Beginner perfumer #115: the formula says lemon oil, and the most-supplied of its 98 candidates is lemongrass

Last article I computed that the corpus's lemon oil matches 98 database records. This one opens them up. The record with the most suppliers is lemongrass oil at 88, which is a grass with no relation to the lemon tree; the two share four letters at the start of a string. Among the real candidates, the widest gap is this: lemon oil lists a substantivity of 4 hours and lemon oil folded lists 264, and both carry the identical CAS number 8008-56-8. A folded oil is what remains after more than 90% of the terpenes are removed, and that sentence comes from the database's own notes field. It leaves a verifiable trace in the optical rotation: the crude oil reads +57 to +65.5, and the deterpenated oil reads -8 to -10. The sign flips. Specific gravity separates them too, 0.849 to 0.855 against 0.885 to 0.897, with no overlap.

About a 5 minute read.

The short version

98 database records begin with lemon; 88 of them have suppliers
The most-supplied is lemongrass oil (88 suppliers), a grass with no relation to lemon
Among the real candidates: lemon oil lists 4 hours, lemon oil folded lists 264 hours
Their CAS numbers are identical (8008-56-8), and so are their FEMA numbers (2625)
Folded means more than 90% of the terpenes removed. The database's own notes field says so
Verifiable trace: the optical rotation flips from +57 to +65.5 down to −8 to −10, and specific gravity separates them too
127 corpus formulas use lemon, and 52% name no qualifier at all

The top of the 98 is a different plant

Last article I found the corpus's lemon oil matching 98 database records, the largest candidate set of any natural material. This one opens it.

The first thing you find is awkward: sorted by supplier count, number one is lemongrass oil at 88.

Lemongrass is Cymbopogon, a grass. Lemon is Citrus limon, a rue-family tree. Their only relationship is four shared letters at the start of an English name. Fifteen of the 98 are lemongrass.

There is also a batch of synthetics dragged in by the same string match: lemonile (8 formulas), precyclemone b (8), isocyclemone e (4). None of them are lemon either.

Search a database by string and your first page may be entirely things you did not want.

Four real candidates

Database record	Suppliers	CAS	FEMA	Substantivity	Optical rotation	Specific gravity @25 °C
`lemon oil`	74	8008-56-8	2625	4 h @100%	+57.00 – +65.50	0.849 – 0.855
`lemon oil folded`	57	8008-56-8	2625	264 h @100%	(empty)	0.840 – 0.879
`lemon oil terpenes`	53	68917-33-9	4848	28 h @100%	(empty)	0.840 – 0.855
`lemon oil terpeneless`	22	68648-39-5	2626	98 h @0.10%	−8.00 – −10.00	0.885 – 0.897

One CAS number, one FEMA number, and 4 hours against 264: a factor of 66.

What folded means, in the database's own words

The notes field on lemon oil folded contains an unusually complete paragraph of process:

Expressed Lemon Oil consists of more than 90% terpenes which have but little flavor value and whose insolubility in alcohol is a nuisance in flavor compounding. By careful vacuum distillation, most of the terpene fractions of Lemon Oil can be removed. This method of concentration leads to…

Three facts:

The crude oil is more than 90% terpenes (mostly d-limonene)
Those terpenes carry little odour value and do not dissolve in alcohol
They come out by vacuum distillation

So "5-fold" means the volume was concentrated to a fifth. What you are buying is not something added; it is something removed.

And the 4 hours becoming 264 now makes sense: what was removed is the most volatile 90%.

He and colleagues were still working on new deterpenation methods in Food Chem in 2024 (extractive deterpenation using deep eutectic solvents, PMID 37603995). This is a process still being improved, not a relic.

The optical rotation flips sign

This is the part I like most.

d-Limonene is called "d" because it rotates polarised light to the right. It is 90% of the crude oil, so the whole bottle reads +57 to +65.5, strongly dextrorotatory.

Take the terpenes out and what remains, the oxygenated compounds, is laevorotatory overall: −8 to −10.

The sign flips.

In practice: a polarimeter will verify whether an oil has really been deterpenated. If a seller says terpeneless and you measure +50, it is not.

Specific gravity separates them too (0.849–0.855 against 0.885–0.897, no overlap), but refractive index does not (1.472–1.474 against 1.478–1.485, too close).

That 98 hours is not directly comparable

lemon oil terpeneless's substantivity field reads:

98 hour(s) at 0.10 %

Measured at 0.10%, where lemon oil's 4 hours and lemon oil folded's 264 are both at 100%.

I measured this problem earlier: the concentration marker after the substantivity value decides whether two numbers can sit side by side. Lasting 98 hours at 0.1% and lasting 264 hours at 100% are different claims.

(lemon oil terpeneless's notes is one line: Furo Coumarin Free. That is a separate selling point: deterpenation leaves the coumarins behind in the residue too. Naganuma and colleagues studied the phototoxicity of lemon oil in 1985, PMID 4096528.)

The CAS split is a little odd

Look at the CAS column:

lemon oil and lemon oil folded: the same CAS (8008-56-8) and the same FEMA (2625)
lemon oil terpeneless: its own CAS (68648-39-5) and FEMA (2626)
lemon oil terpenes: its own CAS (68917-33-9) and FEMA (4848)

Folding and deterpenation are both terpene-removal processes differing in degree. Yet one inherits the crude oil's numbers and the other gets its own.

Last round in the cananga entry I wrote that you should not decide whether two materials are the same from one field. This is the same sentence in another example: two records with an identical CAS and a 66-fold difference in substantivity.

What the corpus writes

127 formulas use lemon (excluding lemongrass and those three synthetics), at a median dose of 4.44%, quartiles 2.00–8.00%.

Corpus writes	Entries
`lemon oil` (unqualified)	66 (52%)
`lemon oil italy`	29
`lemon oil c.p. california`	11
deterpenated or terpene fractions	7

More than half do not say which. And not one writes folded.

Folding and deterpenation are genuinely used across the corpus though: 39 formulas (4.1%) use some deterpenated form or terpene fraction, just concentrated on other citrus. Orange 17, petitgrain 8, lemon 6, plus one or two each of geranium, bergamot, bay leaf, nutmeg and juniper berry, including orange peel oil 5 fold, 7 fold and 10 fold.

What to do

When a formula says only lemon oil:

Buy the expressed crude oil (the record with 74 suppliers). That is the default.
Know that it holds for 4 hours. Lemon is the opening, not the body. Something else has to carry it.
For anything going on skin, look for furocoumarin reduced or terpeneless. The crude oil's notes field says Do not use on skin.
If you see folded or 5 fold, that is a concentrate. Scale the dose down by the fold factor rather than copying the crude oil's proportion.
If you buy terpeneless, measure the optical rotation. It should be negative.

What this doesn't establish

I measured no optical rotation and no specific gravity. These are comparisons of field values.
"5-fold means concentrated fivefold" is a trade convention, and the database's lemon oil folded record carries no fold number. The corpus writes 5, 7 and 10 fold, but the database has no graded records to match them.
I did not verify the 264 hours. I have measured that the substantivity field should be read as a ranking; what I want here is the direction, not the figure.
The classification of the 98 used keyword rules, and I only spot-checked a few of the 55 in "other".
Furo Coumarin Free is what the field says; I did not verify it. Deterpenation and coumarin removal are different operations, and one process incidentally doing the other is plausible, but I have no source for it.
I have not smelled a folded oil against a crude one. This rests on fields and one paragraph of notes.