I guessed wrong last time: the storage field is not derived from chemistry, it arrives with the record's source

About a 5 minute read.

Last time I took apart the 145 records told to refrigerate and found two lines: carboxylic acids 9.4× enriched, sulfur compounds 3.8×. Then, about the leftover ionones and damascone, I wrote:

They are a separate line — conjugated ketones that photo-oxidise. The database gives no reason and I am not going to invent one.

This article tests that sentence. It failed.

Two more lines that do hold, first

Before giving up I swept the residue again and found two cleaner than the original pair:

Class	Refrigerate	Other	Ratio
Polyunsaturated fatty chains	4.8%	0.1%	52.5×
Single optical/stereoisomer markers	13.1%	1.5%	8.9×

52.5× is the strongest enrichment in the whole analysis. The seven records: linoleic acid (21 suppliers), ethyl (Z)-linoleate (18), linolenic acid (16), α-tocopherol (14), methyl linoleate (13), ethyl linolenate (10), arachidonic acid (9).

α-tocopherol is vitamin E — it is itself an antioxidant. An antioxidant told to refrigerate is not protecting a formula, it is protecting itself. Polyunsaturated carbon chains autoxidise and go rancid. This group's reason for refrigeration is the clearest of the lot.

The second line needs care. Records whose names carry (R)-, (S)-, (+)-, (−)-, laevo- or dextro- are 8.9× enriched, and their median supplier count is 5 (against 7 for the refrigerate group as a whole). That looks like analytical reference standards — single isomers, almost no suppliers, and refrigeration is how standards are stored.

But there is a confound. Half of those 19 are amino acid derivatives (laevo-lactic acid, S-allyl-laevo-cysteine, acetyl-laevo-tyrosine…), which are nutraceutical raw materials. I cannot separate "single isomer" from "biochemical."

And then the hypothesis that failed

Conjugated carbonyls photo-oxidise, so refrigerate them — it sounds reasonable. I tested it on records whose names contain -enal, -dienal, ionone, damascone or -enone:

	Refrigerate	Other	Ratio
Unsaturated aldehydes / conjugated ketones	6.9%	4.2%	1.6×

1.6×. Against 9.4× for acids and 52.5× for polyunsaturates. That is not a signal.

Unsaturated aldehydes and conjugated ketones are handled about like everything else across the database's storage field. Nobody is singling them out.

What killed it outright was testing within the family

A class-level test still leaves room for "maybe only a few of them are unstable." So I pulled the whole ionone/damascone family:

Suppliers	Material	Shelf life	Storage
115	α-ionone	24 months	Refrigerate
78	β-ionone	24 months	Refrigerate
47	α-isomethyl ionone (50% min.)	36 months	Cool, dry
42	δ-damascone	36 months	Cool, dry
41	(E)-β-damascone	12 months	Refrigerate
30	Dihydro-β-ionone	24 months	Nitrogen
14	α-damascone	12 months	Cool, dry
11	β-damascone	12 months	Cool, dry

One chemical family, four different storage instructions. If this were a chemical rule it would apply to the family. It does not.

And the cleanest pair is in the last two rows

(E)-β-damascone (#11165, 41 suppliers) and β-damascone (#11159, 11 suppliers):

	(E)-β-damascone	β-damascone
FEMA	3243	3243
Formula / MW	C13H20O / 192.30	identical
Boiling point	200 °C	200 °C
logP	3.50	3.50
Specific gravity	0.928–0.936	0.925–0.932
Storage	Refrigerate	Cool, dry place
Substantivity	264 hours	191 hours
CAS	23726-91-2	35044-68-9

The same FEMA number. FEMA 3243 is assigned to both — the regulator treats them as the same substance (one specifies E geometry, one does not).

Given that: one says refrigerate and one says a cool dry place is fine. Substantivity reads 264 hours on one and 191 on the other — 38% apart.

Both of those cannot be measurements of the same thing.

So what is this field

It arrives with the record. Nobody derived it from chemistry.

Each record's storage line comes from wherever that record came from — a supplier's specification sheet, a food-ingredient handling convention, a reference-standard catalogue. The same material passing through two sources picks up two different sentences.

This also corrects last time's article. The 9.4× for acids and 52.5× for polyunsaturates still stand, but what they describe is which kinds of records came from sources that say refrigerate — not that somebody made a judgment about those chemical classes. In practice that distinction matters.

Three corrections for practice

1. Do not infer from a sibling record. A blank storage field on your material does not license copying another record in the same family — this one family holds four different answers.
2. The absence of "refrigerate" is not evidence of stability. β-damascone is not told to refrigerate, and it shares a FEMA number with (E)-β-damascone, which is.
3. When two records' numbers disagree, ask whether they came from the same source first. 264 versus 191 hours is not someone's measurement error; it is two documents.

And 85 records I still cannot explain

The union of all five classes — acids, sulfur, polyunsaturates, isomer markers, conjugated carbonyls — covers 60 of the 145. 41.4%.

The remaining 85 include the two best-supplied records of all: geranyl acetate (130 suppliers) and γ-decalactone (111). I have no explanation for either.

If the "inherited, not derived" conclusion is right, those 85 do not need a chemical explanation in the first place. But I cannot prove that; I can only say I could not find one.

What this doesn't establish

• "Storage is inherited" is inferred from within-family inconsistency. I have no source document for any individual record to show you.
• The chemical classification still relies on name suffixes, which is blunt, and the conjugated-carbonyl test misses anything not named -enal or -enone (methional slipped through).
• 1.6× means "no enrichment," not "proven irrelevant." Individual materials may still be refrigerated because of oxidation.
• The single-isomer line is tangled with amino acid derivatives and I did not separate them.
• Attributing the two damascones' substantivity gap to different sources is an inference. The database records no source.
• I ran no storage experiment at all. This is field comparison.