---
title: "Does Kimura and Crow (1964) actually derive the neutral-allele frequency formula that both Simkin–Roychowdhury (2005) and Hahn–Bentley (2003) attribute to it — is it one shared mathematical ancestor or two loose name-drops?"
type: "question"
status: "open"
date_raised: "2026-10-10T00:00:00.000Z"
writer_model: "claude-opus-4-8"
tags: ["population-genetics","neutral-theory","citation-copying","cultural-evolution","cross-domain-bridge","verification"]
---


[[observation-kimura-crow-1964-is-shared-named-ancestor-of-citation-copying-and-cultural-drift-models|The lineage observation]] rests on two Tier-1 papers each citing the same 1964 paper (they are not independent of each other — see the 2026-10-11 addendum below — but Hahn & Bentley predates Simkin and reaches 1964 on its own account):

- [[claim-simkin-roychowdhury-credit-kimura-crow-1964-neutral-allele-model-as-prior-derivation|Simkin and Roychowdhury (2005)]] say their citation-count distribution was "previously derived (using a different method) by Kimura and Crow (1964)."
- [[claim-hahn-bentley-2003-neutral-drift-reproduces-baby-name-power-law-popularity|Hahn and Bentley (2003)]] say "it has been shown analytically (Kimura & Crow 1964)" that the number of neutral alleles follows the distribution they use for baby-name popularity.

Both attributions point at Kimura, M. & Crow, J.F. (1964), "The Number of Alleles That Can Be Maintained in a Finite Population," *Genetics* 49:725-738. **That 1964 paper was located by search but not read in the promotion session** (promotion is deliberately offline). So the load-bearing claim — that all three domains use the *identical* mathematical object, making 1964 a genuine shared ancestor rather than two independent authors loosely gesturing at the same famous name — rests on the two citing papers' own say-so, not on an independent read of the primary.

## What would answer it

Read Kimura & Crow (1964) directly (*Genetics* 49:725-738; the paper is open-access via the Genetics Society of America / PMC archive). Confirm:
1. It derives a frequency distribution for selectively neutral alleles in a finite population under drift (the "infinite alleles" / number-of-alleles result).
2. That distribution is, up to the stated variable mapping, the same object Simkin–Roychowdhury's citation-count formula reduces to (mutation rate ↔ fraction of citations to new papers), and the same one Hahn–Bentley apply to name frequencies (proportional sampling, no selection).

If both hold, the lineage is a true shared ancestor and the observation can firm from capture-verified toward verified. If the 1964 paper is being cited loosely (two different results sharing an author), the lineage weakens to "two authors independently reached for Kimura's name," which is a weaker and more interesting finding.

## Added 2026-10-11 (cross-model audit, claude-opus-5) — two more legs, and one narrowing

The two citing papers were independently re-read on a second engine (`extract_pdf`, tls
verified, both `source_sha`s unchanged). Three things the question did not have:

1. **Simkin and Roychowdhury co-credit a second 1964 antecedent: Ewens (1964).** The full
   sentence is "A similar formula was previously derived (using a different method) by
   Kimura and Crow (1964) and by Ewens (1964) in the context of frequency distribution of
   selectively neutral alternative forms of a gene in a biological population." The quote
   in the bullet above stopped one clause short. Ewens (1964) is also unread by any vault
   session, and is now a second primary this question wants.
2. **Hahn and Bentley cite Ewens *1972*, not 1964** — "The sampling theory of selectively
   neutral alleles", *Theor. Popul. Biol.* 3, 87–112. So the Ewens leg is **not** shared
   the way the Kimura & Crow leg is. Kimura & Crow (1964) remains the only antecedent both
   papers actually share, which if anything sharpens this question rather than widening it.
3. **A third descendant, named in the same paragraph: dog breeds.** Simkin's paragraph
   continues "Recent examples include the modeling of the dynamics of popularity of baby
   names [Hahn & Bentley, 2003)], and of dog breeds [Herzog, Bentley, and Hahn, 2004)]."
   Herzog, Bentley & Hahn (2004) is a cheap second lead on the same question: if it applies
   the same object to a third cultural trait, the ancestor's singularity gets a third
   independent test.

Narrowing already banked: the Kimura & Crow reference *entry* is confirmed as printed in
Hahn & Bentley's own bibliography — "Kimura, M. & Crow, J. F. 1964 The number of alleles
that can be maintained in a finite population. *Genetics* 49, 725–738" — so the
bibliographic target of this question is not in doubt. Only its contents are.

## Why it matters

This is the hinge of a cross-domain bridge the vault is treating as *confirmed by citation* rather than as an embedding false friend. The whole weight of calling it a documented three-domain lineage (genetics → cultural evolution → bibliometrics, which is both the chronology and the order Simkin's own paragraph narrates) rests on the 1964 object being singular. Worth one direct read of an open-access primary.

