# Sources — Selberg/Levinson primer (critical-line proportions)

Gathered 2026-11-11, block 1 (≈9:15–11:15 PM PT); updated block 2 (11:30 PM–1:15 AM PT) and block 3 (Nov 12, 1:36–3:20 AM PT: Hardy folio, Titchmarsh §10, Selberg/Levinson access): §2026 re-verified on arXiv abs pages, Hardy URL, Levinson 1975 venue, numerical history, Speiser. "Checked" = the figure/year was read in the source itself
(or, where noted, in a source that quotes the original's own theorem statement). Local cache of PDFs used for
checking: /tmp/slsrc (scratch, not kept).

Notation: N(T) = zeros with 0<γ≤T; N0(T) = those on Re s = 1/2; κ = liminf N0/N; κ* = same for simple zeros.

## Primary papers (the proportion ladder)

| Year | Who | Result (as stated in source) | Link | Checked how |
|---|---|---|---|---|
| 1914 | G. H. Hardy, "Sur les zéros de la fonction ζ(s) de Riemann", note transmise par J. Hadamard, séance du 6 avril 1914, C. R. Acad. Sci. Paris 158, 1012–1014 | infinitely many zeros on the line ("parmi les zéros de ζ(s), il y en a une infinité sur la droite σ = ½") | Gallica, C. R. t. 158 (Jan–Jun 1914): https://gallica.bnf.fr/ark:/12148/bpt6k3111d (volume ark found via search index; Gallica itself returned a Cloudflare block to curl, so the exact folio for p. 1012 is not confirmed on Gallica itself). **Block 3:** the archive.org copy (https://archive.org/details/ComptesRendusAcademieDesSciences0158) has metadata source = http://gallica.bnf.fr/ark:/12148/bpt6k3111d (i.e. it *is* that Gallica volume), 2116 page images; its OCR (djvu.xml, read) puts the running head "1012 ACADÉMIE DES SCIENCES. ANALYSE MATHÉMATIQUE. — Sur les zéros de la fonction ζ(s)…" on **image 1014** (1-based; = archive.org page n1013), p. 1013 on image 1015, p. 1014 on image 1016 — the same offset (+2) as Wikisource. Hence **Gallica folio f1014 is the best inference** (https://gallica.bnf.fr/ark:/12148/bpt6k3111d/f1014.item), *unverified on Gallica* (still Cloudflare-blocked; the Gallica Pagination service also failed). Block 1/2's "likely f1016" was p. 1014, the note's last page — corrected. Direct archive.org page link: https://archive.org/details/ComptesRendusAcademieDesSciences0158/page/n1013/mode/1up . Page-level transcription (Wikisource, from the same scan): https://fr.wikisource.org/wiki/Page:Comptes_rendus_hebdomadaires_des_s%C3%A9ances_de_l%E2%80%99Acad%C3%A9mie_des_sciences,_tome_158,_1914.djvu/1014 (printed p. 1012; djvu/1015 = p. 1013, djvu/1016 = p. 1014). Full volume also at https://archive.org/details/ComptesRendusAcademieDesSciences0158 . Collected Papers reprint: not checked. | **Block 2: opened** the Wikisource transcription of pp. 1012–1014: title, "Note de M. G.-H. Hardy, transmise par M. J. Hadamard", the italicised claim quoted, method = Cahen–Mellin formula → integral of Ξ(2t) against (e^{αt}+e^{−αt})t^{2p}/(¼+4t²), contradiction if Ξ(2t) keeps one sign for t > T. Footnote: result first communicated to the London Mathematical Society, 12 March 1914. |
