Strongest exact-panel delta. Not a causal result.
What the evidence currently says · 13 July 2026
What we know. What we don’t.
This study compares reported trigger exposures with tolerated controls. It keeps observations separate from molecule-level guesses.
Current answer
Leading pair: citronellol + alpha-isomethyl ionone.
Four bad or likely-bad panels share eight named ingredients. Exact English control panels subtract six, leaving citronellol + alpha-isomethyl ionone as the strongest pair-level label hypothesis.
Repeated observation. Exact ingredient unknown.
Best-supported mechanism in this study.
Local label overlap. Exact eaten packs unknown.
01 · scent
Four bad or likely-bad labels narrow to one pair.
The exact English-panel delta is useful; it is not proof of causation or zero presence in controls.
Imaginary Authors — Falling Into the Sea and Maison Margiela REPLICA — Flower Market are confirmed reported triggers. Bleu de Chanel and Versace Eros are suspected or likely bad, but variants and episode details remain incomplete. CHAMO liquid perfume and BERGA SANDAL perfume balm were tolerated.
Compare the four bad or likely-bad panels with the two exact tolerated-control panels.
All four bad or likely-bad lists contain citronellol + alpha-isomethyl ionone. The exact English tolerated-control panels list neither.
That either candidate caused the headache. Parfum and Fragrance hide mixtures, and no concentration or air sample was measured.
Photograph actual packages and batch codes. Record only naturally occurring Bleu/Eros outcomes, including exact variant, distance, timing, and recovery. No deliberate re-challenge.
Maison Margiela · Flower MarketReported episodes
Exact English-list candidatesCitronellol + alpha-isomethyl ionone
alcohol crossed outparfum crossed outlimonene crossed outlinalool crossed out
Falling physical-label transcription: Alcohol Denat., Parfum, Aqua, Alpha-Isomethyl Ionone, Benzyl Alcohol, Citronellol, Coumarin, Limonene, Linalool, Benzyl Cinnamate, Farnesol, Geraniol.
Four-panel intersection: alcohol, water, parfum, limonene, linalool, geraniol, citronellol, and alpha-isomethyl ionone. Exact English control panels subtract the first six, leaving the final pair.
Spelling check: the shared transcription said “alpha isomethyl iodine.” Standard INCI wording and an independent matching retailer panel indicate alpha-isomethyl ionone. A package photo and batch code are still needed to confirm it.
Secondary label footnote: same-SKU Japanese [+/−] panels for CHAMO and BERGA balm list both candidates as may-contain. These broad regulatory umbrella lists do not show that an actual tolerated bottle contains every item—especially if it was bought elsewhere. Actual package label first; product-specific English panel next; Japanese umbrella panel only prevents an absolute-zero claim.
BERGA identity: the tolerated control is PERFUME BALM BERGA SANDAL (SKU 12001156). Its exact English list separately names neither final candidate, but Parfum hides the mixture. The related liquid perfume (SKU 1662463458) is a same-scent formula—not a reported exposure.
Evidence ladder: Imaginary Authors — Falling Into the Sea + Maison Margiela REPLICA — Flower Market are confirmed reported triggers. Bleu de Chanel + Versace Eros are suspected or likely bad, not confirmed. CHAMO liquid perfume + BERGA SANDAL perfume balm are tolerated controls.
Alpha-isomethyl ionone: a common fragrance INCI. An EC SCCS survey found it on 21.7% of 516 analysed consumer-product labels, and human skin-contact evidence supports allergen status. This audit found no evidence isolating inhalation as the cause of this headache.
Where citronellol comes from
- The molecule
- Citronellol (C₁₀H₂₀O) is a rose/geranium-smelling monoterpene alcohol.
- Natural sources
- Rose, rose-geranium, and citronella oils. One steam-distilled rose-geranium sample contained 29.1%; natural composition varies by plant, season, and processing.
- Industrial routes
- It can be recovered by distilling geranium or citronella oil, or manufactured by reducing citronellal or hydrogenating geraniol/nerol.
- What the label reveals
- Only the name. It does not reveal natural versus industrial origin, enantiomer ratio, purity, concentration, or batch impurities.
Three different “by-product” questions
Depending on route and purification, plausible residues or side-products include geraniol/nerol, geranial, and over-hydrogenated 3,7-dimethyloctanol. Their presence in these perfumes is not established.
