---
id: "20261120-0207-bipartite-two-things-this"
title: "Bipartite bridge check: Eliot's hedged Antarctic footnote and Griewank's hedged priority prologue are both unflagged instances of the vault's own Stigler's-Law-of-Eponymy claim"
type: "capture"
status: "promoted"
origin: "batch"
promoted_to: ["claim-griewank-2012-prologue-complicates-trefethens-sole-credit-attribution","observation-eliot-griewank-bridge-is-stiglers-law-hedging-intermediary-eponym","entity-ernest-shackleton","entity-stephen-stigler","entity-t-s-eliot"]
not_promoted: ["Eliot-hedge / Third-Man-Factor etymology claim (the capture's first ## Claim) — duplicate of the existing claim-third-man-factor-name-descends-from-eliot-waste-land, same quote and same conclusion; not re-promoted, cited by wikilink instead.","Griewank-declines-credit claim's downstream-citation half (the capture's second ## Claim, its back half) — duplicate of existing claim-griewanks-name-reaches-differentiable-physics-literature-further-than-linnainmaas-or-speelpennings and claim-difftaichi-2020-names-griewank-and-wengert-but-not-linnainmaa-or-speelpenning; folded into the new Trefethen-specific note by reference, the new Trefethen material promoted on its own.","Further-leads items (Stigler 1980 primary still unread, Trefethen's own listing not independently fetched, StereoMath/Raman hedging-intermediary re-check, Road-to-Emmaus eponymy angle) — none are load-bearing doubts a kept claim rests on beyond the already-routed Stigler question; left as leads in this capture, not promoted to 50-questions/.","Entity candidate Nick Trefethen — person, UNSURE per the entity promotion test: only one quote this session, his own listing not independently fetched, no second vault mention beyond the new claim-note. Bias against the flood; no hub page created. Named in the new claim-note's prose without a wikilink.","Entity candidates Robert K. Merton, Seppo Linnainmaa, Andreas Griewank — all already have hub pages; checked for new material and updated in place (Griewank, Linnainmaa) or found already covered (Merton — the Stigler-credits-Merton fact is already reachable from his page via claim-credit-detectors-are-themselves-misattributed; no new fact about Merton himself surfaced this session)."]
writer_model: "claude-sonnet-5"
date_created: "2026-11-20T00:00:00.000Z"
provenance: "Batch capture run, 2026-11-20, investigating the bipartite (epistemology/mathematics) pairing of claim-third-man-factor-name-descends-from-eliot-waste-land and claim-griewanks-name-reaches-differentiable-physics-literature-further-than-linnainmaas-or-speelpennings, cosine 0.85"
derived_from: []
tags: ["eponymy","stiglers-law","credit-assignment","automatic-differentiation","third-man-factor","linnainmaa","griewank","eliot","cross-domain-bridge","bridge-investigation","bipartite"]
source_url: "https://papers.baulab.info/papers/also/Griewank-2012.pdf"
source_sha: "12b9cbdd0897a8e0715f39c85bb99a844970e9bd64a7c2e4898eeaa758b51bce"
source_title: "Who Invented the Reverse Mode of Differentiation?"
source_author: "Andreas Griewank"
source_date: "2012"
source_venue: "Documenta Mathematica, Extra Volume: Optimization Stories (2012), pp. 389-400"
source_tier: 1
seek_code_commit: "d40ee39"
---


**Bridge-seed check performed first:** searched `10-inbox/raw/` for any existing capture naming both `claim-third-man-factor-name-descends-from-eliot-waste-land` and `claim-griewanks-name-reaches-differentiable-physics-literature-further-than-linnainmaas-or-speelpennings` by slug. No capture names both; the only file matching the Third Man Factor slug is `2026-07-09-hop-third-man-factor.md`, the ancestor capture that was promoted into the claim-note itself, and it never engages automatic differentiation. This is a fresh investigation, not a duplicate.

The two seed notes share no vocabulary: one is about a modernist poem's footnote and a clinical hallucination's name; the other is about citation patterns in differentiable-physics papers. Per the task framing, if the cosine-0.85 resemblance is real, it has to be a mechanism, not a shared word. Reading both seed notes' own primary sources side by side turns up exactly that — and the mechanism is already sitting in the vault, unconnected to either side of this pairing until now.

## Claim: Eliot's note names its own source but hedges on which account it was, and the clinical/popular term descends from his hedge-containing reformulation, not from the explorer's own words

**Claim type:** historical/biographical (uncontested per the sourcing floor; Eliot's own published note, quoted directly). Tier 3 — already the vault's own standing sourcing tier for this document per `claim-third-man-factor-name-descends-from-eliot-waste-land`; re-verified fresh this session via `archive_page` (new receipt below) rather than reused from the existing claim-note's frontmatter.

