Rejecting a quotation from publication and erasing it from the record are two different decisions. This page is the record.
What this page is
Most of this article's evidence came from papers in the source library, several of them open access, so its quotations verified straightforwardly. Seven items did not, and they are held here rather than deleted. Three of them are in circulation on this site's own side of the argument, and one of those is an arithmetic error in this project's own long-standing notes.
Status words.
- REFUTED — the attribution, wording or number is demonstrably wrong. A better copy will not help.
- UNVERIFIABLE — no route to the named source has been found yet. That is a statement about this project's reach, not about the claim.
- MISUSED — the text is real and correctly attributed, and quoting it the way it is normally quoted would invert or overstate what its author meant.
- PRINTED WITH A FLAG — the wording is genuine and correctly attributed, but something about it could not be checked, so the article prints it as somebody's claim and says what is unverified.
A standing rule applies to every entry below. Nothing here was established with Claude's WebFetch tool, which converts a page to markdown and answers through a small summarising model that can paraphrase without saying so. Every item on this page was read off downloaded bytes.
1. "Only 234+45+77+4=391 of our 200,000 ERVs are not orthologous"
Status: REFUTED, in three separate places, and it is in this project's own notes at /home/share/notes/eric/research/biology/erv's.md.
Claimed source: the Chimpanzee Sequencing and Analysis Consortium, Nature 437:69–87, 2005, Table 2.
What was found. The PDF and its text extraction are in the source library. Table 2 on p. 75 reads ERV class 1 234 (>1 Mb) 5 (8 kb) and ERV class 2 45 (55 kb) 77 (130 kb). Three errors. The human class-1 cell is 5, not 4. The arithmetic as written, 234 + 45 + 77 + 4, gives 360, not 391. And the correct sum of the four cells is 361.
A fourth problem, which is not arithmetic. Table 2 counts lineage-specific elements in the aligned parts of the two genomes. It is not a whole-genome census, and using it as one overstates how complete the comparison is.
The denominator has no source either. "200,000 ERVs" does not appear in that paper. The citable figures are ~98,000 HERV elements and fragments (Belshaw 2004 p. 4894, itself a citation to the HERVd database) and ~443,000 LTR-retrotransposon copies covering 8.3% of the genome (IHGSC 2001, not obtained for this article).
What the article prints instead: 361 lineage-specific ERV insertions in the aligned genomes, against a background of order 10⁵ shared loci, with the aligned-regions caveat stated.
2. Yohn 2005: "such a series of independent deletion events … is unlikely"
Status: MISUSED. The quotation is verbatim and correctly attributed. Quoting it alone reverses the paper's meaning.
Wording as it circulates. "If these sites were truly orthologous and, thus, ancestral in the human/ape ancestor, it would require that at least six of these sites were deleted in the human lineage. Moreover, the same exact six sites would also have had to have been deleted in the orangutan lineage if the generally accepted phylogeny is correct. Such a series of independent deletion events at the same precise locations in the genome is unlikely."
What was found. Read from the PLoS PDF, p. 579, in the source library. The sentence is Eichler's argument that the apparently shared PtERV1 sites are not really shared, not a concession that the primate tree is in trouble. The next paragraph tests it: "In two cases, we were able to refine the map location to single basepair resolution. Based on this analysis, we determined that the sites were not orthologous between chimpanzee and macaque."
How the article handles it. Both paragraphs are printed together, and the article says plainly that quoting the first without the second misrepresents the source. The genuinely useful part of the paper for the creationist side — that apparent sharing at map resolution can dissolve at base-pair resolution — is printed too, along with the fact that the same test does not dissolve the human-chimpanzee HERV-K sharing.
3. Tomkins's 84% and 68% GULO identity figures
Status: PRINTED WITH A FLAG. The figures are genuine and correctly attributed. They were computed by a method their own author later documented as faulty, and they have not been recomputed.
Wording. "The 28,800 base human GULO region is only 84% and 87% identical compared to chimpanzee and gorilla, respectively. The 13,000 bases preceding the human GULO gene … is only 68% and 73% identical to chimpanzee and gorilla, respectively." Answers Research Journal 7:91–101, 2014, abstract.
What was found. The paper's own body attributes the computation to "the previously established technique of optimized sequence slices and the BLASTN algorithm (Tomkins 2013b)." In Answers Research Journal 8:379–390, 2015, Tomkins reported that every BLASTN version after 2.2.25+ "omitted significant levels of query sequence with the problem generally increasing with progressive algorithm releases," said plainly that this "potentially affects the results of past research done by this author," and revised his whole-genome figure from 70% to 88%. No recomputation of the GULO region was found.
What would change the verdict. A rerun of the GULO comparison with BLASTN 2.2.25+, nucmer and LASTZ, published or supplied by the author. This is a small job and the article says so.
Note. The ARJ PDF of the 2014 paper has a broken font encoding that shifts most characters by 29 ASCII positions and destroys every digit, so no number can be read from it. The figures above were read from the journal's HTML. Anyone checking this paper from the PDF alone will get nonsense.
4. Chang & Slightom 1984, "one-third of the neutral rate"
Status: RELEASED 2026-09-06. Verified against the original and now cited directly.
