New Update on the Alleged Allergens in Essential Oils.

Modified from a previous website article by the author. Copyright © Tony Burfield Oct 2004-July 2005.


The utilization of flawed dermatological research findings by the SCCNFP (2002), resulted in a published Opinion on alleged fragrance allergens (SCCNFP/0392/00), and its subsequent adoption by EU legislators, looking for suitable evidence to support the thrust of their Chemicals Policy in the cosmetics area. The unfortunate incorporation of this Opinion led to the 2003/15/EC Directive (7th Amendment) into the Cosmetic Products (Safety) Regulations 2004. In fact the SCCNFP Opinion (now the SCCP) on the alleged allergens does not relate to essential oils as such at all, but 16 of the 26 listed substances do happen to occur in essential oils, but the committee (incorrectly as we will see) decided not make a distinction as to the origin of the alleged allergens, reasoning that natural and synthetic allergens would be chemically identical.

You may recall that the infamous 7th Amendment to the EU Cosmetics Act (amongst other things) requires a labelling obligation by March 2005 for a final cosmetic product containing any of the 26 identified allergens present at 0.01% in products rinsed off the skin, products or 0.001% in leave-on products. This may therefore affect retailed aromatherapy massage oils and essential oil blends, creams and lotions and bath products intended for a cosmetic purpose, as well as natural and part-natural fragrances. The result of this hasty and scientifically unsound legislation has been to cause concern amongst fragrance customers, who, rather unrealistically have demanded (and continue to demand) perfumes with either no or only a few allergens present. Since, as mentioned above, essential oils and absolutes naturally contain 16 of these 26 alleged allergens, a subsequent fall in usage of essential oils and natural aromatic materials in fragrances has since occurred across Europe. The SCCNFP, in their published musings “did not consider that the production of essential oils would be threatened”, although of course they had no actual expertise in this area to safely predict this outcome. IFEAT famously disagreed (IFEAT 2002) and sure enough, annual production volumes of essential oils markedly dropped in 2000-2002, and oil-trading companies folded. 

Many other commentators and organisations made protests in print at the time of the disclosure of the original 26 alleged allergens list (e.g. Bassett (2002), Duclos (2003), Chavigny C. (2002), Roberts (2002), PCA (2002), Subrenat (2001) etc.). Basically the aroma trade was well and truly screwed by a powerful alliance of dermatologists and cosmetologists who had the ear of the EU legislators, which ironically, probably acted to the advantage of the big synthetic fragrance chemical producers. This is because essential oils and other natural aromatic raw materials are subject to unpredictable availability and price fluctuations from climatic and political factors, and a decline in their usage potentially suits modern management policies of multi-national aroma concerns, who no doubt, would prefer to base perfumery around more price-stable in-house synthetic fragrance chemicals.

Newer news

The July edition of the German consumer magazine "Öko-Test", No. 7/2004, 55, carries an article (in German) reporting on studies done by the IVDK, an information network association of dermatologists, headed up by Professor Axel Schnuch. It concludes that not all the 26 allergens identified by SCCNFP Opinion and enshrined in the 7th Amendment to the Cosmetics Act, bear the same risk, and criticises the EU commission for treating them all as equal. The report classifies allergens accordingly (those occurring in essential oils and those naturally occurring aromatics are highlighted in blue):

1. Strong potent allergens:
oakmoss, treemoss, iso eugenol and cinnamic aldehyde.

2. Less potent allergens: cinnamic alcohol, hydroxycitronellal**, HMPCC.

3. Rarely found as allergens: amyl cinnamic aldehyde, citral, eugenol, farnesol, lilial, methyl heptine carbonate. 

4.  Risk of being an allergen too small to consider: amyl cinnamic alcohol, benzyl alcohol, benzyl salicylate, geraniol, anisyl alcohol, benzyl benzoate, benzyl cinnamate, citronellol, hexyl cinnamic aldehyde, d-limonene, linalool, coumarin and alpha-keton.

** previously found by the author to be an added constituent (adulterant) of  linden blossom absolute from Tilia spp..

However the above article does not make a distinction in degree of risk between substances presented as allergens in essential oils and the same allergenic substances presented as synthetic chemicals.

