The literature

This page exists so that nobody has to take the article on trust.  Every source it rests on is listed here with what it actually measured, who wrote it, and which way it cuts.

It is deliberately not a reading list of things that agree with this site.  The source that does the most damage to the popular creationist model is Martin 2017, and the damage is done by a heritability estimate that no creationist source has yet cited.  The source that helps the creationist side most is Wilde 2014, a mainstream ancient-DNA paper that measures a population's colour changing fast on standing variation.  The most honest document on the creationist side is Sibley 2025, which concedes dark-skinned ancestral Europeans and UV-driven selection and rests everything on radiocarbon.  All three are tagged by what their data show, not by what their authors believe.

How to read an entry

Measurement or model.  A melanin-index reading, an allele frequency, or a count of nonsynonymous sites is an observation.  A coalescent TMRCA, a sweep-onset date and a predicted ancient phenotype are models, and each inherits every assumption in the machinery that produced it.  Both are useful and they are not the same kind of thing.

Direction.  Assigned by what the source shows, not by what its author believes.

Verified.  Whether this project read the text directly (full), only its abstract or a supplement (abstract), or only through another source that quotes it (via).  Nothing is quoted in the main article unless it is marked full or abstract, and every quotation was read out of downloaded bytes.  No quotation, page number or numeral on this site was established with Claude's WebFetch tool.

Field definitions

Coverage and its limits, stated plainly.  This catalogue covers the pigmentation-genetics literature from Harding 2000 to Perretti 2025, the four creationist primaries that carry the "one race, mid-brown Adam" argument, and the two commentaries that explain why the field itself once believed the few-loci picture.  Five items could not be obtained and are listed at the bottom.  The most damaging of those, for both sides, is Lamason 2005, whose body text is the origin of a figure the whole popular literature repeats.

Architecture: how many genes

Loci associated with skin pigmentation identified in African populations

Crawford, Nicholas G., et al.  Science 358(6365):eaan8433, 2017.  https://doi.org/10.1126/science.aan8433

camp: mainstream topic: architecture method: measurement direction: complicates verified: full

Key figures: melanin index measured with a DSM II ColorMeter on 2,092 Africans in Ethiopia, Tanzania and Botswana, 1,570 genotyped on an Illumina Omni5 array;  four genome-wide significant regions, SLC24A5, MFSD12, DDB1/TMEM138 and OCA2/HERC2;  28.9% ± 10.6% of the pigmentation variance at those four, of which SLC24A5 alone is 12.8% ± 3.5%;  largest single-SNP effect 7.69 melanin-index units per allele copy;  coalescent TMRCAs of 612 kya and 996 kya for derived dark MFSD12 alleles, 1.7 Mya for all OCA2 lineages, 921 kya for the derived light allele at HERC2, 29 kya for Eurasian lineages carrying the light SLC24A5 allele;  Mfsd12 knockdown raised melanin 30–50% in cultured melanocytes.

The single most important source in the article, and it cuts both ways.  It supports the creationist claim creationists most care about:  light-skin alleles are old and African, not European inventions, at every locus except SLC24A5.  It undercuts the model on architecture, because four loci leave 71% of the heritability elsewhere, and on timing, because the antiquity it establishes is coalescent depth of hundreds of thousands to millions of years.  Its "simpler than height" sentence is widely quoted by creationists without the sentence in front of it.  The NIH author manuscript is unpaginated, so every locator is a section heading, and its OCR mangles two RSIDs and one en dash;  the article says so in the footnotes.  Cited heavily.

An Unexpectedly Complex Architecture for Skin Pigmentation in Africans

Martin, Alicia R., et al.  Cell 171(6):1340–1353.e14, 2017.  https://doi.org/10.1016/j.cell.2017.11.015

camp: mainstream topic: architecture method: measurement direction: undercuts verified: full

Key figures: 465 KhoeSan, 278 ǂKhomani San and 187 Nama, with spectrophotometric melanin index, array genotypes, exomes and targeted resequencing, plus a compiled global set of 4,712 individuals from 32 populations;  heritability of baseline colour 0.96 ± 0.12 from pedigrees, 0.97 ± 0.15 from a genomic relatedness matrix, 0.95 ± 0.26 from exomes, 0.97 ± 0.15 from identity-by-descent;  candidate gene sets explain 0.08 and 0.09 against 0.79–0.82 for the rest of the genome;  tanning heritability not significantly greater than zero;  Irish versus Ghanaian variance ratio F = 0.03, p = 6.7e-23.

