---
id: "20260827-0210-verify-nicelys-exact-pentium"
title: "Verify Nicely's exact Pentium FDIV report-to-Intel and public-disclosure dates against a primary"
type: "capture"
status: "promoted"
origin: "batch"
promoted_to: ["30-notes/claim-nicely-twin-prime-project-surfaced-pentium-fdiv-bug.md (corrected in place: unverified-quant flag discharged with the primary, a second unflagged inaccuracy about Usenet/CompuServe propagation fixed, status seedling → budding)","30-notes/claim-nicely-2011-faq-corroborates-1994-dates-but-is-a-memory-reconstruction.md (new)","40-entities/entity-thomas-nicely.md (new hub)","40-entities/entity-intel-corporation.md (updated in place, additive Updates line)","50-questions/question-verify-nicely-fdiv-report-dates-primary.md (ruled status: answered with a substantive answered_log, discharging a mis-scoped mechanical closure)"]
not_promoted: ["Richard M. Smith's 1 November 1994 CompuServe repost as a standalone claim-note — a single secondhand (2011-recollection) date detail, not atomic enough alone; folded into the corrected twin-prime note as clarifying context instead.","Alexander Wolfe's 7 November 1994 Electronic Engineering Times article ('fully public' milestone) — a further lead per the capture's own text, not yet read; no primary fetched this session.","Steve Young's 21 November 1994 CNN report (first mainstream-media coverage) — further lead, not yet read.","The 'Tom Kraljevic' attribution for Intel's internal discoverer — explicitly unconfirmed in both Nicely primaries read here; correctly left unpromoted per the capture's own flag, not asserted in any new or corrected note.","Entity hub for Richard M. Smith — real, named, one concrete act, but a single mention with no claim-note of his own; declined per the person test's 'say why in one sentence' bar being thin, not absent. Bias against the flood.","Entity hub for Alexander Wolfe — lead only, source unread; declined.","Entity hub for A. B. Copsey (first independent bug confirmation, 28 October 1994) — single mention, source unread; declined.","Entity stub for the unnamed Intel internal P6 FPU test engineer(s) — no confirmed name (Kraljevic explicitly unconfirmed); UNSURE → don't promote."]
writer_model: "claude-sonnet-5"
date_created: "2026-08-27T00:00:00.000Z"
provenance: "2026-08-27 batch verification run, direct answer to 50-questions/question-verify-nicely-fdiv-report-dates-primary.md"
derived_from: []
tags: ["pentium-fdiv-bug","thomas-nicely","verification","history-of-computing","primary-source","dates"]
source_url: "https://web.archive.org/web/20251218063712if_/https://faculty.lynchburg.edu/~nicely/pentbug/bugmail1.html"
source_sha: "f0708960e4f9e8a4c5e33450e558be79e1dd2ee907e101124712e4627e0d65f9"
source_author: "Thomas R. Nicely"
source_date: "1994-10-30T00:00:00.000Z"
source_title: "Original Pentium FDIV flaw e-mail"
source_venue: "Thomas R. Nicely's personal faculty page (faculty.lynchburg.edu), reproducing his own 30 October 1994 e-mail; original URL now 404, read via Wayback Machine capture of 18 December 2025"
source_quote: "I contacted Intel Tech Support regarding this bug on Monday 24 October (call reference number 51270)."
source_tier: 1
source_delight: "The document is Nicely's own 1994 e-mail reproduced verbatim on his own page — the actual dispatch that started the public Pentium FDIV story, not a retelling of it, complete with his 1994 phone and fax numbers and an Intel call-reference number."
seek_code_commit: "7d6d9ed"
---


Answers [[question-verify-nicely-fdiv-report-dates-primary]] (50-questions/question-verify-nicely-fdiv-report-dates-primary.md), which flagged that [[claim-nicely-twin-prime-project-surfaced-pentium-fdiv-bug]] carried the dates "24 October 1994" (report to Intel) and "30 October 1994" (public disclosure) with no primary source — a 2026-08-26 direct re-read of Vaughan Pratt's "Anatomy of the Pentium Bug" had found only "the end of October" in that text, not exact dates.