The case of Coumarin.

The SCCNFP position paper on 26 (alleged) allergens identified coumarin amongst 16 of the 26 substances that occur in natural products, this also being the findings of previous COLIPA & RIFM opinions. This led to coumarin being regulated within the 7th Amendment of the EU Cosmetics Directive requiring the labelling of the ingredient, if the substance is present in conc. higher than 10 ppm (=0.001%) in leave-on products, and 100 ppm (=0.01%) in products rinsed off the skin.

Coumarin is a lactone (an ester group integrated into a carbon ring system), and is naturally found at 1-3% concentration in tonka bean (Dipteryx odorata Wild.), deer tongue (Trisilia odoratissima Walt.) in woodruff (Galium odoratum L.) and clover flowers (Melilotus officinalis & M. alba), and in trace amounts in Narcissus spp, Artemisia vulgaris, Michelia champaca, spike lavender, Artemisia annua, cinnamon bark etc. etc. (Burfield 2001). Tonka bean absolute contains some 65% of coumarin. It is also frequently found in commercial Chinese cassia oil at 4-11% concentration (Burfield 1999: unpublished data), which is seemingly at variance with some of the reported literature.

Now, a manufacturer of 99.99% pure coumarin (Rhodia), has protested (summer 2005) that pure coumarin is not an allergen, and forwarded evidence to DG Enterprises who have submitted their evidence back to the SCCP committee, for their education in this matter. The original dermatological testing work, which had been carried out by Malten K.E. et al. (1984), De Groot A.C. et al. (1988), Larsen W. et al. (1996), and Van Joost T. et al. (1985) probably used an impure synthetic form of coumarin, and which gave a false result. Floc’h (2002) heavily criticised work by the above authors, as lacking scientific rigor, found no statements of the purity of the materials used in their work, and questioned the homogeneity and the stability of the coumarin in petrolatum suspension (used for dermal application). Floc’h further indicated the work failed to distinguish allergy to coumarin and cross-reaction to allergens for which coumarin might be an indicator – overall these remarks add up to a pretty serious shortfalls in the standard of previous work carried out, and it is a complete mystery why the SCCNFP 'expert ' committee was apparently unable to spot that the coumarin studies were deeply flawed.

Hopefully with this new departure regarding coumarin, we will now see an investigation into the why the SCCNFP listed some of the other substances as alleged allergens. An apology to the aroma industry for associated loss of confidence, lost sales, and compensation for those innocent European workers who lost their jobs through no fault of their own would not be amiss either.

The case of Frosch et al.’s work (2002) on fragrance sensitisers

It may be worth looking at a by Frosch et al. (2002) in some depth, since it is typical of publications which seemingly much have weight with ‘expert’ committees in the cosmetics area. Interestingly, following the publication of this work, as yet no further set of alleged allergens has been outlined in any consequent SCCP Opinion.

Frosch et al. (2002) further their previous work on fragrance sensitisers [Frosch et al. (1999), Frosch et al. (2002a)] by testing a further eighteen natural aromatic materials across six dermatological centres in Europe, this time mainly employing ‘essential oils’ in petrolatum (eleven out of eighteen substances tested), in order to supplement the current fragrance mix of Larsen (1977). Altogether a total of 1606 dermatology patients were consecutively tested with these substances and with the current fragrance mix.


The authors collectively refer to essential oils as including absolutes by inference throughout the text. Absolutes are solvent extracts, not essential oils.

The authors initially state that essential oils “such as ylang-ylang or sandalwood have gained special attention in aromatherapy. They are often applied directly to the skin in high concentration …”. In fact most professional and teaching organisations in aromatherapy recommend from a 0.5% (children, the elderly) to 2.5% (mature adults) concentration of essential oil in vegetable carrier oil when used for aromatherapy massage, with the use of neat oils not recommended. It is true that aromatherapists are reported as applying undiluted essential oils to the skin in certain ‘minor emergency’ situations – tea tree oil for small skin traumas, lavender oil for very minor burn areas, cajeput or niaouli oils for insect bites, stings etc. etc. If ylang ylang oil is used in aromatherapy, most holistic therapists would choose ‘ylang ylang oil complete’ out of choice, not the fractions selected by the team. In reviewing the components of ylang-ylang oil the authors describe the botanical nomenclature as ‘Canaga odourata’ (sic), which of course corresponds to Cananga oil.