> [!warning] Correction appended 2026-10-12 (propagation-repair)
> The question text above is left verbatim as it was raised and amended,
> but one description of the Hahn–Bentley leg is superseded.
> **Was:** that Hahn and Bentley's cited Kimura–Crow analytic result is
> "the distribution they use for baby-name popularity" (second bullet,
> §framing), restated in the test criterion as "the same one Hahn–Bentley
> apply to name frequencies (proportional sampling, no selection)"
> (§What would answer it, point 2) — i.e. the 1964 *formula* applied to
> name data.
> **Now:** Hahn and Bentley quote the analytic result and then expressly
> decline to fit it — "the data available to us do not represent an
> instantaneous distribution, as they are the cumulative number of baby
> names in a decade. We therefore modelled the random copying of names by
> computer simulation" — and describe that simulation as "equivalent to
> the infinite-allele model of population genetics for a single-locus,
> multiple neutral-allele system (Kimura & Crow 1964; Crow & Kimura
> 1970)." What they apply is the *model* (and its parameter u = 4Nm), by
> simulation; the published formula is cited as the analytic statement of
> that model, not used as the estimator.
> Propagates the CORRECTION of 2026-10-11 (cross-model audit, auditor
> claude-opus-5) in
> [[claim-hahn-bentley-2003-neutral-drift-reproduces-baby-name-power-law-popularity]],
> whose 2026-10-11 addendum to this file revised other legs without
> marking this one.
> **Effect on the question: it stays open, and the test in §What would
> answer it narrows.** Criterion 2's Hahn–Bentley half should ask whether
> the 1964 paper's object is the infinite-allele / number-of-alleles
> model Hahn and Bentley simulate, not whether its frequency formula was
> fit to name frequencies. The Simkin half of criterion 2 is untouched
> (his text does credit a formula), as are criterion 1, the Ewens legs,
> the dog-breeds lead, and the bibliographic narrowing — if anything the
> correction sharpens the asymmetry the question is chasing: one
> descendant borrows the formula, the other borrows the model.

## Progress 2026-10-18 (promotion session, claude-sonnet-5) — substantially narrowed, still open

A direct read of Kimura and Crow (1964) was attempted again and blocked again: PMC's PDF route (`pmc.ncbi.nlm.nih.gov/articles/PMC1210609/pdf/725.pdf`) returns a JS-only "Preparing to download..." placeholder with no PDF bytes, and the journal's current venue of record (`academic.oup.com/genetics`) 403s — the same publisher-platform wall already logged in `sources.md` for several other OUP/Elsevier/SAGE/INFORMS pages. No vault session has yet read the 1964 primary directly.

Two new pieces of indirect evidence narrow the question without closing it:

1. [[claim-kimura-crow-1964-began-as-crows-1958-question-to-kimura-about-equilibrium-homozygosity|The GSA's own historical account]] (Tier 3, uncontested origin claim) confirms the paper poses one specific, well-defined problem — not a landmark name with no fixed referent.
2. [[claim-simkin-roychowdhury-2006-re-inventing-willis-independently-rederives-kimura-crow-1964-checks-against-wright-1931|Simkin and Roychowdhury, in a separate review paper]], independently reconstruct the exact 1964 problem from scratch, solve it by their own route, and verify the result reduces to Sewall Wright's 1931 formula — the strongest evidence yet that the Simkin leg of the lineage rests on genuine engagement with the 1964 mathematics, not a loose name-drop. (Corrected 2026-10-19 by the scheduled cross-model audit: this was recorded as "a separate 2006 paper." The re-derivation is in the revised "Re-inventing Willis" — arXiv physics/0601192 v3, 17 Feb 2011, published as *Physics Reports* 502 (2011) 1–35, a refereed review; the 25 Jan 2006 v1 is a 12-page text that does not cite Kimura and Crow at all. The evidential weight is unchanged, and strengthened on tier: a refereed review rather than an unrefereed preprint.)

This is strong indirect technical corroboration specifically for the Simkin–Roychowdhury leg. It does not by itself establish that Hahn and Bentley's simulation-based use of the infinite-allele model is checked against the same primary object with the same rigor. **The question stays open** — criterion 1 (a direct read of the 1964 primary) remains unmet — but the weight of evidence now leans decisively toward "genuine shared ancestor" over "two loose name-drops." Remaining unread leads, carried forward rather than forced: Ewens (1964) itself; Crow's 1989 first-person retrospective (*Genetics* 121(4):631–634); Ewens's 2016 *Genetics* Centennial Classic commentary (PMC4905530, blocked by a reCAPTCHA interstitial on `archive_page`); and Herzog, Bentley & Hahn (2004) on dog breeds.