The hardest single source against the popular creationist model, and no creationist source cites it.  In the population with the world's greatest genetic diversity, skin colour is 97% heritable and the known pigmentation genes explain effectively none of it.  The nine-SNP European predictor has no significant association with measured colour there.  Its conclusion is more than fifty loci of mostly small effect.  It also supplies the fairest available explanation of why the few-loci picture ever looked right, which the article quotes in full.  The HTML strip destroys superscripts throughout, so no exponent is printed in the article that is not recoverable from the surrounding prose.  Cited heavily.

Focus on African diversity confirms complexity of skin pigmentation genetics

Quillen, Ellen E.  Genome Biology 19:13, 2018.  https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5793402/

camp: mainstream topic: critique method: commentary direction: undercuts verified: full

Key claim: the focus on derived alleles in strongly selected low-UV populations "may have contributed to the mistaken idea" that pigmentation is set by a few large-effect SNPs.

An independent commentary on both 2017 papers by neither team, which is why it is here.  It is the cleanest published statement of why the popular creationist model looked like good genetics for twenty years.  CC-BY, so it can be quoted at length.  Cited.

A GWAS in Latin Americans highlights the convergent evolution of lighter skin pigmentation in Eurasia

Adhikari, Kaustubh, et al.  Nature Communications 10:358, 2019.  https://doi.org/10.1038/s41467-018-08147-0

camp: mainstream topic: architecture method: measurement direction: undercuts verified: full

Key figures: 6,357 CANDELA individuals from five Latin American countries at 8,896,142 genotyped and imputed SNPs;  melanin index 20 to 65, mean 34.98, SD 5.34;  eighteen independent signals at twelve regions;  strongest MFSD12 signal is the Y182H missense variant, common only in East Asians and Native Americans;  significant epistasis between the large-effect SNPs.

The convergence result is the one that has no creationist answer:  light skin in East Asia runs through different alleles at overlapping genes from those in Europe and Africa.  Under a shared created heterozygosity the same alleles should be light everywhere.  Its epistasis result also weakens this article's own arithmetic, which assumes additivity, and the article says so.  CC-BY.  Supplementary Table 5, which carries the effect sizes of the six conditioning SNPs, was not fetched.  Cited.

Human pigmentation genes under environmental selection

Sturm, Richard A. & David L. Duffy.  Genome Biology 13(9):248, 2012.  https://doi.org/10.1186/gb-2012-13-9-248

camp: mainstream topic: uv-model method: review direction: complicates verified: full

Key figures: open-access review of the pigmentation gene set;  in cultured human melanocytes the light SLC24A5 genotypes carry roughly twofold higher tyrosinase activity;  the SLC24A5 light allele "shows no evidence of selection" in the indigenous American populations Quillen studied.

Used for one thing the article needed and could not get from Lamason 2005:  a CC-BY statement of the functional mechanism, plus the awkward fact that the flagship light allele carries no selection signal on one continent.  Cited once.

Rate: how fast a population changes

Direct evidence for positive selection of skin, hair, and eye pigmentation in Europeans during the last 5,000 y

Wilde, Sandra, et al.  PNAS 111(13):4832–4837, 2014.  https://doi.org/10.1073/pnas.1316513111

camp: mainstream topic: selection-rate method: measurement direction: supports verified: full

Key figures: ancient DNA from 63 of 150 Eneolithic and Bronze Age skeletons on the Pontic-Caspian steppe, 48 with pigmentation genotypes;  SLC45A2 rs16891982 G from 0.432 to 0.927, TYR rs1042602 A from 0.043 to 0.367, HERC2 rs12913832 G from 0.160 to 0.651;  fitted selection coefficients 0.030, 0.026, 0.036;  neutrality rejected for all three.

The strongest single paper for the creationist case in the whole catalogue, and it should be given its full weight.  A continuous population's pigmentation moved a long way in about 5,000 years using alleles it already carried at 4% to 43%.  Nothing new had to be invented.  It says nothing about how many loci exist or where the alleles came from, and the paper is candid that its estimates assume population continuity and that the selective pressure itself "remains unknown."  Cited heavily.

