--- title: "Mitochondrial Eve source audit" description: "Access gaps, citation defects, misleading quotations, and scope limits inherited from the original mitochondrial-Eve dossier and its research files." theme: common/themes/berean/theme.php --- # Source audit This audit records what still needs checking before the new article can treat a source as decisive. It was assembled from the original `literature-review.md`, all original footnote definitions, and the research dossiers. The new article should cite primary bytes where available and state when it is relying on an abstract, a transcription, a secondary report, or a model output. ## Access gaps and partial reads The following source identities were retained because they matter to the history or to fairness, but the original dossier did not obtain a full primary text or a stable page-level locator: Brown 1980; Anderson et al. 1981; Avise, Neigel & Arnold 1984; Wainscoat 1987; Vigilant et al. 1991; Barinaga 1992; Templeton 1992; Hedges et al. 1992; Maddison, Ruvolo & Swofford 1992; Wills 1992; Gibbons 1992; Templeton 1993; Stoneking 1994; Howell et al. 1996; Cavelier et al. 2000; Forster et al. 2002; the full Parsons et al. 1997 paper; Guo et al. 2013; several abstracts in the other-species comparison; and selected blog or popular items. The most consequential incomplete items are Templeton 1993 and the 1991–1992 technical comments. Their conclusions are currently represented through contemporaneous reporting and later reviews. The new article should not put an unverified number or quotation in their mouths. Henn 2009 was previously treated as missing, but a working PDF is now stored in the source library and its abstract and discussion were read. The live Oxford page remains hostile to scripted retrieval. Keep the archived route and the licence warning with any future quotation. The original dossier also contains local computations and downloaded GenBank records. These are research artifacts, not peer-reviewed sources. Cite the underlying accession list and script output when presenting a computed distance or regression, and state the masks, sample inclusion rules, and limitations. ## Citation and quotation defects - **Gibbons 1998:** the sentence about a “mere 6000 years” is authentic, but the next sentence is “No one thinks that's the case.” Humphreys 2005/ICR presents the first sentence as a revision researchers made. That claim is absent from Gibbons and should be described as a misrepresentation. - **Tomkins and related reposts:** several popular pages cut the Gibbons and Parsons passages before their qualifications. Tomkins also inserts a gloss equating the mitochondrial MRCA with “the first human woman.” That gloss is not in the source and changes a gene-tree claim into an individual-origin claim. - **MacAndrew:** the standard online rebuttal says Parsons and Howell used RFLP analysis. Parsons directly sequenced about 610 bp of HVR1 and HVR2. The RFLP criticism is therefore a technical error, even though MacAndrew is right that Parsons did not claim a young human origin. - **Parsons 1997:** older notes say “~60 bp” and “average age.” The primary wording and independent transcriptions support about 610 bp and “an age of the mtDNA MRCA.” The source's own rate arithmetic also rules out 60 bp. - **Loewe & Scherer:** the sentence often attributed to Gibbons about “about 6500 years ago” comes from a different 1997 *Trends in Ecology & Evolution* item. The DOI commonly repeated for it resolves to another article. Do not print the quotation until the journal page is obtained. - **Soares 2009:** the “severe shortage of such uncontroversial episodes” line is a criticism of Endicott & Ho inside Soares, not a concession by the ancient-DNA calibrators. Keep the attribution attached to the correct speaker. - **Madrigal 2012:** the abstract's 1.24 × 10^-6 per-site-per-year headline does not reproduce from the Results or Table 2. The table gives 2.51 × 10^-6 (maximum) and 0.89 × 10^-6 (minimum). Treat the abstract number as an unresolved internal inconsistency. - **Wei 2019:** Table S2 prints 13 D-loop variants, but its rate requires 131; 131 also nearly reconciles the table with the 477 total. Record the dropped-digit correction rather than silently changing the source. - **Connell 2022:** the all-nine mutation count and the fixed-only count are different quantities. The original Otzi table once labeled the all-nine rate as fixed-only. The corrected fixed-only prediction is about 1.1 changes, not 5.1. - **Árnadóttir et al. 2024:** the study detects mutations at a 5% read-frequency threshold. Its raw 0.0703 per-transmission figure is an arising rate, not a fixation rate. The bottleneck conversion is a model step and may over-correct a multi-generation matriline. - **Rebolledo-Jaramillo 2014:** the paper reports an association with