Versioning a formula

Purpose

To give one mixed batch a name of its own, pointing at a record from which somebody could rebuild the same litre — so that a strip, a plot or a print is attributable to a solution rather than to a recipe.

When to use it

Before anything is weighed, every time a batch is made up from raw chemicals: a developer, a fixer, a stop bath, an emulsion melt. Again whenever an entry in the record would now read differently — a different tub, tap water instead of water that stood, another mixing temperature.

Not for a working solution drawn from a stock: step 12 gives what that gets instead.

Why a formula’s name and a litre’s name are different objects is set out in the developer laboratory report, which owns the scheme.

Before you start

You should already be able to read a formula, say which hydrate each mass belongs to and assign a formula version, from reading a developer formula and the report.

The hazard belongs to the mix, not to the paperwork. The session beside this one is classified by its own procedure — mixing a developer, mixing a fixer — and worked at the controls that classification sets. Write at a dry bench, clear of the open jars.

On the bench: a printed formula version record, FV-1 and FV-2 back to back; the register of codes already issued; the formula with the document and section it came from; a blank label; and the lab notebook.

Procedure

Issue the code

1. Write the stem: the published formula this mix claims as its parent, without punctuation — D76, D23, F5. It is a claim about parentage and has to be true.
2. If any quantity, form or make-up volume departs from that formula, do not use the parent’s stem. Add a variant marker — D76b, D76c — declared once on FV-3 with the parent and the single change, as the formulary marks one with status: variant. If unsure, declare it.
3. Add the initials of whoever will mix it. A systematic error travels with the hand.
4. Add the sequence: three digits, counted per person per stem, in the order batches were mixed — 001, 002, 003. Never reused and never renumbered.
5. Write the finished code — STEM-INITIALS-SEQUENCE — at the head of a new record, on the label before the bottle is filled, and in the register.

Fill the record while the batch is made

6. Enter the formula and the source it came from, by document and section, not by name alone.
7. Enter every mass actually weighed, to the balance’s resolution, with the balance used and the date it was last checked by the bench check.
8. Enter the water — source, whether it stood, temperature at each dissolving stage — the order actually followed with any departure and its reason, and the final volume with how it was found. Kodak’s 1949 handbook makes that order part of the formula.
9. Enter the measured pH with the temperature of the measurement, the instrument’s stated accuracy, and the colour and clarity when fresh. For a fixing bath, enter the fresh clearing time from mixing a fixer as well.
10. Enter the date, the time, the bottle and its headspace, where it is stored, and who mixed it.

Close the record, and keep the code honest

11. Apply the reconstruction test: hand the record to somebody who has your balance and your tap — could they make this same litre? Every “they would have to ask me” is a blank field with a name, so fill it now.
12. Give a working solution the stock’s code with its dilution and the date, never a code of its own: D76-EB-003 @ 1+3, 2026-09-05. The date is not optional — Kodak dilutes D-76 at 1:1 just before use and discards it after one batch, and ILFORD keeps a 1+1 or 1+3 dilution no more than 24 hours.
13. Cite the code — with the dilution and the volume in the vessel — on every sleeve, strip, negative, print, plot and notebook line the batch produces.
14. Open a new version, with the next sequence number, the moment any entry from steps 6 to 10 would change. A different tub of borax is a new version; so is 35 °C instead of 45 °C.
15. When the bottle is emptied, date the close of the record and file it with the strips it produced. The number stays spent.

When the batch is a proposed variant

16. Before mixing, complete FV-3 in its five parts and no more: the code and its parent with the edition; the single change as a quantity; the evidence from your own strips by cell name; the prediction, stated so it can fail; the test that decides it. Prefer a published quantity — 20 g/L of borax is Kodak’s D-76R level.
17. After the run, complete FV-4: what the control gave, what the variant gave, whether the prediction failed and where, your resolution limit, and the next version or none.

How you know it worked

• Every bottle carries a code; every code in the register points at one record and one bottle; no number appears twice and none has been renumbered.
• Every mass in the record is a reading somebody took, not a target aimed at.
• The reconstruction test passes, or the questions it raised are written where they are visible.
• Nothing downstream of the bottle cites the formula’s name where the code was meant.

If a step fails

Step 1, the formula has no published parent — it is your own. Take a stem of your initials and the substance, and declare it in the register as having no parent. A stem borrowed from a Kodak or ILFORD number claims a provenance that does not exist.

Step 4, a batch is spoiled before use. The number is spent. Strike the record, write why, and mix the next batch under the next number.

Step 4, an emulsion or another formula does not fit the three fields. Use the shape the owning part uses and record which. Part V codes a melt CB-01a and a second coating of that melt by the letter, so a different halide ratio is CB-02a and not CB-01b — see the chlorobromide project.

Step 7, a mass was not written down at the time. Do not reconstruct it from the formula. Record it as unknown; the batch is then usable for work but is not evidence about the formula.

Step 9, the meter will not settle. Record the range you could resolve honestly. ILFORD’s sheet tells users to take their own control measurements from their own fresh solutions; “between 8.5 and 9” is one, and a fabricated 8.63 is not.

Step 12, a diluted bath outlived its 24 hours. Discard it under the waste procedure and the disposal caveat; do not attach its results to the stock’s code. Dilute again from the stock.

Step 13, a strip cannot be traced to a bottle. Mark it provenance not established and keep it. It is an observation, not evidence about a formula.

What to record

Sheets FV-1 and FV-2 of the formula version record, one pair per batch, filled as it is made and closed when the bottle is finished; FV-3 and FV-4 where it is a proposed variant. The register line — code, date, mixer, formula, bottle — which is what stops a number being spent twice. And the code in three more places: the label, the notebook, and everything the bottle produced.