Direct detection of natural selection in Bronze Age Britain

Mathieson, Iain & Jonathan Terhorst.  Genome Research 32:2057–2067, 2022.  https://doi.org/10.1101/gr.276862.122

camp: mainstream topic: selection-rate method: measurement direction: supports verified: full

Key figures: seven loci with genome-wide significant selection in the past 4,500 years;  of 28 complex traits, skin pigmentation the only one with significant polygenic selection;  SLC45A2 rs16891982 s = 0.043, HERC2 rs12913832 s = 0.017, LCT s = 0.064, DHCR7 s = 0.051;  independent confirmation of Wilde's 40% to 90% change at s ≈ 0.03.

The best available measurement of the actual rate, and the number the article builds on.  It is simultaneously a refutation of the creationist claim that no selection is needed:  the frequencies were still moving in the historical period, in Britain, under measured selection.  Its row was missing from sources/index.tsv despite the file being on disk;  the row was added in the session that wrote this article.  Cited.

The evolution of skin pigmentation-associated variation in West Eurasia

Ju, Dan & Iain Mathieson.  PNAS 118(1):e2009227118, 2021.  https://doi.org/10.1073/pnas.2009227118

camp: mainstream topic: selection-rate method: measurement direction: supports verified: full

Key figures: 1,158 ancient West Eurasian genomes from 45,000 to 715 years before present at about 1.2 million sites;  170 UK Biobank skin-colour SNPs;  polygenic selection test on all 170 fails at P = 0.27, is overwhelming on the top five at P < 2.5 × 10⁻⁷, and vanishes again with those five removed at P = 0.59;  the 170 SNPs' PBS values do not deviate from the genome-wide distribution.

Helps the creationist argument more than any mainstream paper except Wilde, and the creationist literature has apparently never cited it.  European lightening was a handful of alleles plus migration, not a slow grind across a hundred loci.  Its phenotype is self-reported UK Biobank skin colour on a six-level scale, much cruder than the reflectometry elsewhere in this catalogue, and it cannot predict ancient individuals' phenotypes.  Cited.

The Timing of Pigmentation Lightening in Europeans

Beleza, Sandra, et al.  Molecular Biology and Evolution 30(1):24–35, 2013.  https://doi.org/10.1093/molbev/mss207

camp: mainstream topic: selection-rate method: model direction: complicates verified: full

Key figures: forward Monte Carlo simulation with rejection sampling on compound SNP-plus-microsatellite haplotypes, effective population size fixed at 10,000;  sweep onsets of 11,000–19,000 years at TYRP1, SLC24A5 and SLC45A2 and about 30,000 at KITLG;  SLC24A5 additive-model onset 11,368 years with a 95% interval of 1,036 to 55,832;  measured effects in admixed populations of 7.6–11.4, 5, and 2.8–3.8 melanin units at SLC24A5, SLC45A2 and KITLG, with the authors' own warning that these "could be overestimated."

Complicates both sides, and the intervals are the part both sides drop.  The modes are two to three times any biblical date for Babel.  The lower bounds of three of the four sweeps sit inside a post-Babel window.  It also supplies the effect-size substitute for Lamason 2005, which could not be obtained.  Two-column PDF;  every quotation was reassembled from the layout-preserving extraction with the columns kept apart.  Cited.

Genome-wide patterns of selection in 230 ancient Eurasians

Mathieson, Iain, et al.  Nature 528(7583):499–503, 2015.  https://doi.org/10.1038/nature16152

camp: mainstream topic: ancient-dna method: measurement direction: complicates verified: full

Key claim: the derived SLC45A2 allele was at much lower frequency in ancient Europeans than today;  the derived SLC24A5 allele appears fixed in the Anatolian Neolithic, so its rise in Europe "was mostly due to migration."

Complicates both sides.  Against a naive "Europeans evolved white skin in Europe," one of the two big loci arrived already fixed with farmers.  Against the creationist account, the other one swept in place, in Europe, long after any biblical Babel date.  Cited once.

