The twin who didn’t grow breasts …

… the trouble with believing ‘the science’ … completely!

In 2007, the New England Journal of Medicine published a brief report describing three prepubertal boys who developed breast tissue. All three were otherwise healthy, with normal serum steroids. All three had been using personal care products said to contain lavender or tea tree oil — a healing balm, a styling gel, a scented soap and lotion. In each case, the gynecomastia resolved after the products were discontinued. The authors ran the oils through in vitro assays, found weak estrogenic and antiandrogenic activity in human cell lines, and concluded that repeated topical exposure had probably caused the breast development.

One of the boys had a fraternal twin. The twin used the same lavender-scented products, in the same household, and developed nothing.

That detail is in the original paper. It is not hidden. And in twenty years of increasingly bitter disbelief about this report — an argument that has consumed regulatory attention, industry rebuttals, follow-up studies at a national laboratory, and an entire population of hyper-concerned parents — almost nobody on either side has treated the twin as the clarifying sentence in the case series. Everyone argued about the oil. Nobody wanted to talk about the boys.

The fight, briefly

The rebuttals arrived fast, and some of them landed. The reported estrogenic activity was extraordinarily weak. The products were self-reported and never chemically analyzed; nobody measured what was actually in them or how much reached the skin. Dermal penetration studies suggested the absorbed dose of the relevant constituents would be far too low. A 2024 review ran the full case literature — twelve patients across five publications, all told — through Hill’s criteria for causation and concluded the association was weak and cause-and-effect had not been shown.

I will note, because my whole argument here is about intellectual hygiene, that several of the loudest early rebuttals came from people with commercial stakes in tea tree and lavender — a conflict they disclosed and their readers mostly ignored. A disclosed conflict is not a refutation of a refutation. But it is context, and this is a story about what happens when nobody’s context gets examined.

Then in 2019, the original group’s successors at NIEHS escalated. Ramsey and colleagues added new cases — including premature thelarche in girls, one of whom was exposed by sitting near a diffuser at school — and did something the 2007 paper never did: they tested constituents. Eight of them, the ones ISO mandates for lavender and tea tree oil, in receptor-transfected human cells. Estradiol drove a roughly fifty-fold increase in estrogen-response activity; the whole oils drove about twenty-fold. And the constituents scattered: alpha-terpineol was the most active, terpinen-4-ol and linalool somewhat less, linalyl acetate weaker still, and eucalyptol, limonene, and alpha-terpinene showed essentially nothing.

Hold that last sentence, because it quietly detonates the terms of the entire debate. If eucalyptol is inert and alpha-terpineol is active, then ‘lavender oil’ was never a single exposure. It is a mixture whose endocrine-relevant chemistry varies with species, chemotype, harvest, and adulteration — which means that for twenty years, both camps have been arguing about a variable that does not exist.

Then the products were finally analyzed

It gets better, by which I mean worse. In 2020, letters to the journal reported chemical analyses of the actual products implicated in the new cases. In some of them, linalyl acetate — the signature ester of true lavender — could not be reliably detected at all, and linalool appeared only in trace amounts in one product. Whatever was making these products smell like lavender, it apparently wasn’t much lavender.

What one product did contain, at 3.61 percent:

diethyl phthalate — a synthetic compound with a well-documented endocrine-disruption literature of its own (which the case report had filed under ‘unidentified substances.’)

Another contained Lilial, a suspected endocrine disruptor since banned in EU cosmetics.

So the strongest recent cases in the ‘lavender disrupts hormones’ file involve products that may have contained little or no lavender, but did contain known synthetic endocrine disruptors that nobody initially looked for. The critics, meanwhile, raised the possibility that the in vitro assays themselves were contaminated by estrogenic compounds leaching from plastic labware. At every layer of this controversy, the exposure was assumed rather than characterized. The word ‘lavender’ on a label was treated as ‘beneficial’. It was, and continues to be, marketing.

Now, about that twin

Here is where I part company with both camps, because the twin discordance cuts in a direction neither of them wants to face. If the products caused one boy’s gynecomastia, why not his brother’s? There are two honest answers. Either the lavender products didn’t cause it — prepubertal gynecomastia has a background rate; most cases are labeled idiopathic; and resolution over months proves nothing, as the condition frequently resolves on its own. Or the products affected one boy and not the other, because the boys were different:

• different skin barrier function and absorption
• different steroid metabolism
• different receptor sensitivity
• different everything that makes fraternal twins siblings rather than replicates

Notice that both answers destroy the same thing: the population-level claim. ‘Lavender causes gynecomastia’ fails either way — either as a false alarm, or as a statement missing its essential qualifier, in a susceptible child, under an exposure nobody measured. The one reading the twin will not support is the one the entire debate was conducted in: that a substance called lavender has a property called endocrine disruption.

What the contradiction is telling us

I want to be precise about my own position, because this is a case where my field’s instinct — circle the wagons, the oils are fine — deserves as much scrutiny as the alarm. The in vitro receptor activity is real, replicated, and constituent-specific; alpha-terpineol and linalool do things in those assays that eucalyptol does not, and pretending otherwise is its own evidence-laundering. Whether that activity survives dermal absorption, systemic dilution, and metabolism to matter in an actual child remains unshown, and the Hill’s-criteria analysis says so plainly. Weak in vitro signal plus uncharacterized exposure plus twelve heterogeneous case reports is not causation. It is also ‘not’ nothing. It is a hypothesis that was never tested properly because no one characterized either side of the interaction.

And that is the lesson this file adds to the series. Installment one argued that the receiving nervous system is a variable our trials refuse to measure. This one shows the mirror failure: the exposure itself was never measured either. Twenty years, a national laboratory, the world’s most prestigious medical journal — and at no point did anyone establish, at the constituent level, what twelve children actually absorbed, or what distinguished them from the millions of children, including one fraternal twin, who used the same shelf of products and developed nothing. Both camps treated ‘lavender’ as a monolith and children as interchangeable. The contradiction was never in the data. It was manufactured by the resolution at which everyone insisted on looking.

Next in the series: Linalool — the constituent that showed up in both installments so far — is documented in the literature as a sedative. It is also documented in the literature as doing nothing of the kind, and in some conditions the arrow points the other way. Same molecule, opposite nervous systems. We’ll look at what determines which one shows up.

Tammy L. Davis, MCN, is an instructor at the University of New Mexico and Santa Fe Community College as well as the founder of Aromagenomics and developer of the ANIS™ methodology.