| 1921 | Hardy & Littlewood, "The zeros of Riemann's zeta-function on the critical line", Math. Z. 10, 283–317 | N0(T) ≫ T | https://doi.org/10.1007/BF01211614 (Crossref-verified) | citation via Montgomery–Vaughan notes; statement via Iwaniec Thm 15.1 ("For T ≥ 15, N0(T) ≫ T") |
| 1942 | A. Selberg, "On the zeros of Riemann's zeta-function", Skr. Norske Vid. Akad. Oslo I 1942 no. 10, 1–59 | N0(T) ≫ T log T, i.e. κ > 0 (constant tiny, not computed in paper) | IAS archive record https://archives.ias.edu/repositories/2/archival_objects/35119 ; companion notes scan (Arch. Math. Naturvid. 45 (1942) 101–114 and Norske Vid. Selsk. Forh. 15 (1942) 59–62) https://sites.math.rutgers.edu/~zeilberg/EM18/selberg1942.pdf | scan read: Arch. Math. note proves N0(2T)−N0(T) > K T logloglog T; Forh. note states Theorem 1: N0(T) > A T log T. Main Skr. paper not opened. **Block 2:** re-checked — the Rutgers PDF is only 9 image pages (the two companion notes); no open scan of the 59-page Skr. paper found (IAS record, ULB Düsseldorf catalogue only). Main statement N0(T) > A T log T therefore quoted from the Forh. note + Iwaniec (16.1) + Titchmarsh §10.9 title, **not** from the Skr. paper itself. **Block 3:** archive.org advanced search (Selberg / Skrifter 1942 / Collected Papers) found nothing; web search again finds only the IAS record, ULB Düsseldorf catalogue and the Rutgers companion-notes PDF. Still not opened. Full proof now read in Titchmarsh 2nd ed. §§10.9–10.22 (Thm 10.22), which follows Selberg's method. Side facts seen: the Arch. Math. note (Rutgers scan) says in a proof-correction remark that log log log T "may be replaced by log T"; Montgomery–Vaughan's references list 1942b as Collected Papers vol. 1 pp. 156–159 (4 pages for a 59-page paper — looks like the Forh. note; not resolved, don't cite those page numbers). |
| 1974–75 | N. Levinson, "More than one third of zeros of Riemann's zeta-function are on σ = 1/2", Adv. Math. 13 (1974) 383–436; PNAS notes: "At least one-third of zeros of Riemann's zeta-function are on σ = ½", PNAS 71 (1974) 1013–1015; "Generalization of recent method giving lower bound for N0(T) of Riemann's zeta-function", PNAS 71 (1974) 3984–3987; "Deduction of semi-optimal mollifier for obtaining lower bound for N0(T) for Riemann's zeta-function", PNAS 72 (1975) 294–297 | κ > 1/3; paper's constant κ = 0.3420 (Iwaniec (16.2); Arias de Reyna 2026 §1); **0.3474 is the 1975 PNAS semi-optimal-mollifier paper** (Arias de Reyna 2026 cites [17] = PNAS 72 (1975) 294–297 for κ ≥ 0.3474) | https://doi.org/10.1016/0001-8708(74)90074-7 ; https://doi.org/10.1073/pnas.71.4.1013 (PMC388151) ; https://doi.org/10.1073/pnas.71.10.3984 (PMC434311) ; https://doi.org/10.1073/pnas.72.1.294 (PMC432290) | **Block 2:** Adv. Math. full text NOT accessible (ScienceDirect blocks scripted access; WebFetch timed out). **Block 3:** Crossref lists an Elsevier *open-archive* user