maternal age, but it does not test linear versus curved or establish an exponential biological law. The apparent exponential shape comes from a Poisson regression log link. - **Carter 2025 and related CMI pages:** Carter doubles to allow for mutations outside the control region, a defensible approximation. The faulty step assigns the fast control-region per-site rate to all 16,569 sites first. Howell's displayed two-lineage divergence rate also requires halving to get a lineage rate. Calculating the control and coding contributions separately gives about 0.030 changes per 30-year generation from Howell's central values, instead of roughly 0.5–1.5. Carter also presents a sample-dependent 84.1% invariant figure as a species-level fact. - **Jeanson 2015b:** the table arithmetic reproduces from its inputs, but the paper treats variants that disappear from the child as if they were new changes and omits a factor of two used in his other papers. Some “non-African” comparison values are global values. The new article should quote the table with these qualifications. - **Auto-caption media:** Jeanson's 2022 clip and 2026 interview are useful records of what he said, but the VTT files are automatic captions. They have no reliable punctuation, can merge speakers, and can mishear technical words. Do not use a long caption as a verbatim quotation. Cite the timestamp and describe the wording as caption-checked unless the audio has been manually verified. ## Scope and interpretation limits The mitochondrial MRCA is a common ancestor at one maternally inherited locus. It is not automatically the first woman, the only woman alive, or the genealogical ancestor of every person. A genealogy result for all living humans, such as Rohde et al. 2004, cannot be substituted for a mitochondrial coalescence result. Gupta et al. 2026 is a useful new scope-control source. Its UK Biobank and All of Us analysis concerns low-level somatic mtDNA variation detected in blood as age-related clonal haematopoiesis expands. It does not measure a maternal germline mutation rate or mitochondrial-Eve age. The verified CC BY PDF and text extraction are kept at `/home/projects/bereanarchive/sources/papers/gupta-2026_age-related-mtdna-blood.*`. Pedigree studies usually count heteroplasmies appearing above a detection threshold. Phylogenetic and ancient-DNA studies count substitutions that survive drift, selection, bottlenecks, and recurrent mutation. A direct numerical comparison must state the threshold, region, unit, and whether the count is arising or fixing. Ancient-DNA rates after Fu 2013 are independently calibrated by radiocarbon-dated tips, but the date and authenticity of each specimen still matter. Damage treatment can reduce errors without making a contaminated or misdated specimen valid. The source register therefore keeps the damage papers, tip tables, and independent re-dating evidence separate. The Neanderthal and Denisovan distance work includes a complete/near-complete GenBank inventory and in-house tests. The distance ratios and transversion checks are useful computations, but they do not become published measurements merely because the input sequences are public. The Denisovan high-altitude argument has an additional scope limit: the divergent Denisovan mtDNA came from Denisova Cave at roughly 700 m, while high-altitude Baishiya evidence is sedimentary mtDNA or proteomic identification. ## Research artifacts to retain with provenance The original dossier's branch-shortening regression, root-topology reanalysis, archaic distance matrix, site-class analysis, Otzi panel, and repair-gene scans should be copied only as clearly labeled research artifacts. Preserve the input accession lists, output tables, exclusion masks, and code references. They are not substitutes for the source papers and should not be described as independently replicated peer review. The complete original footnote ID map is in `citations.md`. `link-check.tsv` records HTTP results from a bounded request sweep. A 403 or 429 is classified as blocked/rate-limited, not dead. A 404, timeout, or connection failure still needs a second route before the source is declared unobtainable. ## Priority follow-up 1. Obtain and read Maddison 1991, Templeton 1992–1993, Hedges 1992, and Stoneking 1994 before using their exact statistics. 2. Verify any proposed quotation against the stored PDF, HTML, or audio, including its page, section, table, or timestamp. 3. Resolve the Loewe & Scherer bibliographic identity and retrieve the printed *Creation Ex Nihilo Technical Journal* item behind Wieland's reference numbering. 4. Keep the full deCODE article and the oocyte paper as primary sources for threshold and maternal-age claims; do not rely on press summaries. 5. Reconcile the new article's final footnotes against the retained-source labels after prose is written.