In short
A perfume iteration log records each version of a formula with its hypothesis, its exact change, the result and the decision that followed, so a reader can follow how you got there. It is the backbone of your capstone. When you have finished, the log will show what you did, including the wrong turns, which a finished formula never can.
What you will learn
- Plan a capstone formula in stages
- Keep a log where each version has a hypothesis and decision
- Know when to stop iterating
Before this lesson: Concept Development
Design in stages
- Skeleton. Build each main cue or sub-accord separately, in small batches.
- Assembly. Combine them in a first draft that totals 100%. Use your unit method if it is long.
- Structure. Fix how it evolves. Check the handover between stages, as in balancing volatility.
- Character. Make the small changes that give it identity.
- Constraints. Bring cost, restrictions and allergens into range. Do this throughout, and not only at the end.
- Refinement. Remove what isn't working, using removal tests.
The log format
| Field | Example |
|---|---|
| Version and date | V5, November 3 |
| Parent version | V4 |
| Hypothesis | Removing the benzoin will reduce sweetness at 2 hours without losing warmth |
| Change | Benzoin resinoid from 3.0 to 0 parts; solvent adjusted to keep the total |
| Test | Blind against V4, on strips, at 15 minutes, 2 hours and 8 hours |
| Result | Less sweet at 2 hours, as expected; thinner at 8 hours, which I did not expect |
| Decision | Keep, then address the 8-hour warmth in V6 |
| Cost and constraint effect | Cost down slightly; no new restriction concern |
The full formula for each version lives in your formula record. The log points to it. RUŌOD Lab keeps formula versions and notes, and a spreadsheet or notebook can do the same as long as you can retrieve the exact formula for any entry.
Rules that keep the log honest
- Write the hypothesis first. Once you have smelled the result, you can't recover what you expected.
- One change per version where possible. If you make several, say so, and accept that you can't separate their effects.
- Date everything. A later comparison might be testing age rather than formula.
- Retain samples. Keep a labeled sample of each version until the capstone is finished.
- Do not rewrite history. Correct an entry with a dated note, leaving the original readable.
Backtracking
Good logs have branches. If V7 is worse than V5, make V8 from V5 and record that V8's parent is V5. Run a blind comparison of your current version against an earlier one every few rounds. Your memory of V3 cannot be trusted, and your nose has adapted to your direction of travel.
Knowing when to stop
Stop when one of these is true:
- It meets the success test on your concept page.
- Your last two rounds did not produce a change that was reliably better in a blind comparison.
- You have reached your stated time limit. In that case, record where it stands.
Unresolved flaws can be listed in your case study. An honest "what I would do next" is more impressive than a claim of perfection, as the finished formula lesson explains.
Practical exercise: At least six logged versions
- Build your skeleton, then assemble a first draft.
- Make at least six logged versions, using the table's fields.
- Include at least one blind comparison with an earlier version, and one backtrack if warranted.
- Run at least two removal tests.
- Write your stopping reason.
You have done this well if someone could rebuild any version from your records, and they could follow why each change was made, including those that did not work. Keep this as a study composition until the later reviews are done.
Common mistakes
- Writing the log afterward. It then becomes a story and stops being a record.
- Testing only against the previous version. Drift can go unnoticed.
- Hiding failures. They are evidence of real work.
Knowledge check
Written and reviewed by the RUŌOD Lab team. This is general education, not legal, regulatory, medical or safety advice. Example dilutions and exercises are for learning, are not validated products, and have not been safety-assessed. Completing a lesson confers no qualification. How we check these lessons.