Ethoxylates (PEGs & polysorbates)
Everything this page can show comes out of one table. Fulton's 1989 survey prints five ethoxylation series, and we have checked every rung of every one of them against a transcription of the journal scan. Four of the five come down at the top of their range as the ethylene-oxide count rises, and the two that print a short rung go up before they come down: steareth-2 scored 2 of 5, steareth-10 a 4, steareth-20 a 2, steareth-100 a 0. None of it is a dose-response curve. Each rung is a different ingredient, read once, by one investigator, in one paper, and we identified no second laboratory that read any of them.
What they are
An ethoxylate is a fatty molecule with a chain of ethylene-oxide units bolted onto it. Take a fatty alcohol — lauryl, oleyl, stearyl — and add ethylene oxide, and you get a laureth, an oleth, a steareth. Do the same to a fatty acid and you get a PEG stearate; to lanolin and you get a laneth or a PEG lanolin. The point of the exercise is water-solubility: the ethylene-oxide chain is water-loving, the fatty tail is not, and the ratio between them is what makes the molecule an emulsifier. The Cosmetic Ingredient Review's 2012 assessment of the alkyl PEG ethers describes this family as the reaction products of a fatty alcohol and ethylene oxide, functioning as surfactants and emulsifiers. THE NUMBER IN THE NAME IS THE THING TO UNDERSTAND, because the whole of this page turns on it. Steareth-10 means stearyl alcohol carrying, on average, ten ethylene-oxide units. It is a nominal average across a distribution of chain lengths, and it does not describe the full commercial distribution behind the name. So a "series" — oleth-3, oleth-5, oleth-10, oleth-20 — is a ladder of averages, which is both what makes it readable as a trend and what makes the trend impossible to pin down. We file six ingredients here, and their Fulton scores run the length of his scale: Laureth-4 at 5 of 5, Oleth-3 at 5, Steareth-10 at 4, PPG-5-Ceteth-10 Phosphate at 4, Laureth-23 at 3, PEG-100 Stearate at 0. This family is unusual on this site in one respect worth stating plainly. Most of the scores we take apart belong to some other material — a neighbouring row, a derivative, a dropped word. All six of these are exact-name rows in Fulton's Table I, read off a transcription of the journal scan. The problem in this family is not the numbers. It is the shape they get summarised into. ONE MEMBER IS NOT ON A LADDER AT ALL. PPG-5-Ceteth-10 Phosphate is propoxylated as well as ethoxylated and carries a phosphate head; Fulton scored it 4, and there is no series of it in his table to read that 4 into. Nothing on this page about the shape of the ethoxylation curve applies to it. And the ladders themselves cross our families, which is worth saying before the prose starts naming things the member table below does not contain. We file PEG-16 Lanolin under LANOLIN; Polysorbate 20 under EMULSIFIER; oleyl alcohol and cetearyl alcohol under FATTY-ALCOHOL. PEG 75 lanolin, which Fulton scores 0, belongs to no family on this site at all. Every one of those is quoted below, because the family-level fact is only visible if you read across them, and not one of them will appear in this family's list of members. Fulton's table does not group by our families either: the lanolin ethoxylates sit among the lanolin rows near the start of our transcription, while the laureths, steareths and oleths run together in one unbroken stretch much further down. His order and our families are different things, and we mean ours unless we say otherwise.
