Material guide: musk xylol — the database says "not used for fragrances", and another field on the same record says "fragrance"

About 6 minutes.

Short version

• Name: musk xylol (musk xylene), CAS 81-15-2
• Formula C12H15N3O6, MW 297.27; three nitro groups
• Odour: musk type — fatty, dry, sweet, soapy, musk; medium strength, rank 2
• Substantivity: the field reads "400 hours at 20% in DPG"
• Appearance: pale yellow crystals, melting point 235.4 °C
• Suppliers: 20
• GHS: H351 suspected of causing cancer, H410 very toxic to aquatic life with long lasting effects
• Acute toxicity: oral rat LD50 > 10000 mg/kg, dermal rabbit > 15000 mg/kg
• Do not use this one. What follows is why, and why you can still buy it.

Two fields on the same record contradict each other

The category and uses fields read:

information only not used for fragrances or flavors

The cosmetic_uses field on the same record reads:

fragrance; perfuming agents

Across all 42,225 records, only two satisfy both conditions at once. The other is sodium cyclamate — an artificial sweetener.

The control case sits right next to it: musk ambrette (83-66-9, 18 suppliers) has cosmetic_uses reading "not used anymore". Two equally retired nitro musks, two different answers.

That is its own article — How do you tell a material has been retired — where the two retirement flags turn out to overlap on 18 records.

The practical conclusion is simple: do not use cosmetic_uses to decide whether something is usable.

Why GHS says "suspected" rather than "confirmed"

The GHS classification field:

H351 — Suspected of causing cancer (Carcinogenicity Category 2) H410 — Very toxic to aquatic life with long lasting effects

Category 2 "suspected" is a different tier from Category 1 "known or presumed", and there is a reason this one stops at Category 2.

Lehman-McKeeman and colleagues laid it out in Molecular Carcinogenesis in 1997. Their opening line:

Musk xylene is a synthetic nitromusk perfume ingredient that, although uniformly negative in genotoxicity testing, causes liver tumors in B6C3F1 mice.

Two things hold at once. Nothing detects it damaging DNA, and it gives mice tumours.

The paper is chasing the mechanism. It induces cytochrome P450 enzymes in a phenobarbital-like manner, which points at an epigenetic — non-genotoxic — route. At the same time it is nitroreduced in vivo by intestinal flora into aromatic amine metabolites, and those metabolites themselves strongly induce CYP2B10.

That is where "suspected" comes from: animal tumours, no genotoxicity, and a mechanism of enzyme induction. That combination leaves room in the cross-species extrapolation, so the classification stops at Category 2.

This does not mean it is safer. The next section is about a different endpoint.

The acute toxicity is very low, and that is not the point

Oral rat LD50 > 10000 mg/kg, dermal rabbit LD50 > 15000 mg/kg, both from the Bulletin of Environmental Contamination and Toxicology in 1981.

Those are handsome numbers, and they answer the wrong question.

Acute toxicity measures whether a large single dose kills you. Nitro musks were phased out over accumulation — they are lipophilic, resist degradation, and stay in fat.

Müller and colleagues analysed 15 human adipose tissue samples from Switzerland by GC/high-resolution MS in Chemosphere in 1996, measuring five nitro musks and three polycyclic musks. Musk xylene ranged up to 288 ng/g lipid.

Suter-Eichenberger and colleagues ran a rat feeding study in the same journal in 1998, and the result has three layers:

1. It accumulates in adipose tissue, with significant amounts in other organs — ovary and adrenal.
2. It crosses the placenta and passes to pups in milk. Milk levels were comparable to adult female adipose tissue.
3. Female tissue levels were 3.7 to 6.8 times higher than male, a difference unrelated to lipid content, which the paper calls unexplained, and which was absent in the offspring.

The third point is the most interesting thing I read here — a sex difference that was never explained away, left standing in the paper.

(The rest of this line is in Synthetic musks are not one category and Material guide: musk ketone. Musk ketone survives on a single specification — musk xylene content below 0.1%. So the existence of this one is a drag on the one that is still usable.)

The solubility field unusually gives five numbers

This record's solubility field is numbers, not adjectives:

Solvent	Grams per 100 g
Benzyl benzoate	25.2
Dimethyl phthalate	16.8
Diethyl phthalate	14.4
Methyl carbitol	5.5
95% alcohol	0.88

Benzyl benzoate and ethanol are 28.6× apart.

A crystalline solid melting at 235.4 °C that barely dissolves in ethanol explains how it was actually used — dissolved in benzyl benzoate as a pre-solution. The odour description field carries the trace: "at 10.00% in benzyl benzoate: fatty dry sweet soapy musk".

This is one of the few records in this series that uses a non-alcohol solvent as the evaluation basis. Most read "10% in DPG".

That qualifier in the substantivity field

The raw string in the substantivity field is:

400 hour(s) at 20.00 % in dipropylene glycol

20% in DPG, not neat. And the numeric column drops the qualifiers and keeps 400.0.

400 is the ceiling of that column (334 of the 2,027 records with a value read exactly 400), and the dropped-qualifier problem is taken apart in That "400 hours" for Ambroxan. This record adds a third variant: Ambroxan is "> 400 at 10% in DPG", Javanol is "400 at 100%", and this one is "400 at 20% in DPG". Three different measurements, one 400 in the numeric column.

What to do

• Do not buy it, do not use it. Its own category field says so.
• If you have old compounds or old formulas, it is worth checking. This was widely used through the 1980s and 1990s.
• If you use musk ketone, note that its specification is "musk xylene below 0.1%" — a spec to confirm with your supplier, not something that holds automatically.
• Twenty suppliers does not mean it is usable. Buyable and usable are different things.

What this doesn't establish

• I have not smelled it and do not intend to. The whole article is fields, database-wide counts and three papers.
• I did not check current regulatory text. This article is about database fields and published toxicology, not the legal status in any jurisdiction. To establish that, check the IFRA standards and your own region's cosmetic regulation.
• The Lehman-McKeeman paper is mice and mechanism, not human risk assessment. What I take from it is why GHS says "suspected", not a conclusion at any exposure level.
• Müller 1996 has n = 15, from Switzerland, in the mid-1990s. That is not a current exposure level; nitro musks largely left the market afterwards.
• The 3.7–6.8× sex difference is called unexplained by the paper itself. I did not check whether later work explained it.
• The five solubility figures are supplier data the database carries, with no stated method or temperature.
• The acute toxicity data are from 1981 and 1990 literature, as the field itself states.
• Twenty suppliers describes this database's coverage, not whether they still stock it.