Porecelain

Flawless skin starts with the fine print.

Botanical extracts

Four ingredients sit in this family and exactly one of them has a comedogenic score in the numeric documents we hold. It is allantoin, at 0 on Fulton's 0 to 5 scale, and allantoin is the only member that is not plant material at all - the Cosmetic Ingredient Review records that its source in cosmetics is synthetic. The three actual plants carry low numbers on consumer checkers that we could trace to no primary document. What they do carry, and most untested ingredients do not, is randomised human trials on acne-prone faces - because they were being tested as the cure rather than as the cause.

4 rated 0 Clogging 0 Caution 4 Clear 3 never assayed

What they are

Materials taken from a plant, or in one case filed alongside them. Four members: aloe barbadensis leaf juice, centella asiatica extract, tea tree oil and allantoin. They are in a formula as skin-conditioning or blemish-directed ingredients rather than as emollients, bases or emulsifiers, and none of them is there to build the oil phase. THE FOUR ARE NOT ONE KIND OF MATERIAL, and the differences decide most of what follows. Aloe barbadensis leaf juice is a leaf gel; the inner-leaf gel it is normally prepared from is described as more than 98% water, which a concentrate or a dried powder is not, so the figure belongs to a form rather than to the label name. Its small solid fraction is mostly polysaccharide rather than lipid. Centella asiatica extract is a solvent extract of a herb, standardised loosely on a group of four pentacyclic triterpenes whose proportions move with plant part, growing conditions and solvent. Tea tree oil is a steam distillate - a volatile mixture of roughly a hundred terpenes, not a fat - and it is the one member for which a compositional standard exists, ISO 4730 - which sets limits for the named material and is not a statement that any given batch conforms to them. Allantoin is none of these. It is a single small heterocyclic compound, CAS 97-59-6, a hydantoin derivative, and CIR's 2010 assessment states plainly that "its source in cosmetics is synthetic." We file it here; that is a shelving decision about where a reader will look for it, not a claim about where it comes from, and the allantoin page says so too. SO THE FAMILY IS THREE PLANT MATERIALS AND ONE MANUFACTURED MOLECULE, which is worth holding on to, because the member with the score is the molecule. A GREAT DEAL OF PLANT MATERIAL IS NOT IN THIS FAMILY and will not appear in the table below. Coconut, olive, argan, rosehip, sunflower and jojoba are filed as OILS; cocoa and shea as BUTTERS; candelilla and carnauba as WAXES; algae extract, algin and carrageenan as ALGAE; tocopherol and ascorbic acid as ANTIOXIDANTS; bakuchiol, which is a plant phenol, as an ACTIVE. The member table on this page renders from our ratings file, so it will show four rows and only four, and nothing on this page is a claim about any plant-derived ingredient we file elsewhere. "BOTANICAL" IS OUR CATEGORY, NOT FULTON'S, and his nearest equivalent is a different set. Our transcription of his 1989 table records a heading reading "X. Vitamins and herbs" and does NOT record which rows sit under it. What it records is the printed order, and in that order a run of vitamin rows - A and D additive, tocopherol, tocopheryl acetate - is followed immediately by Black walnut extract, Papain and Chamomile extract, and then by Vitamin A palmitate and Panthenol. None of our four members appears anywhere in the table. Where this page names one of his rows it names it as his, never as ours, and we rate none of them.