No botanical origins to any of the ‘essential oils’ used for the actual testing were verified by the authors of the article, although the ‘leading manufacturers’ supplying the samples, and their respective countries of origin are stated. None of the substances were analysed by the authors for authenticity, but instead literature references apparently describing typical compositions are presented. These two facts – no defined botanical origin of the substances under study, and no precise (independent) chemical analysis, are enough to regard this paper as showing insufficient scientific rigor from which to draw safe conclusions.

It gets worse. In the case of patchouli oil Indonesian, an old discredited literature reference is used claiming that cinnamic aldehyde, benzaldehyde and eugenol are components found in the oil (!) – these items are, of course gentle reader, not the normal constituents of Patchouli oil. The authors similarly reproduce unreliable data when they report that a 50:50 mixture of Chinese cedarwood oil and Moroccan (Atlas) was used in the tests, claiming that a-ionone as a constituent of Atlas cedarwood oil (!).  It is unclear why such an unusual mixture in was used in these tests anyway. The authors go on to report the major constituents of Texan and Virginian cedarwood oils (which weren’t used in the study) but fail to report the constituents of Chinese cedarwood oil (which was used as a mix ingredient).

The choice of substances chosen for the study reflects (in my opinion) a trend for dermatologists to rush into print with alarming positive sensitisation results for fragrance ingredients. It would have been more expedient however to take a cool look at the ingredients selected, with respect to their respective frequencies of occurrence in perfumes: peppermint and spearmint oils aren’t exactly common fragrance ingredients of most modern perfume compounds, and usage of narcissus absolute (also one of the eleven “oils”) is relatively rare. Two grades of ylang ylang oil were used in the experimental work – I and II. On reflection it would have probably been more useful to have used ylang ylang extra and ylang ylang III to illustrate the any difference in results between a relatively oxygenated-substance rich oil (ylang ylang extra) and a more sesquiterpene-rich oil (ylang ylang III) – but the authors state the extra grade was not tested because of its high price (!!).

The text indicates the current fragrance mix applied at 8% gave a mean of 11.4% positives. In comparison a whacking 10% concentration of Ylang ylang I essential oil applied to the skin gave 2.6% positives, and Ylang ylang II 2.5% positives. I suppose it might – but you wouldn’t usually find these equivalent levels in a hydroalcoholic fine fragrance containing, say, 20% perfume concentrate. The remaining results are:

Lemongrass oil (East India) applied at 2% gave 1.6% positives – but “doubtful or irritant reactions” were placed at 2.2%.

Narcissus absolute at 2% giving 1.3% positives.

Jasmine absolute at 5% gave 1.2% positives.

Similar remarks apply here: an applied hydroalcoholic fragrance giving a 5% jasmin absolute concentration to the dermis would be beyond the purchasing limits of most fragrance buyers. 

The results which fall below 1% positives:

Sandalwood oil E.I. at 10% gave 0.9% positives

Patchouli oil Indonesian at 10% gave 0.8% positives

Spearmint oil at 2% gave 0.8% positives

Dwarf Pine needle oil 2% (we aren’t told which one) gave 0.7% positives.

Cedarwood oil 50:50 mix Atlas: Chinese at 10% gave 0.6% positives

Limonene at 3% (enantiomeric purity and peroxide value not stated – presumably dextro-) gave 0.6% positives

Peppermint oil (US) at 2% gave 0.6% positives

Sandalwood oil E.I. at 2% gave 0.4% positives

“Alpha”-terpineol natural (geometrical isomer composition and optical purity not confirmed) at 5% gave 0.1% positives.

Interestingly, limonene applied at 3% concentration fails to deliver a positive result over the 1% level, in spite of it being listed as one of the 26 alleged allergens in the 7th Amendment to the EU Cosmetics Act. It is probable of course that the impurities in the limonene (hydroperoxides or oxidation breakdown products) are the actual cause of the sensitisation properties, rather than the pure substance itself.

