In short
Release testing checks a batch against a written specification, then records a decision to pass, hold or reject, signed by a named person. Your specification says what must be true. Release is the routine that applies it, batch after batch. A small maker can do a lot here with a scale, a refractometer if available, retained samples, and honest comparison.
What you will learn
- Set provisional ranges from batch data
- Write a batch release record
- Decide among pass, hold and reject using stated rules
Before this lesson: Allergen Calculation in Depth
What to test
| Test | What it catches | What it misses |
|---|---|---|
| Record check | Wrong lot, wrong weight, missed step | Everything the record doesn't show |
| Appearance | Haze, deposit, color drift | Most odor problems |
| Odor against a retained reference, blind | Character changes, some mistakes | Small chemical changes, and it depends on your nose |
| Refractive index | Gross errors, wrong dose, wrong material | Compensating errors. See the refractive index guide |
| Density | Dose and mixing errors | Small differences |
| Alcohol content, if measured | Dilution errors | Method limits; see the guide |
| Fill weight | Underfilling | Content quality |
| GC, through a laboratory | Composition against a standard | Cost; method limits |
Setting ranges
You usually don't start with ranges. You grow them.
- Measure carefully across your first batches, under controlled conditions.
- Check your method: repeat a measurement on one sample, several times. How much does it vary by itself?
- Set a provisional range around your values, wide enough to allow for method noise, and narrow enough to catch a real error.
- Test it: make a deliberately wrong small batch, say with one line missing. Does your method catch it?
- Review the range after more batches, and record each change with a reason.
Worked example, with invented readings. Five batches read 1.4628, 1.4632, 1.4630, 1.4634 and 1.4629 at a stated temperature. Mean: 1.4631. Spread: 0.0006. You might set a provisional range of 1.4620 to 1.4640. A wrong batch reading 1.4655 would be clear. Is your range too wide? Try the deliberate error batch to find out.
A release record
Product, version, batch code, date made. Inputs: all lots traced. Weighing record reviewed. Maceration and filtration as specified. Tests: each with method, result, range and pass or fail. Reference used. Deviations. Decision. Name, date.
Pass, hold and reject
| Status | When | Next step |
|---|---|---|
| Pass | All criteria met and the record is complete | Release; keep a retained sample |
| Hold | Marginal result, doubtful odor, gap in record | Mark and separate it; investigate. See deviations |
| Reject | Clear failure, wrong material or unsafe outcome | Don't use it; record what was done with it |
Don't test a failed batch repeatedly until a pass appears. If you retest, have a stated reason and report both results.
The odor test, done fairly
Retained references change with time. So store them well, in full, dark, cool containers, and renew them under a stated rule, using a passed batch. A triangle test against the reference is a good check. A nose that's already tired, or knows the expected answer, will pass nearly anything. Blind yourself, or use someone else, as in blinded design.
What release doesn't mean
Release says this batch matches what you specified. It doesn't, by itself, establish that the product is safe, that it meets every law, or that its shelf life is what you claim. Those rest on other documents in your file.
Practical exercise: A release decision on one batch
- Choose a fragrance you've made more than once, or make two small batches.
- Write a specification with at least three tests that you can perform.
- Measure your method's repeatability, for one number or by repeating a triangle test.
- Make a small deliberate error batch, marked as such, and see if your tests catch it.
- Complete a release record for one real batch.
- Write your hold procedure on one page.
You have done this well if your ranges are labeled provisional and rest on data, your error batch was detected or you understand why it wasn't, and the record has a named decision.
Common mistakes
- Retesting until it passes. Report all results.
- Using numbers alone. A batch can pass and still smell wrong.
- Letting the reference age unnoticed. Date it, and renew it by rule.
Key terms
- Batch release
- The documented decision that a batch meets its specification and can be used or sold.
- Retained sample
- A stored sample from a batch, kept so it can be compared or investigated later.
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.