What the evidence actually says

Two things are printed, and together they are the reason this family has a page. FIRST, WHERE THE ONE NUMBER ACTUALLY IS. Fulton's 1989 survey is the 220-row rabbit-ear table nearly every comedogenic number now in circulation descends from, and we hold it transcribed verbatim off a 400dpi render of the journal scan. It contains a row reading "Allantoin" at a comedogenicity grade of 0 on his 0 to 5 scale. No range, no asterisk, and the row sits outside the asterisked oils section whose footnote - "results depend on source of raw material" - governs every oil and butter he printed and does not reach this one. That number is the right number for the right molecule, which is rare enough on this site to be the finding rather than the throat-clearing. It is also the only comedogenic grade any member of this family has in the numeric documents we hold, and it belongs to the member that is manufactured. For the other three we searched and came up empty, and the honest form of that is bounded. We identified no comedogenicity assay of aloe barbadensis leaf juice, centella asiatica extract or tea tree oil in the corpus we searched, which is the nine numeric documents indexed at /how-we-decide#corpus. Fulton's 1989 table has no aloe, centella or Melaleuca row. His 1984 paper, whose Table I we have read in full on its own different key, has none. Morris and Kwan's 1983 assay, an independent laboratory on the same 0 to 5 scale that contradicts Fulton in print, tested thirty-two materials, all of them lanolins, vegetable oils, esters or surfactants - nine of its rows are vegetable oils, so plant material was plainly not out of its reach - and no member of this family is among them. Nguyen and Maibach's 2007 rabbit-ear paper names fourteen materials in the abstract we hold, which is all of that paper we have read, and no member of this family is among those either. Consumer checkers do print low numbers against all three, and we traced no primary document behind any of them. We do not know who entered those numbers or why, and we are not going to guess. SECOND, AND THIS IS THE PART A READER CANNOT GET FROM THE MEMBER PAGES ONE AT A TIME - THE ABSENCE IS NOT EXPLAINED BY PLANTS BEING OUT OF SCOPE. Fulton put botanical extracts on a rabbit's ear. In the printed table, the three rows that run together immediately after the vitamin rows are Black walnut extract at 0 of 5, Papain - an enzyme from papaya - at 0 of 5, and Chamomile extract at 0 of 5, in that order. Which heading they sit under our transcription does not record. Those are his rows, not ours; we rate none of the three. But they establish that the gap where our four would sit is not a category he declined to test. What those two extract rows are worth is the family-level point. "Black walnut extract" and "Chamomile extract" are the entire specification Fulton prints: no species, no plant part, no solvent, no concentration of the extract itself, no supplier. For an ester or a fatty alcohol the name on the label is close to a statement of composition, so a score attaches to something another laboratory could obtain. For a plant extract the name is a plant plus a process, and two materials sold under one INCI name need not be the same material. So the two botanical readings the corpus does contain are numbers that cannot be reattached to anything a formulator could buy. Our centella page makes this argument about centella. The table shows it is a property of the category. AND THE EXCEPTION, IN THE SAME BREATH, BECAUSE IT CUTS AGAINST US. Tea tree oil is the member the naming argument does not fit. A compositional standard exists for it, ISO 4730, and in the edition Carson and colleagues summarised in 2006 that standard set minima and maxima for fourteen components, terpinen-4-ol at 30% or more and 1,8-cineole held low. A standard is a specification a material can be sold against, not a guarantee that a given batch met it. It is the one plant member of this family that could be specified tightly enough to assay reproducibly, and we identified no comedogenicity assay of it. Whatever keeps this family out of the rabbit-ear literature, it is not only that the materials are hard to pin down. THIRD, AND IT RUNS THE OTHER WAY. The absence of an assay here is not an absence of human exposure, and this is the sentence a page like this one gets wrong. Two of the four members have been applied to acne-prone human faces in randomised trials, and the lesion counts fell. Bassett and colleagues (1990) ran a single-blind trial in 124 patients, 5% tea tree oil gel against 5% benzoyl peroxide lotion, and both arms significantly reduced inflamed and non-inflamed lesion counts, the non-inflamed count being open and closed comedones. Enshaieh and colleagues (2007) ran a double-blind placebo-controlled trial in 60 patients over 45 days, and the tea tree arm's mean total lesion count fell from 21.16 to 11.33. Hajheydari and colleagues (2014) randomised 60 patients, double-blind, to 0.05% tretinoin cream plus 50% aloe vera gel or to tretinoin plus vehicle for eight weeks, and the aloe arm reduced non-inflammatory, inflammatory and total lesion scores significantly more than tretinoin alone. Allantoin has no such trial we identified. We name no centella trial here either, because we have graded none to the standard the three above are held to. Every one of those trials measured treatment - whether an existing crop of lesions shrank - and not induction, which is what "comedogenic" asks. None is a comedogenicity measurement and we will not launder one into one. But the shape is the family's: these plants have been on more acne-prone faces than most of what we rate, and the reason is that they are sold to treat the condition a pore-clogging checker is warning about. The evidence in this family runs backwards, and that is why its numbers are missing rather than disputed.