licence for the VoR from 2013-07-17 (so it should be free to read at the DOI in a normal browser), but scripted access again 403 (curl) / timeout (WebFetch); Unpaywall had no record. Also reprinted in Levinson's Selected Papers (Birkhäuser), https://doi.org/10.1007/978-1-4612-5335-8_28 (not opened). Still not read. Abstracts of the three PNAS notes read on PMC (quoted in mechanism-notes §9). PMC PDFs blocked. Correction: block 1 said ".3474 from a 1975 follow-up" without the venue — it is PNAS 72 (1975) 294–297, whose abstract says the optimized mollifier improves the 1974 result "only 1.4%" (0.3420×1.014 ≈ 0.347, consistent). Simplicity: Heath-Brown, Bull. LMS 11 (1979) 17–18, and Selberg independently (Conrey 1989 p. 1; Iwaniec ch. 23). |
| 1983 | J. B. Conrey, "Zeros of derivatives of Riemann's ξ-function on the critical line", J. Number Theory 16, 49–74 | κ ≥ 0.365 (general P, θ = 1/2) | https://doi.org/10.1016/0022-314X(83)90031-8 (Crossref-verified) | quoted in Conrey 1989 intro ("at least 0.365") |
| 1985 | Balasubramanian, Conrey, Heath-Brown, J. reine angew. Math. 357, 161–181 | mollifier length past 1/2 (θ < 9/17) | https://doi.org/10.1515/crll.1985.357.161 (Crossref-verified) | quoted in PRZZ §1 |
| 1989 | J. B. Conrey, "More than two fifths of the zeros of the Riemann zeta function are on the critical line", J. reine angew. Math. 399, 1–26 | **Theorem 1: κ ≥ 0.4077, κ* ≥ 0.401**; "Note added in proof": κ > 0.4088. Mollifier length T^θ, θ = 4/7 (Deshouillers–Iwaniec Kloosterman bounds) | https://doi.org/10.1515/crll.1989.399.1 ; PDF https://maths.dur.ac.uk/users/herbert.gangl/morethan2_5.PDF | read in PDF (Thm 1 on p. 3; note in proof at end) |
| 2011 | H. M. Bui, B. Conrey, M. P. Young, "More than 41% of the zeros of the zeta function are on the critical line", Acta Arith. 150, 35–64 | **κ ≥ .4105, κ* ≥ .4058** (two-piece mollifier) | https://arxiv.org/abs/1002.4127 ; https://doi.org/10.4064/aa150-1-3 | read Theorem 1.1 in arXiv v2 |
| 2012 | S. Feng, "Zeros of the Riemann zeta function on the critical line", J. Number Theory 132, 511–542 | **κ ≥ 0.4128** (mollifier with ζ-derivative-type pieces) | https://arxiv.org/abs/1003.0059 ; https://doi.org/10.1016/j.jnt.2011.10.002 | read Theorem 1 in arXiv. **Caveat:** PRZZ §1 records that a gap was found (Bui 2014, Kühn–Robles–Zeindler 2017, Pratt–Robles 2018, Robles–Roy–Zaharescu 2016) reducing θ_F to 3/7 (κ > 0.4107); Pratt–Robles (Res. Number Theory 4:9, 2018) then validated θ_F = 1/2 − ε and hence κ > 0.4128. |
| 2020 | K. Pratt, N. Robles, A. Zaharescu, D. Zeindler, "More than five-twelfths of the zeros of ζ are on the critical line", Res. Math. Sci. 7 (2020), Paper No. 2, 74 pp | **κ > 0.417293, κ* ≥ 0.407511** (5/12 = 0.41667) | https://arxiv.org/abs/1802.10521 ; https://doi.org/10.1007/s40687-019-0199-8 | read abstract/intro in arXiv PDF; journal year 2020 vol 7 confirmed (Pratt CV, Lancaster EPrints lists 2019 online) |