The authors give an extended discussion over 4-5 pages regarding the significance of their findings, concluding that the fragrance mix should include at least 4 essential oils producing positive test results in patients at a frequency of over 1%, these being Ylang Ylang oils I & II, lemongrass oil, narcissus absolute and jasmine absolute. The ‘marked variations’ between the results of the assessment centres are not satisfactorily explained by the authors, and may hinge on the individual assessors giving differing interpretations of the intensity of any observed reactions, or that the results are purely related to how well the substances have been stored.

In conclusion the authors state that “if intensive human skin exposure to ylang ylang I & II, lemongrass oil, narcissus absolute, jasmine absolute and possibly sandalwood & patchouli oil as found in aromatherapy (?), then contact sensitisation may develop”. This smacks of fear culture. No-one I have spoken to in the aromatherapy professional organisations has ever noticed this trend at normal usage levels, or seems to have been approached by the named authors on this subject. In any case narcissus absolute is almost unused in aromatherapy, sandalwood oil E.I. is scarce, and where obtainable, is almost inevitably purchased adulterated with other woody oils. The relative expense of jasmin absolute to aromatherapists means that it is less commonly used than other oils (such as lavender or rosemary), and that lower concentrations are often employed.

In conclusion I believe that the end-user experience of aromatherapy professionals using these stipulated essential oils, data for which to this authors’ certain knowledge is currently being collated in two or more on-going schemes, will not bear out a hypothesis concerning adverse levels of sensitivity at normal levels of essential oil use as is floated in this paper.

Further, from the high number of technical gaffs appearing in articles such as this, it is quite apparent that dermatology investigative teams such as the one above need some expert input from a member with experience in aroma materials and the normal practice of aromatherapy and perfumery, instead of relying on erroneous reports in the literature.


Bassett F. (2002) “Natural Products: Towards a Common Cause” Parfums Cosmétiques Actualités 168 p32-35.

Burfield T. (2001) Natural Aromatic Materials – Odours and Origins pub. AIA, Tamoa USA.

Chavigny C. (2002) Parfums Cosmétiques Actualités No 166 Sept 2002 pp 50-51; 54.

Duclos T. (2003), president of EFEO, quoted in Landousy M-T. (2003) “Clouds on the Horizon” Parfums Cosmétiques Actualités Dec 2003 No. 174 p69.

Floc’h, François (2002) “Coumarin in Plants and Fruits: Implications in Perfumery.” Perf. & Flav. 27 (Mar/Apr 2002), 32-36.

Frosch P.J., Johansen J.D  &  Menné T. (1999) “Lyral is an important sensitiser in patients sensitive to fragrances” Br. J. Dermatol 141, 1076-1083.

Frosch P.J., Johansen J.D., Menné T., Pirker C., Rastogi S.C., Andersen K.E., Bruze M., Goosens A., Lepitoittevin J.P. & White I.R. (2002) “Further important sensitisers in patients sensitive to fragrances” II. Reactivity to essential oils.” Contact Dermatitis 47, 279-287.

Frosch P.J. (2002a) “Further important sensitisers in patients sensitive to fragrances” Contact Dermatitis 47, 78-85.

de Groot, A.C. et al. (1988) “Allergens in Cosmetics” Arch. Dermatol. 124, 1525-1529.

IFEA (2002) Private Communication to all members IFEAT/20/02

Larsen  W. et al. (1996) “Fragrance contact dermatitis. A worldwide multi-centre investigation (part 1).” Am. J. of Contact Dermatitis 7, 77-83.

Malten K.E. et al. (1984) “Reactions in Selected Patients to 22 fragrance materials” Contact Dermatitis 11, 1-10.

Roberts G (2002) “The Good Times are Over: and They Are Not Coming Back” Parfums Cosmétiques Actualités  166, p 51-59.

PCA (2002) Parfums Cosmétiques Actualités No 167 Oct-Nov 2002 p87.

Subrenat J-P (2001) “European Parliament Regulates Fragrance Ingredients” as featured in “Industry Insight” Perfumer & Flavourist 26, 24-28 (2001).

Van Joost T et al. (1985) “Simultaneous allergy to perfume ingredients” Contact Dermatitis , 115-116 (1985)