Allele age and the archaic comparison

Evidence for Variable Selective Pressures at MC1R

Harding, Rosalind M., et al.  American Journal of Human Genetics 66(4):1351–1361, 2000.  https://doi.org/10.1086/302863

camp: mainstream topic: allele-age method: measurement direction: complicates verified: full

Key figures: complete MC1R sequencing in European, African, African American and Asian samples plus two chimpanzees;  zero nonsynonymous variants in 106 African chromosomes against thirteen segregating sites in 524 non-African ones;  29% of total variance in MC1R haplotype distributions apportioned among populations, against 10–15% for autosomal loci generally;  allele ages of 250,000–100,000 years for three variants and 80,000 for the red-hair variants, on an explicit assumption of "genetic drift in constant-size, randomly mating populations."

Cuts both ways and the creationist side has not engaged it at all.  Africa has no functional MC1R variation to recombine, which is awkward for a model in which one heterozygous couple seeded the world's colour range at the gene creationists most often name.  But Harding's own conclusion is that European MC1R diversity reflects relaxed constraint rather than positive selection, which a young-earth model can use.  Its stated constant-size assumption is also the opening for the best unmade creationist argument in the field, and the article says so.  Cited.

MC1R diversity in Northern Island Melanesia has not been constrained by strong purifying selection

Norton, Heather L., Elizabeth Werren & Jonathan Friedlaender.  BMC Genetics 16:122, 2015.  doi:10.1186/s12863-015-0277-x.  https://doi.org/10.1186/s12863-015-0277-x

camp: mainstream topic: uv-model method: measurement direction: supports verified: full

Key figures: MC1R coding region sequenced in 188 Island Melanesians from New Hanover, New Britain, New Ireland and Bougainville, with rs2228479 genotyped in 635;  three nonsynonymous and two synonymous mutations;  a modified McDonald-Kreitman test finds no significant purifying selection;  no reduced FST or haplotype sharing with Africans;  the one common nonsynonymous polymorphism is not associated with skin or hair colour.

Helps the creationist side against a paper the creationist side has never cited.  Harding's Melanesian sample was 32 chromosomes;  this one is 188 people, and the tidy picture of MC1R locked down by UV everywhere does not survive it.  The authors credit population history rather than selection.  It removes one mainstream argument;  it does not touch allele age.  CC-BY.  Cited.

Ancient DNA

Derived immune and ancestral pigmentation alleles in a 7,000-year-old Mesolithic European

Olalde, Iñigo, et al.  Nature 507(7491):225–228, 2014.  https://doi.org/10.1038/nature12960

camp: mainstream topic: ancient-dna method: measurement direction: undercuts verified: full

Key figures: 3.40x whole genome from a tooth of La Braña 1, León, Spain, radiocarbon-dated 7,940–7,690 cal BP;  homozygous ancestral at SLC45A2 rs16891982 and SLC24A5 rs1426654, whose ancestral alleles are now at 3% and 0% in Europeans;  ancestral at MC1R, TYR and KITLG, derived at TYRP1, ASIP and IRF4;  genomic heterozygosity 0.042%.

The single hardest datum for a young-earth pigmentation account, and no creationist source found addresses it.  These are direct genotype calls, not classifier output, which is why they carry weight that the Cheddar Man prediction does not.  The paper's own hedge, that the phenotype "cannot currently be ascertained in a European genetic background," is a real hook and the article reports it.  Cited.

Ancient genomes indicate population replacement in Early Neolithic Britain

Brace, Selina, et al.  Nature Ecology & Evolution 3(5):765–771, 2019.  https://doi.org/10.1038/s41559-019-0871-9

camp: mainstream topic: ancient-dna method: model direction: complicates verified: full

Key claim: Cheddar Man predicted to have had blue or green eyes and "dark or dark to black skin";  different pigmentation levels coexisting in Europe by about 6000 BCE.

This is the famous case and the weaker one, because it is a 36-SNP classifier applied to a 2.3x genome.  The coexistence finding is the standing-variation picture the creationist model wants, stated by the paper itself.  Cited once, with the accuracy caveats attached from Walsh and Perretti.

Global skin colour prediction from DNA

Walsh, Susan, et al.  Human Genetics 136(7):847–863, 2017.  https://doi.org/10.1007/s00439-017-1808-5

camp: mainstream topic: ancient-dna method: measurement direction: supports verified: full

Key figures: 77 SNPs from 37 loci in 2,025 individuals from 31 populations, reduced to a 36-SNP predictor;  five-category cross-validated AUC 0.74 Very Pale, 0.72 Pale, 0.73 Intermediate, 0.87 Dark, 0.97 Dark-Black;  in an independent test on 194 samples from 17 populations, intermediate AUC 0.63.