T.S. Eliot's note to line 360 of *The Waste Land* (1922) reads:

- source_quote: "The following lines were stimulated by the account of one of the Antarctic expeditions (I forget which, but I think one of Shackleton's): it was related that the party of explorers, at the extremity of their strength, had the constant delusion that there was one more member than could actually be counted."
- source_url_2: "https://wasteland.windingway.org/360/who-is-the-third-who-always-walks-beside-you"
- source_sha_2: "23a7e6d6e6aaffc91abbd2b05a8d1a843e34f7a50093038eac73e88b124fea5e"
- source_title_2: "Who is the third who always walks beside you"
- source_author_2: "T.S. Eliot (note to The Waste Land, l. 360); annotated by windingway.org"
- source_date_2: "1922"
- source_venue_2: "T.S. Eliot, The Waste Land (1922), author's own notes; annotated-text resource"
- source_tier_2: 3

Eliot names a real source (an Antarctic expedition memoir) but is explicit that he is working from an uncertain memory of it, not a checked citation: "I forget which, but I think one of Shackleton's." The modern clinical/popular name, "Third Man Factor," descends from Eliot's own line ("Who is the third who walks always beside you?") — itself already established in the vault's existing claim-note — not from Ernest Shackleton's own phrasing in *South* (1919), "it seemed to me often that we were four, not three." The name that survives into later usage is the hedging intermediary's reformulation, not the original witness's words.

## Claim: Griewank's 2012 prologue explicitly declines sole credit and names the earlier independent discoverers by name — yet it is Griewank's own name, not theirs, that reaches downstream differentiable-physics citation practice

**Claim type:** historical/technical-mechanism (a specific, checkable pattern in how credit is distributed in a primary document versus how it is cited downstream). Tier 1 — direct `extract_pdf` read of Griewank's own essay, fetched fresh this session.

Griewank opens "Who Invented the Reverse Mode of Differentiation?" (*Documenta Mathematica*, Optimization Stories, 2012) by reporting, and immediately complicating, credit that had been assigned to him alone:

- source_quote: "Nick Trefethen [13] listed automatic differentiation as one of the 30 great numerical algorithms of the last century. He kindly credited the present author with facilitating the rebirth of the key idea, namely the reverse mode."
- source_quote_2: "In fact, there have been many incarnations of this reversal technique, which has been suggested by several people from various fields since the late 1960s, if not earlier."
- source_quote_3: "Seppo Linnainmaa (Lin76) of Helsinki says the idea came to him on a sunny afternoon in a Copenhagen park in 1970."
- source_quote_4: "Bernt Speelpenning (Spe80) arrived at the reverse mode via compiler optimization when Bill Gear asked him to automatically generate efficient codes for Jacobians of stiff ODEs."
- source_tier: 1

The prologue goes on to name, by name, Gerardi Ostrowski, Hachtel et al., Iri, Bennett, Paul Werbos, Baur & Strassen, and Speelpenning as independent prior discoverers or contributors to the reverse mode, before Griewank even mentions his own 1987 rediscovery. The essay's own title is phrased as an open question, not an assertion of his own priority. Despite this, the vault's own already-sourced finding in [[claim-griewanks-name-reaches-differentiable-physics-literature-further-than-linnainmaas-or-speelpennings]] shows that in the ChainQueen/Bolliger/DiffTaichi cluster, it is Griewank's name — not Linnainmaa's or Speelpenning's, both named explicitly in Griewank's own prologue above — that reaches downstream citation as the standard reference for the technique ([[claim-difftaichi-2020-names-griewank-and-wengert-but-not-linnainmaa-or-speelpenning]]). The erasure happens after Griewank's own text, not inside it.

## Claim: Both cases are unflagged instances of a mechanism the vault already names — Stigler's Law of Eponymy — which neither seed note, nor the vault's own prior false-friend check on a closely related Griewank pairing, had connected to either side of this bridge

**Claim type:** synthesis / mechanism identification, resting on two already Tier 1–3 sourced primaries quoted above plus an existing vault claim-note. Marked `[unverified-synthesis]`: this is Seek's own connective read linking two already-sourced claims, not a claim any single external source asserts directly.

The vault already holds a claim-note naming exactly this general pattern: [[claim-credit-detectors-are-themselves-misattributed]] states "Stigler's Law of Eponymy" — "no scientific discovery is named after its original discoverer," per Stephen Stigler's 1980 contribution to a festschrift for Robert K. Merton — as a standing vault concept. That note's own `audit_status` flags that the Stigler half still rests only on a Tier-4 Wikipedia source, because Stigler's 1980 essay itself has not yet been read by any session (`[[question-verify-stiglers-law-and-matthew-effect-primaries]]`, item 1, still open as of this capture). This capture inherits that flag rather than resolving it: a search this session for the primary (Stigler, "Stigler's Law of Eponymy," *Transactions of the New York Academy of Sciences* 39, 1980, pp. 147-157, reprinted in *Science and Social Structure: A Festschrift for Robert K. Merton*) turned up only a Wiley paywall page and an Internet Archive lending-library scan (`archive.org/details/sciencesocialstr0039unse`) that serves only its borrow-landing page to automated fetch, not the text — so the primary remains unread. "No scientific discovery is named after its original discoverer" is cited here as a definitional/uncontested framing of a named phenomenon (floor-acceptable at Tier 3-4), not re-asserted as independently confirmed on the self-referential Merton-attribution point, which stays exactly as flagged in the existing note.

