Porecelain

Flawless skin starts with the fine print.

Minerals & pigments

Six of the seven minerals on this site have a comedogenic grade printed against the name on the label, and every one of those grades falls at or below the line its own paper drew. That is an unusually clean record for this subject. What the charts do not carry is that all nine of those readings come out of one laboratory programme — two papers five years apart, Fulton first author on both — and that we identified no reading of any member of this family from a second laboratory. The variable Fulton's own table shows moving a dispersed colourant furthest, the carrier it was mixed into, is the one he never put through a series on a mineral.

7 rated 0 Clogging 0 Caution 7 Clear 1 never assayed

What they are

Insoluble inorganic solids. They are in a formula to colour it, to bulk it out, to absorb oil or water, to scatter ultraviolet light, or to thicken. Seven members: zinc oxide, titanium dioxide, iron oxides, talc, kaolin, bentonite and sodium chloride. ONE OF THE SEVEN DOES NOT FIT THAT DESCRIPTION, and it is worth saying which. Sodium chloride is freely water-soluble. It is a salt sitting among six powders that do not dissolve in anything a cosmetic is made of, and almost everything on this page is about the six rather than the seven. "Mineral" is OUR filing category, not Fulton's. His tables sort by what a material is FOR, and our seven are scattered across his. In the 1989 table, bentonite, kaolin and talc sit together among the powders and thickening agents; iron oxides and titanium dioxide sit in a different run, after a run of D&C red rows; zinc oxide sits in a third, beside lithium, magnesium and zinc stearates. Our 400dpi transcription records section headings for only two blocks of that paper, so we describe those runs by their neighbouring rows and do not give them names the transcription does not hold. The 1984 paper does print its headings, and there our members appear under two of them: "Miscellaneous pigments" and "Miscellaneous ingredients". Some materials a reader will expect here are not in this family and will not appear in the table below. Mineral oil, squalane, petrolatum and hydrogenated polyisobutene are HYDROCARBONS, filed separately because they are not minerals in this sense at all — mineral oil appears on this page only as something Fulton mixed other things into. Magnesium aluminum silicate, hectorite, ultramarine violet, carmine, chromium hydroxide, and the lithium, magnesium and zinc stearates are rows in Fulton's tables that we do not rate, and they are named below as his rows, never as ours.