What the evidence doesn’t tell you

Every reassuring thing above is weaker than it reads, and we would rather say so than let a page about untested plants become an argument that untested means safe. Unproven is not disproven, in either direction. THE ONE SCORE IS ONE READING. Allantoin's 0 comes from three rabbits' ears in one paper, and Fulton's own key says that a grade of 0 to 1 "is not considered significant" - the floor of the band he treated as noise rather than as a result. A 0 from the rabbit ear is as weak as a 4 from it, and this site discounts the model when it condemns an oil. Consistency requires discounting it here. Fulton called the whole survey "not at all definitive but simply designed to stimulate research." A ZERO IS A NUMBER TOO, AND NOBODY AUDITS A ZERO. The point of the first section is not that the plants are secretly bad. It is that three of the four carry a reassuring figure on the checkers that we could trace to no measurement, and that a low number sitting in the same column as coconut butter's measured 4* - the asterisk being Fulton's oils-section footnote for source of raw material - invites a reader to treat an untested ingredient as a tested and cleared one. The correction here happens to be reassuring in shape and is not reassuring in content. THE TWO EXTRACT ROWS ARE FULTON'S, NOT OURS, AND THEY DO NOT TRANSFER. Black walnut extract at 0 of 5 and chamomile extract at 0 of 5 are evidence that he tested botanicals. They are not evidence about aloe, centella or tea tree oil, and reading one plant's score across to another is the exact substitution this site exists to catch. We rate neither material and we carry neither number. THE HUMAN TRIALS ARE SMALL, AND THEY MEASURED SOMETHING ELSE. One hundred and twenty-four patients, sixty, sixty. One single-blind. All three tested finished 5% or 50% formulations rather than the isolated ingredient, and a result about a formula is not a result about a molecule - Hajheydari's two arms both received tretinoin, so what the randomisation supports is an incremental effect from adding the aloe gel, not a comedogenicity result for aloe. None counted comedone induction. A substance could in principle reduce inflammatory lesions while nudging comedone formation, and none of these trials was built to catch that. THE FAMILY-LEVEL CLAIM IS ITSELF THE KIND OF CLAIM WE DISTRUST. Four members is a small family, one of them is not a plant, and a generalisation across three plant materials with nothing in common chemically is a filing decision talking. What is documented is narrow: one printed row, two of his rows for materials we do not rate, three absences bounded to the corpus we searched, and three treatment trials. Everything wider than that is us. The corpus is open - a ninth numeric paper turned up after we had called the list complete once - so an absence here records what we looked at, not what exists. Finally, a 2025 review of the field concludes that no standardised comedogenicity test exists and that "non-comedogenic" is an unregulated claim, which sets the ceiling on how much any of these numbers, present or absent, can be asked to carry.

All 4 botanical extracts we rate

The legacy score is printed exactly as it appears in the paper, including the ranges, the asterisks and the parentheses. Those marks are not decoration. They are the paper telling you what its own number is worth.