The caveat the Cheddar Man press coverage dropped, and a fair point for the creationist side.  The model separates the two ends well and the middle barely above chance.  CC-BY.  Cited.

Inference of human pigmentation from ancient DNA by genotype likelihood

Perretti, Silvia, et al.  bioRxiv preprint, 30 January 2025.  https://doi.org/10.1101/2025.01.29.635495

camp: mainstream topic: ancient-dna method: measurement direction: complicates verified: full

Key figures: HIrisPlex-S applied to 348 ancient Eurasian genomes over 45,000 years using genotype likelihoods;  half of individuals still dark or intermediate into the Copper and Iron Ages;  below 8x coverage direct calling gives 10% error and imputation starts failing at 3x, reaching 50% wrong.

Preprint, not peer reviewed, and the article says so wherever it is cited.  It lets the creationist side keep its best point, that the colours were sorting slowly and coexisting well into the historical period, while keeping the mainstream timescale, because the dates are radiocarbon and stratigraphic.  Its coverage floor is what makes the Cheddar Man caveat quantitative.  CC-BY-NC-ND.  Cited.

The UV model

The evolution of human skin coloration

Jablonski, Nina G. & George Chaplin.  Journal of Human Evolution 39(1):57–106, 2000.  https://doi.org/10.1006/jhev.2000.0403

camp: mainstream topic: uv-model method: measurement direction: complicates verified: full

Key figures: skin reflectance at 425, 545 and 685 nm regressed against satellite UVMED;  Table 4 correlations at 545 nm pooled, 0.957 against UV and 0.964 against latitude;  Table 5 within-hemisphere r² of 0.4572 and 0.3402 at 425 nm, 0.7747 and 0.7013 at 545 nm, 0.3995 and 0.2485 at 685 nm;  females lighter than males in every population;  Robins (1991:208) quoted saying there is "not a scintilla" of evidence for the rickets mechanism.

The founding paper of the UV model, and its own numbers are more modest than its reputation.  The cline is real, strong and replicated;  the r² is between 25% and 77%.  Anyone quoting an unattributed r of 0.96 should say which wavelength.  Its quotation of a named dissenter is the best citation in the catalogue for the article's objections section.  Page anchoring:  printed page = PDF page + 56, anchored twice against the running heads.  Cited.

Human skin pigmentation as an adaptation to UV radiation

Jablonski, Nina G. & George Chaplin.  PNAS 107(suppl. 2):8962–8968, 2010.  https://doi.org/10.1073/pnas.0914628107

camp: mainstream topic: uv-model method: review direction: complicates verified: full

Key claims: the folate and vitamin-D clines;  skin reflectance correlates better with autumn UV than with annual average;  sunburn and skin cancer "can no longer be considered significant selective pressures";  adequate folate associated with a 72% reduction in neural tube defects.

Used for one thing:  Jablonski herself discarding the folk explanation.  Its weak points, which the article names, are that the human evidence for a folate fitness cost is indirect and that the Inuit counterexample is handled only by invoking diet.  Cited once.

Biological Perspectives on Human Pigmentation

Robins, Ashley H.  Cambridge University Press, 1991.  archive.org isbn_9780521365147 (printdisabled).

camp: mainstream topic: uv-model method: review direction: complicates verified: full (pp. 206-210 only)

Key figures: 457 skeletons from an Iroquoian ossuary near Toronto (AD 1490) with no rickets or osteomalacia recorded;  92 skeletons from an Early Bronze Age site near the Dead Sea, likewise;  Beadle's 1977 calculation that dark skin makes a normal annual amount of vitamin D up to the Arctic Circle on 10.5 percent of body area exposed;  Webb et al. 1988 finding no vitamin D synthesis in Caucasoid skin in Boston from November to February.