Controlled ageing produced citronellol hydroperoxides as the major products, with smaller amounts of aldehydes, esters, epoxides, and diols. The established human relevance is delayed skin contact allergy.
Gas-phase work identified acetone, glyoxal, methylglyoxal, and proposed 6-hydroxy-4-methylhexanal. These were chamber-reaction products—not measurements of the reported rooms or symptoms.
Air-aged citronellol can become more skin-sensitising. Perfume odours can trigger attacks in some people with migraine.
No human study found that isolates inhaled citronellol as the headache cause. “Very common,” direct sinus inflammation, sudden brain-vessel changes, and “synthetic is worse” are unsupported.
Important distinction: citronellol is one molecule. Citronella oil is a multi-molecule plant oil. Natural and industrial citronellol share the same constitutional compound/name; stereoisomer and impurity profiles may differ. That does not make one source automatically safe.
Evidence receiptFalling · independent matching transcriptionCHAMO liquid perfume · officialCHAMO Japan · broad may-contain panelBERGA SANDAL balm · officialFlower Market · officialFalling · official notes onlyFalling retailer panel · version context onlyFalling retailer panel · version context onlyEC SCCS · alpha-isomethyl-ionone label prevalenceEC SCCS · skin-contact evidenceNIH PubChem · origin and propertiesRose-geranium composition studyRSC · geraniol hydrogenationCitronellol autoxidation studyHuman patch-test studyGas-phase oxidation study
02 · chicken
The food pattern is stronger than the protein theory.
Chicken is worth tracking. Gal d 7 is one possible explanation, not the established cause.
Chicken meat and soup were repeatedly linked to sharp pain. Egg, duck, fish, beef, and pork were reportedly tolerated.
Either a chicken-specific muscle protein such as Gal d 7, or a repeated stock, seasoning, recipe, or meal-context feature.
Chicken exposure: moderate observation.
Gal d 7: low-confidence mechanism.
An allergy, or a need to avoid every animal with a related muscle protein. Headache alone is not the usual IgE poultry-allergy pattern.
During natural events, record plain meat versus broth/powder, latency, and any oral, skin, breathing, or gut symptoms. Do not provoke a reaction.
- Gal d 7
- Chicken myosin light chain 1f: a 22 kDa, heat-stable skeletal-muscle allergen. It is not an egg protein.
- Closest clinical relative
- Turkey commonly cross-reacts with chicken in poultry-meat allergy, but not everyone reacts clinically.
- Similar proteins, tolerated foods
- Duck, fish, beef, and pork contain related muscle proteins. Their reported tolerance is stronger personal evidence than sequence similarity, so this study does not add them to an avoid list.
- Different chicken pathway
- Gal d 5 is serum albumin involved in bird–egg syndrome and egg yolk reactions. Repeated egg tolerance makes that pathway less fitting.
Practical boundary: keep chicken meat and clearly chicken-derived stock as the specific foods under investigation. Do not expand to egg, duck, turkey, fish, beef, pork, or shellfish solely because homologous proteins exist.
Evidence receiptWHO/IUIS · Gal d 7Primary Gal d 7 studyPoultry allergy and cross-reactivity reviewBovine myosin light-chain studyGal d 5 bird–egg syndrome
03 · mushroom / truffle brain fog
No mushroom molecule has survived yet.
A cheap truffle carbonara is a mixed exposure, not a mushroom test.
Brain fog followed a truffle-carbonara meal; fog can also occur before breakfast. Chocolate reportedly helped once.
Meal size, carbohydrate/fat load, long fasting, sleep, hydration, mushroom polyols, or a flavouring—not one assumed toxin.
No specific molecule is supported. The mixed meal and baseline fog prevent a clean subtraction.
That mushroom, truffle aroma, dairy, gluten, or sugar caused the fog.
Record the full dish, time since the previous meal, sleep, fog before eating, onset, and recovery. A naturally repeated plain-mushroom event would be more informative than another carbonara.
Mushrooms can contain this poorly absorbed polyol. Controlled evidence supports abdominal symptoms in susceptible people—not isolated brain fog.
A common commercial truffle-aroma molecule. Its presence could explain the smell, but no human evidence found ties culinary exposure to brain fog.
A small POTS study found high glucose worsened post-meal tachycardia and reduced stroke volume. This becomes relevant only if fog repeats after other large carbohydrate meals.