What the evidence actually says
What this family has is one table, and the honest way to use it is to read the series out of it rather than to summarise them. Fulton's 1989 Table I was transcribed row by row from a 400dpi render of the journal scan, and every number below is one of those rows, on his 0-to-5 scale. THE STEARETHS are the clearest case, because four rungs are printed and none is missing. Steareth-2 scored 2, Steareth-10 scored 4, Steareth-20 scored 2, Steareth-100 scored 0. It goes up, peaks in the middle and comes down. The qualification belongs in the same sentence as the shape and not in a footnote under it: those are four different ingredients, each read once, each at the same assay concentration. It is not a dose-response curve, and nothing in the table turns it into one. THE OLEYL LINE prints the same shape and one rung more. The parent oleyl alcohol scored 4; Oleth-3, the same alcohol carrying about three ethylene-oxide units, scored 5; then Oleth-5 a 3, Oleth-10 a 2, Oleth-20 a 1. Five ingredients, one reading each, and the same caution attaches to the shape they make. THE LAURETHS AND THE PEG STEARATES print a descent and nothing underneath it. Laureth-4 scored 5 and Laureth-23 scored 3. PEG 8 stearate scored 3, PEG 20 stearate 1, PEG 100 stearate 0. There is no lauryl alcohol row and no unethoxylated stearate row anywhere in the table to set beneath either series, so for these two the rise is not absent from the chemistry — it is absent from the paper. THE LANOLIN ROWS print a rise, and the top of it is a rung Fulton discounts himself. Anhydrous lanolin is 0-1* — the asterisk is his own footnote, results depend on the source of the raw material — Laneth-10 is 2, and PEG 16 lanolin is 4. He states the rise in prose as well as printing it: "Among the lanolins, the classic anhydrous lanolins are not as comedogenic as the moderately ethoxylated derivatives (laneth 10)." But of the top rung he writes that PEG 16 lanolin "is quite comedogenic and irritating, perhaps secondary to the combination of nonlanolin additives: ceteth-16, oleth-16, and steareth-16." He is attributing that reading to what came with the material rather than to the ethoxylated lanolin, so the 4 is doing less work than it looks like it is doing. Further along the same range there is a PEG 75 lanolin row at 0. We print it as a row rather than as the far end of a curve, because the rung it would be measured down from is the one he has just explained away. So four of the five series come down at the top of their range — the oleths from 5 to 1, the laureths from 5 to 3, the PEG stearates from 3 to 0, the lanolins from a discounted 4 to 0 — and the two that print a short rung go up before they come down. That is the family-level fact, and its bounds belong in the same breath as the fact: single readings, one investigator, one paper, no replication, and every rung a different ingredient rather than one ingredient at several strengths. We did not survey the sites that republish these numbers for this page, so we will not tell you what they print. What can be said is that a chart listing one rung's score carries none of this with it, because the shape exists only if the series is read together. Two things keep that shape from being a rule, and they belong here rather than at the foot of the page. FIRST, the fall does not reach the floor. Laureth-23, with about twenty-three ethylene-oxide units, still scored 3 — the top of the band Fulton's own key calls borderline. PEG 200 dilaurate also scored 3, and we report that row without leaning on it: his table uses the molecular-weight convention elsewhere, printing "Polyethylene glycol (PEG 400)", and on that convention a PEG 200 is a short chain rather than a long one. The row does not tell us where a heavily ethoxylated material lands. SECOND, chain length is plainly not the only variable, and the table prints its own counter-examples. PEG 100 stearate scored 0 while PEG 100 distearate — the same chain carrying a second stearate group — scored 2. Ceteareth-20 scored 2 on its own, cetearyl alcohol scored 2 on its own, and the blend Fulton prints as "Cetearyl alcohol + ceteareth 20" scored 4. Those rows sit in the same table as the ladders, and we are not going to explain them. Then the key, printed here once for the whole family because the charts drop it and because a family page is the place to print it once instead of six times. Fulton: "a minimal grade of 0 to 1 is not considered significant. Grade 2 to 3 is borderline. However, a grade of 4 to 5 is uniformally reproduceable and considered positive." So of our six, Laureth-4 and Oleth-3 at 5 and Steareth-10 and PPG-5-Ceteth-10 Phosphate at 4 are in his positive band; Laureth-23 at 3 is borderline, which is neither a warning nor an all-clear; PEG-100 Stearate at 0 is the clean bottom. And a 5 is defined by a reference substance rather than measured against nothing: "Grade 5 is the presence of large comedones throughout the ear, similar to those induced by the application of our standard 'positive' testing agent, isopropyl myristate." That substance is an ester, and it is not in this family. LAST, AND PLAINLY, MOST OF THESE RUNGS ARE NOT MEMBERS. Twelve of the rows quoted above are Fulton rows we hold no rating for: Steareth-2, Steareth-20 and Steareth-100; Oleth-5, Oleth-10 and Oleth-20; PEG 8 stearate, PEG 20 stearate and PEG 100 distearate; Ceteareth-20 and the blend printed as "Cetearyl alcohol + ceteareth 20"; and Laneth-10. PEG 200 dilaurate is a thirteenth. We report them as printed rows, we rate none of them, and not one of them appears in the member table below.