### §2026 — the "≥ 2/3" preprints (block 2: every entry re-checked on its real arXiv abs page, 11:33–11:50 PM PT Nov 11)

The "record = 5/12" framing is out of date as of Nov 2026, **for the proportion of simple zeros on the line**.
These are arXiv preprints: none has a journal-ref on its abs page (checked 2026-11-11). They use a *different*
method (Montgomery pair correlation + linear algebra / Hilbert-space inequality, no mollifier). Upload times
below are arXiv's (UTC), converted to Pacific (PDT, UTC−7 in Aug–Oct).

| arXiv | Authors (abs page) | Title (abs page) | Versions | Main claim (read in the paper) | Comments field / journal-ref | Verdict |
|---|---|---|---|---|---|---|
| [2608.13637](https://arxiv.org/abs/2608.13637) | Levent Alpöge, Ralph Furman | "More than two thirds of the zeta zeros are simple and on the critical line" (PDF/HTML title: "…of the zeros of the Riemann zeta function are simple…") | v1 Aug 13 2026, 10:44 AM PDT; v2 Aug 19 2026, 11:17 AM PDT (HTML body dated Aug 24 2026) | **Theorem A:** N0^s(T,2T) ≥ (2/3 − o(1)) N(T,2T) and N_d(T,2T) ≥ (5/6 − o(1)) N(T,2T), unconditionally; with the Montgomery–Taylor window 0.67250… and 0.83625…; Theorem B: same for primitive Dirichlet L-functions. Method: Weil's explicit-formula Hermitian form compressed to a d×d matrix (d ≈ N(T,2T)), rank–trace inequality rank P ≥ 2 tr P + 4 tr Q − 4b − ‖P+Q‖²_HS, Sylvester inertia for off-line pairs; prime-side input = unconditional Montgomery second moment (Aryan 2022; BGSTB 2024). States its own ceiling ≈ 0.682 for "bandwidth-one certificates" and §1.4 "What the results are not": remaining third not shown off the line; method "insensitive to o(N) off-line zeros"; RH out of reach. | Comments: "21 pages. Proof discovered autonomously by Claude (Anthropic); verified and communicated by the listed authors. See §1 for provenance. Lean formalization available". No journal-ref. Lean repo cited: github.com/anthropics/zeta-23-lean (tag v1.0), "only the three standard axioms" per the paper. | **CONFIRMED** (exists, claim as stated). Refereeing: none shown. |
| [2609.02882](https://arxiv.org/abs/2609.02882) | Youness Lamzouri (Université de Lorraine, IECL) | "A new proof that more than 2/3 of the zeros of the Riemann zeta function are simple and on the critical line" | v1 Sep 2 2026, 10:56 AM PDT; v2 Sep 8 2026, 10:55 AM PDT | **Theorem 1.1:** liminf N0^s/N ≥ C0 = 3/2 − (1/√2)cot(1/√2) = 0.67250…; liminf N_d/N ≥ 0.83625…; new: liminf N_{s∪0}/N ≥ 0.88762…, and (N_s+N0)/(2N) ≥ 0.83625…. Method: one Hilbert-space inequality (Prop. 2.1, Bessel/Gram–Schmidt) replacing the matrix framework, + BGSTB 2024 Lemma 5. Thanks Conrey, Goldston, Gonek, Granville, Milinovich, Tenenbaum for reading a draft. Appendix A: AxiomProver Lean certificates (github.com/AxiomMath/ZetaZerosV2), conditional on BGSTB Lemma 5 + Riemann–von Mangoldt as assumptions. | Comments: "17 pages. We establish two new unconditional results: … 88.76% … 83.62%. We also expand the exposition". No journal-ref. | **CONFIRMED.** Independent human proof of the 0.6725 bound. |
| [2610.08965](https://arxiv.org/abs/2610.08965) | Kristian Muri Knausgård (Kristiansand, Norway) | "More than 83.9% of the zeros of the Riemann zeta function are distinct and more than 67.35% are simple and on the critical line" | v1 Oct 6 2026, 11:31 AM PDT (only version) | **Thm 1.3:** liminf N0^s/N ≥ 1669159/2478195 = 0.67353820…; **Thm 1.1:** N_d/N ≥ 0.83900…; Thms 1.4–1.5: 0.83676…, 0.88931…. Computer-assisted 7-/8-point local inequalities (branch-and-bound, ~5×10^8 boxes); Lean 4 proofs "use three axioms beyond those of Mathlib, one per inequality. Each records that a search program returned true; Lean's kernel does not check the run itself." Abstract: "This work is an experiment in AI-assisted mathematical research." Also lists many unrefereed competing claims 0.67300–0.673492 it has not verified. | Comments: "43 pages, 3 figures, 13 tables. Lean 4 proofs, certificate data and source code are in the ancillary files…" No journal-ref. | **CONFIRMED** (exists, claim as stated). Weakest-checked of the three: depends on computer searches trusted as axioms. |