Sources

  1. Fulton JE. "Comedogenicity and irritancy of commonly used ingredients in skin care products." J Soc Cosmet Chem 40:321-333 (1989) The 220-row rabbit-ear survey nearly every comedogenic number in circulation descends from, read off a 400dpi render of the journal scan rather than through a list quoting it. Method: the ingredient at roughly 10% in propylene glycol, applied to the inner ear of three New Zealand albino rabbits, five days a week for two weeks, follicles scored for keratosis on a 0 to 5 scale whose own key reads 0 to 1 "not considered significant", 2 to 3 "borderline", 4 to 5 "considered positive". Three things on this page come out of it and out of nothing else: the Allantoin row at 0; the heading our transcription records as "X. Vitamins and herbs", one of only two headings that transcription holds and not a membership list; the three rows that run together immediately after the vitamin rows, Black walnut extract 0, Papain 0 and Chamomile extract 0, in that order; and the verified absence of any aloe, centella or Melaleuca row across all 220. Fulton called the survey "not at all definitive". It supports no rating of the three plants in either direction.
  2. Morris WE, Kwan SC. "Use of the rabbit ear model in evaluating the comedogenic potential of cosmetic ingredients." J Soc Cosmet Chem 34:215-225 (1983) An independent laboratory's rabbit ear-canal assay on the same 0 to 5 scale, three rabbits, materials undiluted unless stated, fourteen applications, graded microscopically. It predates Fulton's 1989 table and contradicts him in print on several materials, which is why it is not a republisher of him. Cited here purely for an absence, and for the shape of that absence: its thirty-two test materials are seven lanolins, nine vegetable oils, eleven esters and five surfactants, so the second primary assay in the corpus reached plant material only as pressed oils and reached no plant extract at all.
  3. Cosmetic Ingredient Review Expert Panel (Becker LC, Bergfeld WF, Belsito DV, et al.). "Final Report of the Safety Assessment of Allantoin and Its Related Complexes." Int J Toxicol 29(3 Suppl):84S-97S (2010) The industry expert-panel SAFETY dossier on allantoin, in its final published state, covering an ingredient reported in 1,376 cosmetic products. Cited for one sentence of identity - "its source in cosmetics is synthetic" - which is the fact that makes this family's only measured member its only manufactured one, and for the names that place it chemically as a hydantoin derivative. It is a safety assessment: the words comedogenic, comedone, acne, sebum and pore do not appear in the full text we searched, and its silence on pores is not evidence about pores.
  4. Bassett IB, Pannowitz DL, Barnetson RS. "A comparative study of tea-tree oil versus benzoyl peroxide in the treatment of acne." Med J Aust 153(8):455-458 (1990). PMID 2145499 A single-blind randomised trial in 124 patients, 5% tea tree oil gel against 5% benzoyl peroxide lotion. Both arms significantly reduced inflamed and non-inflamed lesion counts, the non-inflamed count being open and closed comedones; benzoyl peroxide acted faster and more strongly and tea tree oil produced markedly fewer side effects. Characterised precisely, because the precision is the point: this is a TREATMENT-efficacy trial on people who already had acne, not a comedogenicity assay measuring comedone induction, and it tested a 5% gel rather than the ingredient alone.
  5. Enshaieh S, Jooya A, Siadat AH, Iraji F. "The efficacy of 5% topical tea tree oil gel in mild to moderate acne vulgaris: a randomized, double-blind placebo-controlled study." Indian J Dermatol Venereol Leprol 73(1):22-25 (2007). PMID 17314442 A randomised, double-blind, placebo-controlled trial in 60 patients over 45 days, 5% tea tree oil gel against a vehicle placebo, with total lesion count and acne severity index as endpoints. The tea tree arm's mean total lesion count fell from 21.16 to 11.33. Again a treatment trial and not a comedogenicity study: the endpoint is the reduction of existing lesions, and no comedone induction was measured. Cited as a human proxy and only as a proxy.
  6. Hajheydari Z, Saeedi M, Morteza-Semnani K, Soltani A. "Effect of Aloe vera topical gel combined with tretinoin in treatment of mild and moderate acne vulgaris: a randomized, double-blind, prospective trial." J Dermatolog Treat 25(2):123-129 (2014). PMID 23336746 A randomised double-blind trial in 60 patients with mild-to-moderate acne, twice daily for eight weeks, comparing 0.05% tretinoin cream plus 50% aloe vera gel against tretinoin plus vehicle. The aloe arm reduced non-inflammatory, inflammatory and total lesion scores significantly more than tretinoin alone, with less erythema. Both arms received tretinoin, so the randomised difference is attributable to the added aloe gel - and what was measured is lesion reduction under treatment, not comedone induction by aloe on its own. Evidence that adding a large amount of aloe gel to a regimen did not worsen acne-prone skin. Not a comedogenicity measurement.
  7. Carson CF, Hammer KA, Riley TV. "Melaleuca alternifolia (Tea Tree) Oil: a Review of Antimicrobial and Other Medicinal Properties." Clin Microbiol Rev 19(1):50-62 (2006) The standard peer-reviewed review of the oil, from a University of Western Australia group that has published extensively on it - an affiliation worth stating. Cited for composition and for nothing else: it sets out the ISO 4730 specification as it stood in the edition this review summarises, fourteen components with terpinen-4-ol the major one at roughly 30 to 48% and 1,8-cineole held low. That such a specification exists is the reason tea tree oil is the exception to this page's argument about extract naming; whether a given batch conforms to it is a separate question, and the review does not address that. The review says nothing about comedogenicity or pores.
  8. Starzyk T, Aust N, Schur N, Miller R. "Comedogenicity in cosmeceuticals - a review of clinical relevance, regulatory gaps, and future directions." JAAD Reviews 6:78-83 (December 2025) A peer-reviewed clinical review of the field from 1972 to 2025. It is a synthesis of the same legacy literature rather than a new measurement, and independent of Fulton only in authorship. Cited for one field-wide conclusion we take seriously: no standardised comedogenicity test exists, and "non-comedogenic" is an unregulated claim. It sets the ceiling on what any number on this page, present or absent, can be asked to carry. Its Table I prints rabbit-ear scores for a list of materials and the publisher has intermittently blocked the full text from us, so we do not assert what that table does or does not contain for these four ingredients.
  9. Consumer comedogenic checkers and ingredient databases (CosDNA and similar) Aggregators that publish a 0 to 5 "acne" number per ingredient with no disclosed method, provenance or citation. CosDNA lists Centella Asiatica Extract at 1 on its own scale, and checkers of this kind list aloe at 0. Cited as the artefact being corrected rather than as evidence: we traced no primary assay behind any of these figures and make no claim about where they came from or who entered them. They are not carried as legacy values in our ratings file, and they are republishers rather than laboratories, so on an ingredient with no assay behind it they cannot be republishing a measurement.

‹ All ingredients