What the evidence doesn’t tell you
All of it is one table. Every ladder above — the oleths, the steareths, the laureths, the PEG stearates, the lanolin ethoxylates — appears in Fulton 1989 and, as far as we can establish, in no second reading. We hold his earlier 1984 rabbit-ear table in full, transcribed from a print copy, and it contains no oleth, steareth or PEG-stearate row of any length; the only overlap is Laureth-4, which it scores 4 rather than 5 and marks with its concentration asterisk (diluted in mineral oil or made up as a 10% alcohol gel), and an entry called "Ethoxylated lanolin" at 3, a name that matches none of the 1989 lanolin rows. The 1984 paper also runs a different key — there, 1 to 2 is not significant and 3 or above is positive — so its numbers cannot be laid alongside 1989's without carrying both. We identified no second laboratory that read any of these series. The one place two laboratories overlap on an ethoxylate is small and worth stating exactly, because it is all the independent corroboration this family has. Morris and Kwan's 1983 assay, from a different laboratory and on the same 0-to-5 scale, tested thirty-two materials, of which two are ethoxylated nonionics: Octoxynol-9 at 50% aqueous, scored 0-1, and Polysorbate 20 at 1% aqueous, scored 0. Fulton's own Polysorbate 20 row is also 0. That is one name scored 0 in two laboratories at concentrations five to ten times apart, which is less than corroboration and more than nothing — and polysorbate 20 is not one of this family's members (we file it as an emulsifier) and is not a rung on any of the ladders above. Octoxynol-9 is an alkylphenol ethoxylate, a different chemistry again, and we hold no comparison reading for it. And every number on this page came out of a rabbit's ear. The method, stated plainly: the ingredient mixed in propylene glycol at nine to one, 1 ml applied once daily to the entire inner surface of one ear of a New Zealand albino rabbit, five days a week for two weeks, three rabbits per assay, the other ear untreated as control, follicular keratosis then measured with a dial micrometer and confirmed on biopsy. Fulton wrote in the same paper that "not everything that irritates this model will also irritate human skin." We identified no human comedogenicity study of any of the six ingredients in this family — no comedone count, on a face or a back, for laureth-4, laureth-23, oleth-3, steareth-10, PPG-5-Ceteth-10 phosphate or PEG-100 stearate — and no human study of the ethoxylation series as a series. We are not offering a mechanism for the shape, and we are not borrowing one either. The explanations that circulate for it — an argument from HLB, an argument from a molecular-weight window — are guesses about one investigator's ingredient set, published in a survey he called "not at all definitive but simply designed to stimulate research", and this site invented a molecular-weight argument once already and had it refuted by arithmetic. We report the rows and the shape they make. We do not know why. A rung is a nominal average rather than a full description of what was in the bottle. Steareth-10 is a distribution of chain lengths around ten, and Fulton records no supplier, no certificate of analysis and no measured distribution for any material behind any of these rows. So the ladders are ladders of labels. Whether two laboratories buying "oleth-10" would be testing the same mixture is not something the paper lets us check. What the family-level fact does NOT establish. It says nothing about PPG-5-Ceteth-10 Phosphate, which sits on no ladder. It does not tell a reader that a heavily ethoxylated ingredient is safe: Laureth-23's 3 is borderline by Fulton's own key, and unproven is not disproven in either direction. And it does not license reading one rung's score onto another, which is the specific thing a tidy one-line version of the shape invites. The Cosmetic Ingredient Review has assessed the alkyl PEG ethers and judged them safe as used when formulated to be non-irritating. What we read is that assessment as published in Int J Toxicol in 2012; we hold no record of the status line printed on the document itself, so we do not characterise it beyond what the journal carried. It is a safety assessment — irritation, sensitisation, systemic toxicity — and not a comedogenicity assessment; a panel concluding a material is not an irritant has said nothing about whether it plugs a follicle. That review also notes that these ethoxylates can carry 1,4-dioxane and unreacted ethylene oxide as manufacturing impurities, which is a purity question and again not this one. Last, the caveat that sits under every legacy score on this site, and under this page's argument as hard as under the numbers it corrects: a 2025 review in JAAD Reviews, surveying the literature from 1972 onward, found that no standardised comedogenicity test exists and that "non-comedogenic" is an unregulated label. And a family-wide generalisation is precisely the kind of claim this site was built to distrust. That includes ours. What we are asserting is bounded and checkable: five series of rows in one table, and the shape those rows make when each series is read as a series.