What the evidence actually says

Start with the record itself, because it is short enough to print in full. Fulton's 1989 survey, on his 0-5 scale, grades six members of this family: Bentonite 0, Kaolin 0, Talc 1, Iron oxides 0, Titanium dioxide 0, Zinc oxide 1. None of those six rows carries an asterisk and none is a range. His earlier 1984 paper with Pay and Fulton, on the same 0-5 scale, grades three of the same six: "Kaolin*" 1, "Iron oxides" 0-1, "Titanium dioxide*" 0-1. That is nine readings, six materials, two documents. NOW THE TWO KEYS, because a hub is the place to print them once. Fulton 1989, p. 322: "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." Fulton 1984 draws the line somewhere else: "a minimal grade of 1 to 2 is not considered significant. However, a grading score of 3 or above is uniformly reproducible and considered positive." So a 3 of 5 is borderline in one of his papers and positive in the other, and a digit carried across without its key is not the same claim it was. Every one of the nine readings above falls below the line drawn in the paper that printed it. That is nine readings measured against two different thresholds, not one verdict repeated nine times. THE FAMILY-LEVEL FACT IS WHERE THOSE NINE READINGS CAME FROM. Three members — kaolin, iron oxides and titanium dioxide — have two readings each, which on any other subject would read as corroboration. It is not. The two readings in each case are the same first author, the same species, the same laboratory programme, five years apart, and where they differ, as kaolin's 0 and 1 do, the vehicle differed too, so the digit cannot be assigned to the clay rather than to the preparation. The document that would have counted as an outside check exists, and it did not test this family. Morris and Kwan's 1983 rabbit-ear paper is an independent assay on the same 0-5 scale, from a different laboratory, six years before Fulton's famous table, and it contradicts him in print on other materials. Its 32 rows cover seven lanolins, nine vegetable oils, eleven esters and five surfactants. There is no mineral in it. Nguyen, Dang and Maibach's 2007 rabbit study names its fourteen test materials in its abstract: esters, fatty alcohols, hydrocarbons, one butter and one surfactant. There is no mineral in that either. Kanaar's 1971 paper graded six fatty acids in words rather than numbers. We identified no comedogenicity reading of any member of this family from a laboratory other than Fulton's, in the documents we hold and in the searches run for the member pages. The check was available. On this family it was not run. AND THE VARIABLE HIS OWN TABLE MOVES FURTHEST WAS NEVER PUT THROUGH A SERIES ON A MEMBER OF THIS FAMILY, in either of the two papers that grade one. Fulton did run exactly one vehicle experiment, and it is the only table in the 1989 paper given over to a single material. Table II, on p. 331, grades one colourant in four carriers: "D&C red #36 in mineral oil" 3, "D&C red #35 in pentaerythrital tetra capra/caprylate" 2, "D&C #36 in propylene glycol" 1, "D&C red #36 in PEG 400" 0, all on the 0-5 scale. We print those row labels as they appear — the second says #35 and the third drops the word red, and we are not going to silently tidy a paper we are quoting. One material, four carriers, and a grade that runs from borderline by his own key down to nothing. What that table is NOT is a result about any mineral. D&C Red #36 is a synthetic organic colorant, it is not in this family, and Table II says nothing about iron oxides or titanium dioxide in either direction. What it does establish, in Fulton's own paper, is that in this model a dispersed colourant's grade is a property of the material and its carrier together. No mineral in either paper was put through a series like that one. Each of the nine readings is a single grade in a single preparation, and the preparations are not all the same. In 1989 the vehicle was roughly 10% in propylene glycol. In 1984 the asterisk on "Kaolin*" and "Titanium dioxide*" means the sample was "either diluted with mineral oil or incorporated into an alcohol gel at a 10% concentration" — which of those two, the paper does not say. The 1984 iron oxides row is unstarred, and that paper's footnote says an unmarked material "was used 'as is' if the agent was a liquid or soft solid" — a dry pigment is neither, so how that particular sample was carried is not recoverable from the page. FULTON'S OWN FILING IS THE LAST PIECE, AND IT IS A LAYOUT, NOT AN EXPERIMENT. His 1984 Table I prints its colourants in two headed blocks, one after the other. "D & C pigments" holds ten rows running 2 to 4, eight of them at 3 or above, which that paper's key calls positive. "Miscellaneous pigments" holds five — printed "Ultramarine Violet", "Iron oxides", "Carmine", "Chrominum hydroxide" and "Titanium dioxide", the misspelling in the fourth his own — and every one of the five reads 0-1, which the same key on the same page calls not significant. All ten rows in the first block carry the dilution asterisk and four of the five in the second do, so the gap between the blocks is not explained by one set being diluted and the other not. The 1989 table is laid out the same way: a run of D&C reds from 1 to 3, then four colourants at 0. Two disciplines on that observation, and both matter. It is a description of how two tables are arranged, which is what our iron oxides and titanium dioxide pages already say about it, and it is not a further experiment on either material — it does not raise our confidence in the zeros. And the block is not chemically clean, so it cannot be read as a finding about inorganic pigments: carmine is in it, and the FDA defines carmine as an aluminium or calcium-aluminium lake of the colouring principles, chiefly carminic acid, obtained by extracting cochineal insects. Fifteen rows in one table, single readings, one laboratory programme, no replication. We are not going to promote that into a rule. THE SEVENTH MEMBER IS THE EXCEPTION IN BOTH DIRECTIONS. Sodium chloride has no row in the 1989 table, no row in the 1984 table, none in Morris and Kwan, and none in any numeric document we hold. It is also the one member of this family that carries an alarming figure in general circulation. The consumer charts print a 5 of 5 against it. That figure is not a reading and we are not repeating it as one — no document we hold ever put salt on that scale — and it was already being handed around on the acne.org forum on 14 September 2005 by people who could not source it either. So the family divides into six materials whose circulating numbers are real rows read under the right names, and one whose famous number we have not traced to a measurement at all.