The standing scholarly objection to the vitamin-D half of the UV model, and the article now cites it directly rather than through Jablonski.  Robins closes his section "Skin depigmentation:  the vitamin D hypothesis" at p. 208 with "there is not a scintilla of palaeontological or experimental evidence that rickets (or osteomalacia) ever did or would have manifested, regardless of skin colour."  Jablonski & Chaplin 2000 quote that one sentence on their p. 73 and answer it in one sentence, without addressing the skeletal surveys.  Read with bin/iapdsearch.py;  leaf 226 maps to printed p. 208 through BookReaderJSIA.php, which independently confirms Jablonski's own "1991:208".  Only pp. 206–210 were read.

The creationist primaries

One Blood, chapter 4: One race

Ham, Ken, Don Batten & Carl Wieland.  Master Books 1999;  republished on creation.com, 8 February 2006.  https://creation.com/chapter-4-one-race

camp: creationist topic: yec-argument method: hypothesis direction: background verified: full

Key claim: two gene pairs, A/a and B/b;  "an entire range of 'colors,' from very white to very black, can result in only one generation, beginning with this particular type of mid-brown parents."

The canonical statement of the model under test.  Note its logical form:  it does not derive mid-brown ancestors from genetics.  It assumes all the factors were present in Adam and Eve and infers the phenotype.  All the empirical content is in the two-locus additive model.  Cited once.

Are There Really Different Races?

Ham, Ken.  The New Answers Book 1, chapter 17, Answers in Genesis, 29 November 2007.  https://answersingenesis.org/racism/are-there-really-different-races/

camp: creationist topic: yec-argument method: hypothesis direction: background verified: full

Key claims: "four to six genes … control the amount and type of melanin produced";  "we all have approximately the same concentration of melanocytes in our bodies."

The sourcing is the finding.  The "four to six genes" figure is the load-bearing genetic claim of the chapter and its footnote cites J. C. Gutin, "End of the rainbow," Discover, November 1994.  The only technical citations in the pigmentation section are two dermatopathology textbooks, cited for melanosome size, not for gene number.  The melanocyte claim is correct and the article credits it.  Cited.

Chapter 18: How did all the different 'races' arise (from Noah's family)?

Batten, Don (ed.), David Catchpoole, Jonathan Sarfati & Carl Wieland.  The Creation Answers Book, pp. 223–238.  https://creation.com/images/pdfs/cabook/chapter18.pdf

camp: creationist topic: yec-argument method: hypothesis direction: background verified: full

Key claims: the two-locus Punnett square in its most worked-out form;  footnote 9 on p. 229 that "the more genes there are, the easier it is to have a huge range";  p. 233 that natural selection "does not produce skin colours, but only acts on the created capacity";  p. 226 adopting Jablonski's folate mechanism by name.

The best-stated version of the position, and the one with a falsifiable error in it.  Footnote 9 is the load-bearing defence of the two-locus simplification and its arithmetic is backwards:  P(extreme child) = (1/4)ⁿ and the offspring spread narrows as 1/√n.  The "no new information" claim on p. 233 is contradicted by the paper its own footnote 12 cites on the same page, and by Carter's 2017 concession that colour genes are an easy target for mutation.  Page anchoring:  printed page = PDF page + 222, anchored twice on the running feet.  Cited heavily.

Skin colour surprises

Carter, Robert W.  Creation Ministries International, 26 October 2017.  https://creation.com/skin-colour-surprises

camp: creationist topic: yec-argument method: commentary direction: background verified: full

Key claims: the light and dark variants are worldwide, so they were in the original population;  Adam and Eve were middle brown;  the Bantu expansion spread dark variants within the last 2,000 years;  mutations in colour genes are permitted.

The strongest creationist statement available on this topic, published fourteen days after the Science paper it responds to, and it never opens that paper.  The word "Crawford" appears nowhere in it, in the rendered text or the raw HTML;  its sources are a different admixed-population study and Ann Gibbons's Science news summary, and its closing Tishkoff quotation is flagged in Carter's own note as taken from the news piece.  It reports neither the 28.9% variance figure nor any coalescent date.  Its central factual claim, that light-skin alleles are old and worldwide, is nonetheless correct, and the article says so.  Cited.

Were the first Britons dark-skinned?

Sibley, Andrew.  Creation Ministries International, 24 March 2025.  https://creation.com/en/articles/first-britons-dark

camp: creationist topic: yec-argument method: commentary direction: background verified: full

Key claims: early European immigrants most likely had darker skin than today's;  the lighter alleles probably became dominant because of weaker UV-B at northern latitudes;  the disagreement is the "inflated timescale," and the first settlers to Britain arrived 2000–2400 BC.