All 6 ethoxylates (pegs & polysorbates) 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
- Fulton JE. "Comedogenicity and irritancy of commonly used ingredients in skin care products." J Soc Cosmet Chem 40:321-333 (1989) The primary source of every number on this page and of nearly every comedogenic score in circulation. A survey, not a controlled trial: ingredients mixed in propylene glycol at nine to one, 1 ml applied daily to the inner ear of three New Zealand albino rabbits, five days a week for two weeks, follicular keratosis measured with a dial micrometer and confirmed on biopsy, graded 0-5. Read off a transcription of Table I made visually at 400dpi from the journal scan. Two things on this page come out of it and out of nothing else: the five ethoxylation series themselves, rung by rung, including the rungs we hold no rating for; and the grade key the republishing charts omit. Its author called it "not at all definitive but simply designed to stimulate research," and he wrote in the same paper that "not everything that irritates this model will also irritate human skin." One investigator, one laboratory, no replication.
- Fulton JE, Pay SR, Fulton JE III. "Comedogenicity of current therapeutic products, cosmetics, and ingredients in the rabbit ear." J Am Acad Dermatol 10(1):96-105 (1984) Fulton's earlier rabbit-ear paper, two rabbits per test, on the same 0-5 scale but with a DIFFERENT key: there 1 to 2 is not significant and 3 or above is positive, so a 3 is positive in 1984 and borderline in 1989. We hold its Table I in full from a print copy, which is why this page can say what it lacks rather than assume it: no oleth, steareth or PEG-stearate row of any length, so it supplies no second reading of any series on this page. What it does contain is Laureth-4 at 4, carrying the paper's concentration asterisk, against 1989's 5; and a row called "Ethoxylated lanolin" at 3, which matches none of the 1989 lanolin names. It is the same investigator and the same model, so it is not independent corroboration in any sense but the date.
- 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) The one genuinely independent rabbit-ear assay of the era, from a different laboratory, on the same 0-5 scale, and willing to contradict Fulton in print. Cited here mostly for its absences, which we verified against its transcribed rows rather than assumed: of thirty-two materials, none is an oleth, a steareth, a laureth or a PEG stearate, so no rung of any ladder on this page has a second reading. Two of its rows are ethoxylated nonionics — Octoxynol-9 at 50% aqueous, 0-1, and Polysorbate 20 at 1% aqueous, 0 — and the second of those is the only ethoxylate we can show two laboratories both scoring, at concentrations five to ten times apart. Neither material is a member of this family as we file it.
- Fiume MM, Heldreth B, Bergfeld WF, et al. (Cosmetic Ingredient Review Expert Panel). "Safety Assessment of Alkyl PEG Ethers as Used in Cosmetics." Int J Toxicol 31(5 Suppl):169S-244S (2012) An independent expert-panel SAFETY review of a large family of alkyl PEG ethers, the laureths, oleths and steareths among them, and not a comedogenicity assessment. Cited only for what it establishes: that these materials are the reaction products of a fatty alcohol and ethylene oxide, that they function as surfactants and emulsifiers, and that the Panel judged the class safe as used when formulated to be non-irritating. What we read is the assessment as published in Int J Toxicol in 2012; we hold no record of the status line printed on the document itself, so we do not characterise it beyond what the journal carried. Safety is a different endpoint from follicular plugging and the report counts no comedones. Its note that these ethoxylates can carry 1,4-dioxane and unreacted ethylene oxide as manufacturing impurities is a purity-and-safety matter and is not to be borrowed as if it were a comedogenicity finding.
- "Comedogenicity in cosmeceuticals: a review of clinical relevance, regulatory gaps, and future directions." JAAD Reviews (2025) A recent peer-reviewed review of the whole field, covering the literature from 1972 onward, and independent of Fulton and of the charts that republish him. Finds that no standardised comedogenicity test exists and that "non-comedogenic" is an unregulated label. Cited for the field-wide caveat that sits under every legacy score on this site, and under this page's own argument about the shape of Fulton's curve just as hard.