What the evidence doesn’t tell you

The absence above is bounded and has to be. What we can say is that we identified no comedogenicity reading of zinc oxide, titanium dioxide, iron oxides, talc, kaolin, bentonite or sodium chloride from a laboratory other than Fulton's, in the numeric documents we hold and in the searches run for their pages. That is a statement about a corpus we can show you, not about the world. Our own provenance file instructs us to treat that corpus as OPEN, and it says so because it grew: a ninth numeric paper turned up after we had written down that there were eight. THIS FAMILY IS NOT WITHOUT HUMAN EVIDENCE, and the counter-examples are inside our own site. Three of them, and they do not all point the same way. Draelos and DiNardo, in 2006, counted comedones on the upper backs of six people with prominent follicular orifices, using a modification of the Mills and Kligman human assay, on finished cosmetic products rather than ingredients. One of the products was a face powder reported as more than half zinc oxide, and it came back non-comedogenic — which is a result about a formula, and sits inside a paper whose whole argument is that an ingredient's grade does not predict what the product containing it does. Salimi and colleagues, in 2025, randomised 60 people with moderate acne in Sari, Iran, to a bentonite and Alcea sulphurea mask or a placebo gel, both arms also taking oral azithromycin; total lesion count fell 55.45% against 21.74% unadjusted, but after the authors adjusted for baseline differences in age and education the between-group p-values were 0.063 and 0.203, and no arm isolates the clay. And Choi and colleagues, also in 2025, questioned 151 people in Guangzhou and found powder products an independent risk factor for acne after adjustment, odds ratio 3.47 (95% CI 1.58-7.59) — a result that points the other way, names no ingredient, and being observational shows association rather than cause. A SECOND LABORATORY DID TEST THE PRODUCT CATEGORY two of these minerals are the actives in, and its conclusion was about carriers. Mills and Kligman applied 29 proprietary sunscreen formulations to the external ear canal of albino rabbits in 1982 — the paper at Arch Dermatol 118(6):417-419, which is not the other Mills and Kligman paper of that year, and the source note below says why that matters — and found fourteen of them comedogenic. Their abstract states that "the vehicles, rather than the UV-absorbing compounds, seemed to be responsible" and that "a sampling of UV absorbers showed these agents to be noncomedogenic". We have read that abstract and not the full text. It names no mineral, prints no per-material grade in what we read, and UV absorbers are not the mineral filters, so it is not a test of zinc oxide or titanium dioxide and must not be read as one. It is here because it is the nearest thing to an outside check this family has, and because a different laboratory reached the same conclusion about carriers that Table II reaches. Unproven is not disproven, and that cuts in the reassuring direction here rather than the alarming one. A grade below a paper's own significance line is that assay failing to see an effect. It is not a measured zero, and it is not a measurement of a human follicle. Fulton called the 1989 survey "not at all definitive but simply designed to stimulate research", and warned that "not everything that irritates this model will also irritate human skin" — a caution about what his high grades mean. What his low grades imply for human skin is a question the documents we hold do not answer, and we are not going to answer it for them. The materials themselves are not fixed, either, and nothing above pins them down. "Iron Oxides" is a plural covering red, yellow and black oxides, hydrated and anhydrous, blended or single; the tables print the plural and stop. Bentonite's principal mineral varies with the deposit it was dug from. Neither paper records a source, grade, purity, particle size, surface treatment or supplier for anything in this family. Cosmetic iron oxides and titanium dioxide are commonly supplied surface-treated, with the treatment co-listed in the INCI name, and a treated grade is a different physical material from a bare one. WHAT THE FAMILY-LEVEL FACT DOES NOT COVER. The two-block observation reaches only iron oxides and titanium dioxide of our seven; the other five are not in those blocks. The carrier point applies to the six with rows, and it rests on an experiment run on a colorant that is not one of them. Zinc oxide, talc and bentonite have one reading each and no second, so for them there is not even a within-programme concordance to discount. And salt has no reading at all, so nothing on this page is about it except its absence. Finally, the generalisation itself. Seven materials filed together because they are inorganic solids is our decision about a database, and a family-wide claim is exactly the kind of claim this site exists to distrust, including when we are the ones making it. The narrow sentence, and the only one we will defend, is this: for six of these seven the circulating number is a real row read under the right name, and what is missing is not the measurement but any second one.