The current CMI position, and it concedes almost everything except the dates.  It is also the one creationist source in the catalogue that cites its mainstream primary correctly:  reference 2 is the Perretti preprint with the right DOI, reported accurately.  Its closing argument, that the slowness of the change proves the timescale wrong, points the wrong way and the article says so.  Cited.

How Many Races Did God Create?

Answers in Genesis.  https://answersingenesis.org/tower-of-babel/how-many-races-did-god-create/

camp: creationist topic: yec-argument method: commentary direction: background verified: full

Considered and not used in the article.  It is the shortest popular form of the same argument and adds nothing beyond One Blood and the Creation Answers Book, both of which are cited instead.  The page's own datePublished field is truncated in the served HTML to "April 1" with no year, so the year in the library row is unverified and the article should not cite it for a date.  On disk at sources/web/aig-2007c_how-many-races-did-god-create.html.

Considered and not used

• AiG, "How Many Races Did God Create?" Read in full. Not cited; see its entry above.
• Ginger, R. S., et al., J. Biol. Chem. 283(9):5486–5495, 2008, on SLC24A5 knockdown. A block quotation is carried in Eric's notes at /home/share/notes/eric/research/biology/speciation.md, taken from the article's open HTML in an earlier year and not re-verified against the JBC bytes in the research pass. Not cited, because Sturm & Duffy 2012 supplies the same functional point from a CC-BY source that was read directly. It should be re-fetched before any future revision relies on it.
• Hernandez-Pacheco et al. 2017, Sci. Rep. 7:44548. Named only as one of Carter's two references for "the first-ever study on the skin colour of Africans." Not obtained and not cited; it is relevant to the article solely as evidence about Carter's sourcing, which is established from Carter's own page.
• Gibbons, A., "How Africans evolved a palette of skin tones," Science 358(6360):157–158, 2017. Science news, paywalled, not fetched. It matters only as the route by which Carter got his facts, which his own footnote states.

Could not be obtained

• Lamason et al. 2005, Science 310:1782–1786, full text. The origin of the widely repeated figure that SLC24A5 explains 25 to 38 percent of the European-African difference in melanin index. Six routes failed: Europe PMC (not in the open-access subset), the science.org PDF (403), three university mirrors (404), ResearchGate (403), a Wayback id_ fetch of the legacy sciencemag reprint URL (404), and the archive.org sim_science microfilm collection, which does not hold the December 2005 issues. The verbatim abstract was obtained from the Europe PMC REST API and does not contain the figure. A university proxy, an AAAS institutional session, or an interlibrary scan would open it. The article uses Beleza's 7.6–11.4 melanin units and Crawford's 12.8% of variance instead. See quarantine.
• Lalueza-Fox et al. 2007, Science 318:1453–1455, full text. Abstract obtained and cited; body not obtained. OpenAlex reports it closed with no repository full text and there is no PMC record, but the publisher-supplied abstract comes back as raw JSON from the Europe PMC REST service for PMID 17962522, and that is what the article quotes. Nothing from inside the paper is cited. A CSIC institutional repository copy was not tried.
• Crawford et al. 2017 supplementary tables S2–S7. Table S3 lists all genome-wide significant variants per locus and would let this article's arithmetic use the full set rather than the eight top candidates. A separate download from science.org, not fetched; the PMC author manuscript carries the main tables only.
• Adhikari et al. 2019 Supplementary Table 5. Effect sizes of the six well-established conditioning SNPs in the CANDELA sample, which would give a second independent effect-size set. The main PDF is CC-BY and on disk; the supplement is a separate file and was not fetched.
• Elias & Williams on the epidermal-barrier alternative to the UV model, e.g. Am. J. Phys. Anthropol. 2013, doi:10.1002/ajpa.22244. Europe PMC returned NOTFOUND for that DOI. It is the main published rival hypothesis for why dark skin evolved, and an article that reports only Jablonski's model reports one side. Search Europe PMC by title rather than DOI next time.
• Jeanson, Traced (Master Books, 2022), and any Answers Research Journal treatment of pigmentation. Searched and not found; recorded as a search result rather than a fact. See quarantine for the routes tried and the one route left untried.