**Critical comment found (block 2):** J. Arias de Reyna (Univ. Sevilla), "Human Exposition of Claude's 2/3 Theorem
on the zeros of zeta", [arXiv:2610.24612](https://arxiv.org/abs/2610.24612), v1 Oct 17 2026, 3:33 AM PDT, 29 pp., no
journal-ref. He re-proves N0^s ≥ 0.6725 N and N_d ≥ 0.83625 N (his Theorem 7) and writes (§1, quoted from arXiv
HTML): *"It took me a month of study before I was convinced it was correct. The paper [1] is like the class notes of
a poor student. As it's written, [1] doesn't work, but I don't think the necessary modification deserves more than a
note."* — i.e. **a fixable error in the written Alpöge–Furman proof, not a refutation of the theorem.** MathOverflow
question 516206 "Reliability of Anthropic's Lean verifications" (Henri Cohen, Oct 20 2026) discusses this; a
commenter points to footnote 1 p. 15 (§7) of the current arXiv version as the specific wrong assertion; Arias de
Reyna (as "juan") replies it is "notable, but not the only wrong assertion… they are easily corrected."
(An AI-generated "Pith review" of 2610.24612 claims his own Prop. 14 has an algebra error; that is not a reliable
source and I did not verify it — **not to be cited**.) Press: Live Science (Oct 1, via a secondary blog) — not used.
No withdrawal and no refutation of the theorem found in searches on 2026-11-11.

Inputs these rest on: Baluyot–Goldston–Suriajaya–Turnage-Butterbaugh, "An unconditional Montgomery theorem for pair
correlation of zeros", Acta Arith. 214 (2024) 357–376 (arXiv 2306.04799); Aryan, J. Number Theory 233 (2022)
389–404. Background: Goldston–Suriajaya arXiv 2511.20059 (Lamzouri lists it "to appear in Anal. Math. 52 (2026)") —
conditional/expository.

**How to label in the primer (decision, block 2):** keep all three (none DROPPED). Wording:
"[recent preprint, 2026] Unconditional ≥ 67.25% of zeros simple and on the line (Alpöge–Furman, argument found by an
AI system and checked by the authors; independent human proof by Lamzouri; small further gain to 67.35% by
Knausgård, computer-assisted). Not yet refereed; an expository re-proof by Arias de Reyna says the first write-up
needed (easy) corrections." The **Levinson/mollifier-method record stays PRZZ 2020, κ > 0.417293** [known] — say
"record for the mollifier method", not "record". Never present any of this as progress on RH (the papers say so
themselves).