Wording. "The three hominoid psi beta-globin genes show a high degree of sequence correspondence, with the number of differences found among them being only about one-third of that predicted for DNA sites evolving at the neutral rate (i.e. for sites evolving in the absence of purifying selection)." Journal of Molecular Biology 180(4):767–784, 1984.
What resolved it. The sentence is in the paper's abstract, which is printed on p. 767, so Tomkins's locator is right and his wording is verbatim. The abstract was read as bytes from the Europe PMC REST API (search?query=EXT_ID:6098690&resultType=core) and cross-checked against PubMed. No summarising tool was involved. The body of the paper remains closed at Elsevier, but nothing in the article depends on it.
What it added. The same abstract carries the three shared disabling lesions in human, gorilla and chimpanzee, which the article now quotes: the initiator-codon substitution, the codon-15 stop, and the codon-20 deletion with its frame shift. Tomkins quotes the conservation sentence and not this one.
5. Tomkins's "251 cell types", "16 exon variants" and "42 intron variants"
Status: UNVERIFIABLE. Not quoted in the article.
Claimed source. Tomkins's own reading of UCSC and ENCODE browser tracks, reported in Answers Research Journal 6:293–301, 2013. No independent published source is given for any of the three numbers.
Why they are not printed. The article's treatment of HBBP1 does not need them. Huang 2017's deletion experiment settles the mechanism question these numbers were offered to settle, and it settles it against them: abolishing the HBBP1 transcript changed nothing. Reporting a browser-track count as though it were a measurement would give the reader a false impression of how the claim was established.
What would change the verdict. A published expression study of the HBBP1 locus reporting transcript counts by cell type. Ma et al. 2021 (BBRC 552:157–163) was identified on 2026-09-06 and is still closed at Elsevier; its abstract shows an overexpression result routed through ELK1, which the authors themselves report also binds the γ-globin promoter directly, so it does not supply the missing measurement.
6. Peter Borger on the syncytin-1 locus
Status: RELEASED 2026-09-06. Obtained, verified, and now quoted in the article.
Source. Borger, Peter, "The 'VIGE-first hypothesis'—how easy it is to swap cause and effect," Journal of Creation 27(3):105–112, 2013. Note the title: this project's notes recorded it as "The Design of Life", which is wrong.
What resolved it. Plain curl -sL on https://creation.com/images/pdfs/tj/j27_3/j27_3_105-112.pdf returned the PDF, 639 KB, eight pages, no wall and no bot challenge. Saved to the source library as borger-2013. Folios anchored on the running head printed with the page number on all eight pages, PDF page 1 = 105 and PDF page 8 = 112. All three quotations are on printed p. 108.
What it changed. The article's paraphrase became three verbatim quotations, and the third of them is a concession. Borger writes that "The syncytin genes identified in different orders of mammals do no [sic] show a high level of sequence homology and they map to different genomic locations." He is making the improbability objection, and in the same paragraph he reports the non-syntenic distribution that is the article's answer to it. Answering a creationist argument out of the creationist author's own sentence is better than answering a paraphrase of it.
7. Bentwich 2007 on "GAM" genes
Status: MISUSED, by the paper that cites it. The article quotes both the claim and its source and lets the reader see the gap.
Wording. Tomkins, Answers Research Journal 8:403–411, 2015: "highly specialized gene expression experiments have functionally identified these expressed sequences in this gene as belonging to a group of novel 'genomic address messenger' or 'GAM' genes (Bentwich 2007). These GAM genes have been demonstrated to selectively inhibit the translation of known target genes."
What was found. The paper's own reference list gives Bentwich 2007 as: "Bioinformatically Detectable Group of Novel Regulatory Genes and Uses Thereof. In United States Patent: Rosetta Genomics Ltd. http://www.google.com/patents/US7250496." The phrase "have been demonstrated" therefore rests on a patent's assertions about its own bioinformatic predictions. Nothing in that reference is peer-reviewed and nothing in it is an experiment on this locus.
Why this is here rather than simply refuted. The quotation is real, correctly attributed, and printed in the article. What is quarantined is the inference a reader would draw from the words "have been demonstrated", which the reference does not support. The article prints the quotation and the reference number side by side for exactly that reason.
8. The EvoGrad claims about the vitellogenin locus
Status: PRINTED WITH A FLAG, attributed and not endorsed.
Claims. That the mammalian VTG remnant is about five sequences totalling ~3,100 bp rather than the 150 bp Tomkins analyses; that ten genes upstream and ten downstream of the human locus are shared with chicken apart from an IFI44/IFI44L insertion; that the Bentwich "GAM" sequences appear throughout the genome; and that the Maunakea brain data Tomkins cites amounts to a single read at a locus whose own paper required at least five.
What was found. The post is real, was downloaded, and is in the source library. Its author is pseudonymous, so it is a secondary source. Every one of its claims is checkable at the UCSC genome browser and none of them was independently re-checked for this article.
What the article does. Attributes them to EvoGrad by name, says they were not re-checked, and rests none of its own argument on them. The vitellogenin case in the article rests on Brawand's shared indels and on the patent reference above.
What would change the verdict. Re-deriving the remnant lengths and the flanking gene comparison at UCSC, which is a half-day of work and would let the article make the claim in its own voice.