All 7 minerals & pigments 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 circulating comedogenic score descends from — 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, follicular keratosis measured with a micrometer. Held here transcribed verbatim from a 400dpi render of the journal scan, because OCR drops the numeric columns. Six members of this family have rows in it: bentonite 0, kaolin 0, talc 1, iron oxides 0, titanium dioxide 0, zinc oxide 1, all on the 0-5 scale, none asterisked and none a range. Sodium chloride has no row. Also the source of the grade key on p. 322 and of Table II on p. 331, which grades one colourant at 3, 2, 1 and 0 across four carriers. Fulton calls the whole survey "not at all definitive but simply designed to stimulate research". A single-author animal survey, not a clinical trial and not a human result.
  2. Fulton JE Jr, 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). PMID 6229554 The earlier and far less quoted rabbit-ear paper from the same laboratory, and note the byline — Fulton JE Jr, Pay SR and Fulton JE III, three authors and two different Fultons, which is why this page speaks of one laboratory programme rather than of one man. Read off a 400dpi render of the journal scan — two New Zealand albino rabbits per test, one ear dosed daily for two weeks with the other as control, borderline cases confirmed by microscopic biopsy. It is the source of everything on this page about the two pigment blocks, and of the second reading for kaolin, iron oxides and titanium dioxide. Two features belong to this paper alone and must travel with its numbers. Its key is its own — 1 to 2 not significant, 3 or above positive — which is NOT the 1989 key. And its asterisk is a dilution note, not a source note: marked materials "were either diluted with mineral oil or incorporated into an alcohol gel at a 10% concentration", unmarked ones being used "as is" if the agent was a liquid or soft solid. Concordant with 1989 on all three shared members, and from the same programme, so not independent replication.
  3. 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 independent rabbit-ear assay on the same 0-5 scale, from a different laboratory, six years before Fulton's famous table — the document that would have corroborated or contradicted this family's numbers had it tested any of them. It did not. Its 32 rows are seven lanolins, nine vegetable oils, eleven esters and five surfactants, and contain no mineral. Cited for that absence, and because it demonstrates that an outside check was possible: on other materials Morris and Kwan contradict Fulton in print and say so.
  4. Mills OH Jr, Kligman AM. "Comedogenicity of sunscreens. Experimental observations in rabbits." Arch Dermatol 118(6):417-419 (1982). PMID 6212027 A different laboratory testing the product category two members of this family are the actives in. We have read the PubMed abstract verbatim and NOT the full text, so methods, per-material grades and any scale are not in hand. As printed: "Fourteen of 29 proprietary sunscreen formulations, including suntan promoters, were found to be comedogenic when applied to the external ear canal of albino rabbits. Ultraviolet exposures enhanced the comedogenic effect. The vehicles, rather than the UV-absorbing compounds, seemed to be responsible." It also reports that "a sampling of UV absorbers showed these agents to be noncomedogenic". Cited for one thing only: a second laboratory reaching the same conclusion about carriers that Fulton's Table II reaches. It tests FINISHED FORMULATIONS, names no mineral in the text we read, and UV absorbers are a different class from the mineral filters, so it is not evidence about zinc oxide or titanium dioxide in either direction. TWO DIFFERENT PAPERS ON THIS SITE ANSWER TO "MILLS AND KLIGMAN 1982", and a bare year would let them be read as one. This is the sunscreen paper, Arch Dermatol 118(6):417-419, rabbits only. The other is Mills OH and Kligman AM, Arch Dermatol 118:903-905 (1982), a model using both the rabbit ear and the human back with cyanoacrylate follicular biopsy, graded 0-3, which our lanolin family hub cites as an unread table and which appears nowhere on this page. Same authors, same journal, same year, different study, different scale. The page range is what tells them apart.