## Standard references / expositions

- E. C. Titchmarsh, *The Theory of the Riemann Zeta-Function*, 2nd ed. rev. D. R. Heath-Brown, Oxford 1986.
  **Chapter X "The zeros on the critical line"** (p. 254): Riemann's memoir; infinity of zeros on σ=1/2;
  N0(T); Hardy–Littlewood bound; Selberg's theorem (§10.9 — states N0(T) > A T log T); functions for which RH
  fails; Levinson's method; simple zeros; zeros of derivatives. Checked from the book's table of contents
  (GBV TOC scan https://www.gbv.de/dms/ilmenau/toc/214006875.PDF).
  **Block 3: §§10.9–10.22 (pp. 268–279) read** in a full scan hosted for a Rutgers course
  (sites.math.rutgers.edu/~zeilberg/EM18/TitchmarshZeta.pdf — cite the book, not that copy). Gives Selberg's theorem
  (Thm 10.22, N0(T) > A T log T) with all lemmas: mollifier = square of a smoothed partial sum of ζ^{−1/2}, X = T^c,
  h = (ac log T)^{−1}; moment bounds Lemmas 10.17–10.21 (copied into mechanism-notes §10 step 3). §10.9 footnote: Selberg's
  constant "was calculated in an Oxford dissertation by S. H. Min" (not seen). archive.org has only the 1930 Cambridge tract
  "The Zeta-Function of Riemann" (ids zetafunctionofri032076mbp, in.ernet.dli.2015.203927 — predates Selberg, not opened);
  no 2nd-edition copy on archive.org (a search-engine claim of id "theoryriemannzet00titc" is false — that id has no metadata).
- H. Iwaniec, *Lectures on the Riemann Zeta Function*, AMS University Lecture Series 62, 2014 (Crossref date
  2014-10-07), https://doi.org/10.1090/ulect/062 . Part 1 ch. 15 "Zeros on the critical line" (Hardy–Littlewood
  via I(t)=∫Z vs J(t)=∫|Z| on windows of length Δ); Part 2 "The critical zeros after Levinson" ch. 16–29:
  detecting critical zeros (17), Conrey's construction (18), argument variations (19), attaching a mollifier
  (20), Littlewood lemma (21), principal inequality (22), positive proportion (23), moments of Dirichlet
  polynomials (24–28), computations & optimal mollifier (29). Ch. 16 states Levinson κ = 0.3420, Conrey 0.4088,
  "best so far" Feng 0.4128 (written 2012). Read from a PDF copy; **cite the DOI, not that copy** (it is a
  personal-licence copy hosted on a faculty site).
- M. P. Young, "A short proof of Levinson's theorem", Arch. Math. 95 (2010) 539–548, https://arxiv.org/abs/1002.4403 .
  Gives the clean formula used in our computation: κ ≥ 1 − (1/R) log c(P,Q,R,θ) with
  c = 1 + (1/θ)∫₀¹∫₀¹ e^{2Rv} ( d/dx[e^{Rθx}P(x+u)Q(v+θx)]_{x=0} )² du dv; with P=x, Q=1−x, R=1.3, θ=.5:
  c = 2.35…, κ ≥ 0.34… (proved there for θ < 1/2).
- J. B. Conrey, "The Riemann Hypothesis", Notices AMS 50 (2003) 341–353 — good plain overview (not re-read this block).
- H. L. Montgomery & R. C. Vaughan, *Multiplicative Number Theory I*, CUP 2007, §14.2 + notes — Hardy's theorem
  proof via Z(t), clean history paragraph (read: notes to §14.2 list Hardy 1914, HL 1917/1921, Selberg 1942a/b,
  Levinson 1974, Heath-Brown 1979, Conrey 1989).
- S. A. C. Baluyot, "On the zeros of Riemann's zeta-function on the critical line" (Atkinson-method proof of
  κ>0, explicit κ > 0.0001049 / 0.0086729 from that method), https://sieg-baluyot.github.io/criticalline.pdf —
  useful because it gives an *explicit small number* for a Selberg-type (sign-change) method.
- J. B. Conrey, H. Iwaniec, K. Soundararajan, "Critical zeros of Dirichlet L-functions", J. reine angew. Math.
  (2013), https://arxiv.org/abs/1105.1177 — family version with mollifier length θ = 1 via the asymptotic large
  sieve: proves ≥ 56% (0.56001); remarks optimal mollifier gives κ′(1,1.06,0.75) = 0.5865 at θ=1 and
  κ′(1/2,0.96,1.24) = 0.356 at θ=1/2. **These two numbers match our independent computation (see compute/).**

## Degree-2 L-functions (block 4, Nov 12 ~3:50 AM PT — for primer §10.1)

- J. L. Hafner, "Zeros on the critical line for Dirichlet series attached to certain cusp forms", Math. Ann. 264 (1983)
  21–37, https://doi.org/10.1007/BF01458048 (Springer record seen) — N₀(T) > A T log T for even-weight Hecke eigenforms on
  SL₂(ℤ), Selberg's method with a mollifier approximating L^{−1/2}. Statement read in the announcement "On the zeros of
  Dirichlet series associated with certain cusp forms", Bull. AMS (N.S.) 8 (1983) no. 2, from p. 340,
  https://doi.org/10.1090/S0273-0979-1983-15111-X (end page not recorded). Maass forms, level 1: Hafner, J. reine angew.
  Math. 377 (1987) 127–158 (as cited by Andersen–Thorner [13]; not opened).
- N. Andersen, J. Thorner, "Zeros of GL₂ L-functions on the critical line", arXiv:2004.03581 (HTML read block 4; page dated
  Aug 24 2026; journal-ref not checked). **Thm 1.1:** for any holomorphic or Hecke–Maaß newform f on Γ₀(N_f), N_f ≥ 1
  (trivial nebentypus, per §1), κ_f > 0.0696; under GRC κ_f > 0.0896. Levinson's method after Bernard + Blomer–Harcos;
  mollifier range (2.12) 0 < ν < ¼ − θ/2, θ ≤ 7/64 (Kim–Sarnak); (2.11) κ_f ≥ 1 − inf log c(P,Q,R,ν/2)/R. Cites Bernard,
  Acta Arith. 167 (2015) 201–237 (κ_f > 0.0297 holomorphic squarefree level) and Kühn–Robles–Zeindler, Math. Z. 291 (2019)
  (0.02976) — both not opened.
- Not used: Rezvyakova (arXiv 1212.2948; Izv. 2016 doi 10.1070/IM8463) — Selberg-type positive proportion for cusp forms
  on Γ₀(D) of any weight k ≥ 1 with character (search snippet only, not opened; would cover weight-1 forms).