What this capture adds is new: neither the Third Man Factor claim-note nor the Griewank claim-note had been linked to Stigler's Law before, and the vault's own closest prior search for a matching mechanism — [[observation-stereomath-griewank-cosine-pairing-is-embedding-false-friend]], which tested the Griewank/Linnainmaa/Speelpenning asymmetry against "obliteration by incorporation" and the Matthew Effect and rejected both — never tested Stigler's Law against that same asymmetry, despite Stigler's Law being the better fit of the two already-vault-held mechanisms (OBI requires erasure into anonymity; the Matthew Effect requires credit transferring to a *different*, more eminent party than the one who already holds it; Stigler's Law requires neither — only that the name attached to a discovery not be its original discoverer's, which is exactly what the DiffTaichi citation data shows for Griewank over Linnainmaa and Speelpenning).

The refinement this pairing surfaces, beyond plain Stigler's Law: in both the Eliot and the Griewank case, the person whose name ends up eponymously attached (Eliot's line for the clinical term; Griewank's textbook essay for the technique) is also the one whose own primary text explicitly hedges on, or explicitly redistributes, priority toward an earlier named source — Eliot's "I forget which, but I think"; Griewank's naming of eight-plus earlier independent claimants before his own 1987 rediscovery. The hedge does not stop the eponymy. Downstream users stop at the hedging intermediary's text rather than following the hedge back to the name it points at.

## Verdict on the bipartite pairing

Not a false friend. The cosine-0.85 resemblance between the two seed notes is real and is a mechanism, confirmed by reading both notes' own primary sources directly rather than inferring from vocabulary: both are instances of the vault's existing Stigler's-Law-of-Eponymy claim, a connection that existed in neither seed note nor in the vault's own prior adjacent search for a matching mechanism on the Griewank side. The sharper, narrower finding — that the eventually-eponymous name in both chains belongs to a hedging intermediary, not a confident claimant — is this capture's own synthesis and is flagged as such; the general Stigler's-Law claim it specializes remains exactly as under-sourced as the vault's pre-existing note already says, and this capture does not upgrade that flag.

## Further leads

- The Stigler 1980 primary essay remains unread by any session (`question-verify-stiglers-law-and-matthew-effect-primaries`, item 1); a library/JSTOR/Wiley route or a usable Internet Archive lending checkout would close a flag that now bears on three vault claim-notes instead of one.
- Nick Trefethen's own listing ("the 30 great numerical algorithms of the 20th century," SIAM News, 2000ish) that first credited Griewank alone with "facilitating the rebirth" of the reverse mode was not independently fetched this session — reading it directly would confirm whether Griewank's prologue is correcting Trefethen's actual wording or a looser paraphrase of it.
- Whether other "founding-name-erased" cases already in the vault (StereoMath/Raman, per [[observation-stereomath-griewank-cosine-pairing-is-embedding-false-friend]]) also fit the "hedging intermediary becomes the eponym" refinement, or only the plainer Stigler's-Law shape, was not checked this session.
- The unremarked Road-to-Emmaus biblical layer under Eliot's "third" (noted in the existing claim-note) was not re-examined for its own eponymy angle (the gospel account itself is anonymous-author testimony later attached to named evangelists) — a plausible third leg, not chased here.

## Entity candidates

- Robert K. Merton — person — the foundational figure Stigler's own law is credited to, and the deepest point either chain's priority-attribution rests on; already has vault presence via [[entity-robert-merton]] per `claim-credit-detectors-are-themselves-misattributed`, flagged first here per the known blind-spot instruction (the figure the ancestry claim rests on, not the paper's own co-authors).
- Seppo Linnainmaa — person — the actual 1970 originator in the Griewank chain, already a vault entity ([[entity-seppo-linnainmaa]]); flagged again because this capture is the first to frame his erasure specifically as a Stigler's-Law instance rather than an OBI/Matthew-Effect candidate.
- Ernest Shackleton — person — the actual originating witness in the Eliot chain (via his memoir *South*, 1919); no `entity-ernest-shackleton` file located by search — flagged first per the same blind-spot instruction, ahead of Eliot himself.
- Stephen Stigler — person — names the general law this capture leans on; no `entity-stephen-stigler` or `entity-stiglers-law-of-eponymy` file located — the mechanism hub this whole bridge rests on has no dedicated page yet, only the claim-note.
- Andreas Griewank — person — already a vault entity ([[entity-andreas-griewank]]); this capture adds the specific fact that his own 2012 prologue redistributes, rather than claims, priority.
- T.S. Eliot — person — no `entity-t-s-eliot` file located by search; the hedging intermediary on the epistemology side of this bridge.
- Nick Trefethen — person — no vault entity; newly surfaced this session as the one whose confident sole-credit framing is what Griewank's prologue is explicitly responding to and complicating — the trigger for the hedge, not the hedge itself.
