3.9 Misattributed claims
Three independent streams each compiled their own list of claims that do not match their sources. Merged and de-duplicated, they yield 20 items, which fall into five error types: a mixture attributed to a molecule, animals attributed to humans, one route of administration attributed to another, population level attributed to individual level, and figures with no traceable origin.
| Claim | Error type | Actual source of the figure | What the data actually say |
|---|---|---|---|
| Linalool is anxiolytic in humans | Mixture → molecule; oral → inhaled | Silexan: lavender oil taken orally, ~29% of its constituents unidentified | Linalool and linalyl acetate have never been tested on their own by the oral route. The data on pure linalool consist of only one study that reports no effect size, and one between-group comparison study that was negative. |
| The effect of lavender is the effect of linalool | Mixture → molecule | Whole lavender oil (more than 100 compounds) | Wrong even in sign: in a double-blind trial, lavender oil tended to raise blood pressure, whereas pure linalyl acetate lowered it. A mixture is not the sum of its constituents. |
| Limonene and citrus scents are calming | Mixture → molecule; wrong sign | One study with n = 12, an unidentified mixture, no control group and no outcome data published at all — yet cited 217 times | When the pure molecule is tested: (+)-limonene raises systolic blood pressure and increases feelings of restlessness. All four chiral substances in that study raised blood pressure. |
| 1,8-cineole improves cognition | Mixture → molecule; measured → administered | The exposure was rosemary essential oil; cineole was only measured in serum, not administered | No control group; cineole concentration is fully confounded with exposure duration (r = 0.727). And an adverse effect nobody cites: contentment scores fell (r = −0.454; p = 0.044). |
| Cedrol and cedarwood relax through aromatherapy | Route of administration reversed | The effect persists when olfaction is completely eliminated | The mechanism is pulmonary absorption, not olfaction. Scientifically, cedrol is closer to an inhaled antihypertensive than to aromatherapy. The effect is real; the explanation is wrong. |
| Vanillin is calming and analgesic | Animals → humans; mechanism changed | The mechanism stated by the original paper itself is prior olfactory experience, i.e. familiarisation | In that same work vanillin could be replaced by anethole. Without a prior familiarisation phase, a trial with n = 90 was negative (p = 0.36). The richest behavioural dataset is in rats. |
| 2-phenylethanol is an antidepressant | Animals → humans | All the data are from mice | In humans, 2-phenylethanol serves only as a reference substance in olfactory test batteries. We have found no emotion study in humans. |
| α-pinene improves sleep via GABA-A receptors | Animals → humans | The GABA-A mechanism is mouse data | The only human data are 90 seconds of inhalation from an air bag, n = 13, and no sleep outcome of any kind was measured. |
| Sandalwood is calming | Wrong sign at both levels | The title of the original paper states explicitly that it increases physiological and self-rated arousal | The essential oil increased pulse rate, skin conductance and systolic blood pressure; pure α-santalol increased attentiveness. Neither was calming. |
| Jasmine is the most powerful anxiolytic essential oil | Ranking from a single arm | The ranking rests on one trial arm, with a confidence interval about 22 STAI points wide | The jasmine molecule with the best human data is Hedione, and its result was negative, with η²ₚ = 0.00 for cortisol, amylase, blood pressure and self-rating scales. |
| Linalool has two isomers with opposite effects | Intensity confound not ruled out | Detection thresholds: (R)-(−) about 9–11 ppm versus (S)-(+) about 35–40 ppm | The original study gave both isomers at the same concentration, set at threshold. “(S)-(+) has no effect” may well mean only “(S)-(+) was not smelt”. And the original paper reports one substance with an effect and one without — not opposite signs. |
| Pleasant odours have an optimal concentration and then decline (inverted-U curve) | No original paper | The source is a textbook chapter, not a study | Three studies with data show pleasantness rising and then levelling off, not declining. Unpleasant odours decline steeply. |
| Indole and skatole smell floral when diluted | Overgeneralisation | For indole, a 4-fold increase in concentration reversed the sign in 12/27 people (~44%) | The remaining 15 did not change. For skatole, p-cresol, grapefruit mercaptan, civetone, costus and the aldehydes: no quantitative human data exist at all. |
| Formulas with many notes give a multi-layered experience | Exceeds the perceptual limit | Humans can resolve at most about three components | Training and expert experience do not raise the limit. A 30-ingredient formula is not perceived as 30 notes. |
| Blending in more ingredients makes a scent more complex | Ignores convergence | From about 20–30 equal-intensity components spanning the odour space widely | Different formulas converge on the same smell. Coverage matters more than number: mixtures from the same cluster remain discriminable (67.5%), widely spanning mixtures do not (42.4%). |
| Adding ingredients means adding odour | Ignores antagonism and sub-additivity | The intensity of binary mixtures is about 34% lower than the sum | At the receptor level, antagonism events roughly equal activation events — but those data are mouse receptors at 300 µM, orders of magnitude above natural airborne concentrations. |
| Odours disappear after about 20 minutes, so scents must be changed | Untraceable figure | Intensity levels off at 25–40% of its initial level after about 4 minutes and then stops falling | Odour does not go to zero. Some people still perceive it after 15 minutes. And trigeminal pungency does not adapt — pungency even increases with inhalation time. |
| Chemical structure determines pleasantness | Population level confused with individual level | The predictive model reached r = 0.71 at population level but r = 0.41 at individual level | That is about 50% of the variance at the group-mean level, only about 17% at the level of a specific person. One author of that very model wrote that prediction at the individual level is not feasible. |
| Piriform cortex activation proves an olfactory effect | Brain region confused with sense | CO₂ — an odourless substance — activates the piriform cortex in PET, coordinate-based meta-analysis and independent component analysis | The signal may come from the trigeminal nerve. Piriform activation is not evidence of an olfactory response. |
| Aromatherapy reduces STAI anxiety by 6–7 points | Effect generated by small trials | Pooled 44 trials: −6.63 (95% CI −8.17; −5.08) | Excluding trials with fewer than 30 people per arm: −0.84. I² = 93%; not a single comparison reached moderate GRADE certainty or above. |
Two items in the table deserve emphasis because they are not ordinary citation errors but problems of experimental design. First, the difference between the two optical isomers of linalool — still used as the classic example of the importance of stereochemistry in perfumery — has not had the intensity confound ruled out: the two isomers have detection thresholds that differ three- to four-fold, yet the original study gave both at the same concentration, set at threshold. Second, there are signs of publication bias at the molecular level itself: a controlled dose–response study of both pure linalool isomers, carried out in 2005, does not appear in the peer-reviewed literature.
Scientific reference material, not medical advice and not promotional material. All assessments of evidence quality are the compilers’ own. Figures are redrawn from published data.