## Early numerical verification and Speiser (block 2 — verified or removed)

Verified (keep, with these citations):
- **Gram 1903**, "Note sur les zéros de la fonction ζ(s) de Riemann", Acta Math. 27, 289–304,
  https://doi.org/10.1007/BF02421310 (Crossref) — first 15 zeros (count from Moree–Petrykiewicz–Sedunova Table 4).
- **Backlund 1914**, "Sur les zéros de la fonction ζ(s) de Riemann", C. R. Acad. Sci. Paris 158, 1979–1981 (séance du
  29 juin 1914; first page seen on Wikisource djvu/1983) — N(200) = 79, all on the line (MPS §3); Backlund's 1912
  N(T) = θ(T)/π + 1 + S(T) method.
- **Hutchinson 1925**, "On the roots of the Riemann zeta function", Trans. AMS 27, 49–60,
  https://doi.org/10.1090/S0002-9947-1925-1501297-5 (Crossref) — RH to height 300 (138 zeros), Gram-point method; found
  two empty Gram intervals (MPS §2).
- **Titchmarsh 1935**, "The zeros of the Riemann zeta-function", Proc. R. Soc. A 151, 234–255,
  https://doi.org/10.1098/rspa.1935.0146 — Riemann–Siegel with explicit remainder, all zeros up to t = 390.
- **Titchmarsh 1936**, same title, Proc. R. Soc. A 157, 261–263, https://doi.org/10.1098/rspa.1936.0192 — abstract
  (Crossref, quoted): "these calculations have now been carried as far as t = 1468. The number of zeros up to this
  point is 1041, and they all lie on the line σ = ½. I have to thank Dr. L. J. Comrie for planning and supervising the
  calculations…". So "Titchmarsh–Comrie 1936" is fair as *Titchmarsh 1936, computations planned by Comrie*.
- Secondary used: P. Moree, I. Petrykiewicz, A. Sedunova, "A computational history of prime numbers and Riemann zeros",
  arXiv:1810.05244 (Oct 2018), §§1–5 + Table 4 (1903 Gram 15; 1914 Backlund 79; 1925 Hutchinson 138; 1935 Titchmarsh
  1041 — their table lumps the 1935/1936 papers). Modern: first ~3·10^12 zeros verified (Platt–Trudgian 2021, as cited
  in Knausgård §1 [32]) — block 1's "~12 trillion" should be re-checked before use: Knausgård says 0<γ≤3·10^12 (a height,
  not a count; ≈ 1.2·10^13 zeros), consistent.
- **Speiser**, "Geometrisches zur Riemannschen Zetafunktion", Math. Ann. 110, 514–521, https://doi.org/10.1007/BF01448042
  (Crossref issue date Dec 1935; often cited as 1934, e.g. Farr–Pauli). Statement (Farr–Pauli, "More zeros of the
  derivatives of the Riemann zeta function on the left half plane", Thm 1): **RH ⇔ ζ′(s) has no zeros in 0 < σ < 1/2**.
  Levinson–Montgomery, "Zeros of the derivatives of the Riemann zeta-function", Acta Math. 133 (1974) 49–65,
  https://doi.org/10.1007/BF02392141, give "a simpler and more instructive proof" (Farr–Pauli). Primary not opened.

Removed: nothing further; all five early-verification items are now sourced. Not verified: Riemann's own computations
beyond the 14.13… zero (Iwaniec ch. 15 says Riemann found γ1 = 14.13… "with bare hands").

## Not used / flagged
- arXiv 2511.06109 (student report on Levinson & Wu): secondary, not relied on.
