the strip never passes through a camera; that is one of the five things the Analysis section lists as invisible to the method.

Check my answersClear