  5. Draelos ZD, DiNardo JC. "A re-evaluation of the comedogenicity concept." J Am Acad Dermatol 54:507-512 (2006). PMID 16488305 A "Current Issues and Opinion" piece, not a controlled trial: a modification of the Mills and Kligman human assay run on six people with prominent follicular orifices and the ability to form comedones on the upper back, testing FINISHED PRODUCTS built from ingredients the rabbit assay had flagged. It reports no comedogenic score for any individual ingredient, and citing it for one would invert the paper. Cited here for two things it does support: that one of the products it tested was a face powder reported as more than half zinc oxide and did not come back comedogenic, and that an ingredient's grade does not transfer cleanly to the formula it ends up in, in either direction.
  6. Nguyen SH, Dang TP, Maibach HI. "Comedogenicity in rabbit: some cosmetic ingredients/vehicles." Cutan Ocul Toxicol 26(4):287-292 (2007). PMID 18058303 The most recent rabbit-ear study in the corpus we hold, graded on a 0-4 scale — a third scale, easily misread as Fulton's. Abstract read verbatim, full text not read. The abstract names all fourteen test materials: esters, fatty alcohols, hydrocarbons, one butter and one surfactant. No member of this family is among them. Cited for that absence only.
  7. Choi K, Liu H, Zhu Y, Jiang Z, Lu S. "A Case-Control Study Exploring the Association Between Cosmetic Use and Acne Risk: Implications for Prevention and Clinical Practice." Clin Cosmet Investig Dermatol 18:1833-1843 (2025). doi:10.2147/CCID.S533950 A single-centre case-control study of 151 participants — 81 with acne, 70 without — at Sun Yat-sen Memorial Hospital, Guangzhou, using a self-administered questionnaire. Read from its abstract record. Its endpoint is self-reported product-category use against acne case status, not comedone counts and not ingredients: after adjustment for age, occupation, parental history and milk consumption, powder products remained an independent risk factor, odds ratio 3.47 (95% CI 1.58-7.59). Cited because powders are the category most of this family lives in, and because a hub that carried only the reassuring evidence would be choosing its sources. It names no ingredient and establishes association rather than cause.
  8. Salimi S, Fakhri M, Saeedi M, Rahimnia SM, Kazeminejad A, Moosazadeh M, Habibi E, Jokar A. "The Therapeutic Effects of a Bentonite-Based Facial Mask With Alcea sulphurea Extract on Acne Severity and Patient Experience: Add-On Randomized Controlled Clinical Trial." J Cosmet Dermatol 24(12):e70586 (2025). doi:10.1111/jocd.70586 The only randomised human trial of a topical containing a member of this family that we found, read in full text. A TREATMENT study, not a comedogenicity test: 60 people with moderate acne in Sari, Iran, randomised to a bentonite and Alcea sulphurea mask or a placebo gel, both arms also taking oral azithromycin 250 mg, nightly for eight weeks. Funded by Mazandaran University of Medical Sciences; no declared conflicts. What it cannot support, from its own text: the mask combines the clay with a plant extract at 1:1, so no arm isolates bentonite; the arms differed at baseline in age and education, and after adjustment the between-group p-values were 0.063 for lesion count and 0.203 for the severity index, neither reported in the abstract; and the authors list the mask-versus-gel format difference as a threat to blinding.
  9. Kern D. "What Is Comedogenicity, and What Ingredients Are Comedogenic? The Full Story." Acne.org, last updated 24 March 2022 A consumer article by a commercial acne-products site, not a study, and the place most readers will have met a family-level verdict on these materials. Characterised precisely rather than dismissed: it walks through eleven comedogenicity studies with citations, states its own list rule, and says plainly that rabbit results do not necessarily transfer to humans. Inside its summary of the 1989 paper it prints "Mineral materials generally were non-comedogenic. They include clays, bentonite, kaolin, talc, iron oxide, chromium hydroxide, and titanium dioxide." That is a paraphrase rather than a row read off the table, and it names materials we do not rate. Whatever route it travelled, it is not a second measurement: the article attributes it to the same 1989 paper this page reads.
  10. US Food and Drug Administration. 21 CFR 73.100, "Cochineal extract; carmine" (Listing of Color Additives Exempt from Certification), with the cosmetic listing at 21 CFR 73.2087 A federal colour-additive regulation, cited for material identity and nothing else — it contains no comedogenicity data and makes no claim about follicles. It defines carmine as the aluminum or calcium-aluminum lake, on an aluminum hydroxide substrate, of the colouring principles, chiefly carminic acid, obtained by aqueous extraction of cochineal. Cited for one narrow purpose: to stop the reader, and us, from reading Fulton's "Miscellaneous pigments" block as a chemically uniform set of minerals when one of its five rows is an insect-derived lake.

‹ All ingredients