Evidence receiptMannitol absorption and gut symptomsJECFA · 2,4-dithiapentanePost-meal POTS physiology
04 · caffeine and sleep
Dose plus clock time explains the pattern.
Coffee bitterness, roast colour, milk, and brand are poor caffeine gauges.
Hong Kong milk tea, two coffees with an afternoon cup, and one MUJI tea disrupted sleep. Some TWG teas and one earlier coffee were tolerated.
Residual caffeine from an earlier drink stacks with a later dose. A useful working half-life is about 4–6 hours, with large personal variation.
Best-supported mechanism in this study. It explains sleep disruption, not the sharp scent or chicken pain.
That all coffee or tea is unsafe. Brew size, leaves/beans, method, and timing matter more than the name.
Keep the usual morning drink stable; remove only afternoon caffeine for several nights. Log milligrams when known, time, bedtime, and sleep onset.
Common sources that can quietly stack
Coffee, Hong Kong milk tea, black tea, matcha, yerba mate, energy drinks, cola, pre-workout, caffeine tablets, and some headache medicines. Chocolate contains less but can add to the total. Pour-over and light roast are not reliably lower: bean dose, drink size, and extraction dominate.
Do not stop habitual caffeine suddenly just to test this. Withdrawal can itself cause headache and fog.
Evidence receiptHong Kong CFS drink measurementsCaffeine dose-and-timing trialCaffeine-withdrawal review
05 · E102 / tartrazine
Local label overlap. Exact eaten packs unknown.
This is a real formula-level overlap, not a causal result: it does not prove E102 caused any symptom. Exact exposure remains unknown, and false-positive risk remains high.
The Fortune Hokkien example explicitly lists E102. The associated China-origin 180 g Milk and Peanut M&M back-panel photographs explicitly list Tartrazine; their Thai panels also name INS 102.
If the remembered candy matched either photographed formula, E102 would be a shared label candidate with the Fortune noodle example.
Formulation-specific candidate. Both current Singapore examples list E102; the exact eaten products are still unknown.
That E102 caused pain, that every M&M formula contains it, or that the remembered packet matched either photographed SKU.
Match only an existing packet, photo, receipt, weight, origin, or barcode. Do not deliberately test E102.
- Label names
- E102, INS 102, tartrazine, colour (102), Yellow 5, or FD&C Yellow No. 5.
- Photographed formulas
- M&M Milk 180 g (barcode 9300682001076) and Peanut 180 g (9300682001069), both product of China, list tartrazine among their artificial colours.
- Market limit
- The current European M&M formula does not list E102. That proves the ingredient is not universal; it does not erase the Singapore example-label overlap.
- Source hierarchy
- The retailer text fields are visibly malformed. The attached back-panel photographs are the evidence used here; neither is the remembered packet.
- Unresolved shell lane
- Compared with plain chocolate, the shell adds starch, glucose syrup, gum arabic, dextrin, waxes, colours, and flavourings. Current observations do not isolate one of them.
Evidence receiptFortune Hokkien formulaMilk 180 g back panelPeanut 180 g back panelMilk product pagePeanut product pageEuropean M&M formulaSingapore SFA label guideFDA reaction context
06 · smallest useful follow-up
Collect discriminating facts.
No broad diet, perfume-family ban, or deliberate symptom challenge is supported.
Product, variant, batch, restaurant dish, stock base, package photo, or ingredient panel.
Baseline symptom, exposure time, onset, peak, associated symptoms, and recovery.
Keep recording comfortable exposures. They are what crossed out ethanol, atomisation, broad poultry, and broad chocolate theories.
Evidence limits and safety boundary
This page is a de-identified evidence review, not a diagnosis. Ingredient presence, a plausible biological pathway, and personal causation are three separate claims.
A molecule comparison cannot distinguish migraine or scent sensitivity from structural, neurological, or other causes of a recent severe persistent headache with a neck event and radiating pain. If a sudden worst-ever headache, confusion, seizure, fever with marked neck stiffness, repeated vomiting, new weakness or numbness, trouble speaking or seeing, severe imbalance, bladder/bowel changes, saddle numbness, or throat/tongue swelling appears, emergency assessment is appropriate.
NICE headache guidance · Singapore HealthHub neck-pain guidance