**Sourcing note on venue.** Thomas R. Nicely's own page at `faculty.lynchburg.edu/~nicely/` (and its `/pentbug/` subdirectory) now 404s on every live route tried (`archive_page`, `WebFetch`, and the plain URL) — the University of Lynchburg appears to have removed or reorganized his faculty pages. The Wayback Machine's `available` API confirmed a 18 December 2025 capture at 200 status; that capture (fetched via the `if_` raw-content suffix, which succeeded where the plain Wayback URL timed out) is the read of record for both documents used here. Per the found-spec rule on archived captures, the original venue (Nicely's own site) is cited, with the fact that only the archived copy survives stated explicitly. No recognition signals fired on either page — both are plain technical prose, dated and signed, with no addressed-to-AI language, override language, or claimed authority; no safety flags for this session.

## Claim: Nicely's original 30 October 1994 e-mail — reproduced verbatim on his own site — gives 24 October 1994 as the date he reported the bug to Intel, with a call reference number

The e-mail Nicely dispatched to announce the bug, preserved on his own page as `bugmail1.html`, states directly: "I contacted Intel Tech Support regarding this bug on Monday 24 October (call reference number 51270). The contact person later reported that the bug was observed on a 66-MHz system at Intel, but had no further information or explanation, other than the fact that no such bug had been previously reported or observed." This is a quantitative/date claim, and the source is as strong as this kind of claim gets: not a later recollection but the actual document Nicely sent on the day of disclosure itself, carrying a specific call-reference number rather than a vague "I called Intel" — Tier 1, clears the floor outright. It also confirms 19 October 1994 as the date Nicely felt he had eliminated other explanations for the discrepancy ("it was not until 19 October 1994 that I felt I had eliminated all other likely sources of error"), the point at which he concluded the fault was in the Pentium itself.

## Claim: the same 1994 e-mail is itself dated 30 October 1994, fixing the public-disclosure date directly rather than through later report

The document's own header reads "DATE: 30 October 1994," addressed "TO: Whom it may concern" and signed with Nicely's Lynchburg College contact information. This is the actual dispatch that [[claim-nicely-twin-prime-project-surfaced-pentium-fdiv-bug]] describes as being "circulated... via Usenet and CompuServe" — the e-mail itself does not mention Usenet or CompuServe (those were downstream relays: per Nicely's later FAQ, Richard M. Smith of Phar Lap Software reposted the inquiry to a CompuServe forum on 1 November, a day after Nicely's original dispatch). The 30 October 1994 date is the date of Nicely's own e-mail, which is the correct anchor for "public disclosure" — Tier 1, the document is dated at the moment of its own creation.

## Claim: Nicely's 2011 retrospective FAQ corroborates both dates and adds detail, but is explicitly a memory-based reconstruction for anything beyond the two dates already fixed by the 1994 e-mail

A companion page on the same site, an FAQ dated "0900 GMT 19 August 2011" (source_sha `bcf7948942bbf12f042c85ed164451e986092ac57c62be4b12a275574a38b50c`, same Wayback capture, `pentbug.html`), gives a detailed day-by-day chronology including both dates: "24 October — I call Intel tech support and inform them of the problem," and "30 October — With no substantive response to this point from either Micron tech support or Intel tech support, I dispatch my initial e-mail inquiry... to several individuals and groups (at approximately 3:20:49 pm EST)." Both dates match the contemporaneous 1994 e-mail exactly. However, Nicely explicitly frames this table as reconstructed memory, not a contemporaneous log: "Following below is the chronology as best I can now reconstruct it." The two headline dates are independently anchored by the 1994 document itself and do not depend on this caveat, but any additional granularity found only in the 2011 account and not in the 1994 e-mail — the exact dispatch time (3:20:49 pm EST), the claim that an Intel program manager later said Intel had internally found a similar flaw in "May or June of 1994" during P6-related FPU testing, the day-by-day narrative of intermediate steps (17–22 October testing on additional Pentiums) — rests on a 17-years-later recollection and should carry that caveat if promoted.

## Further leads

- Nicely's FAQ (2011) states an Intel program manager told him Intel had already found "a similar FPU intended for the P6" flawed internally "in May or June of 1994" — a widely repeated secondary detail names the Intel discoverer as "Tom Kraljevic, a Purdue University co-op student," but that name does not appear in either Nicely primary read here; needs its own primary before use.
- Richard M. Smith (Phar Lap Software), "An account of the spread of the Pentium flaw announcement across the Internet during the first few days" (27 December 1994), linked from Nicely's bibliography as `rsmith.html` on the same site — not yet fetched; would independently corroborate the November propagation chronology.
- Alexander Wolfe, Electronic Engineering Times, 7 November 1994 — per Nicely's own account, "the matter is now fully public" only with this article; a distinct claim from the 30 October e-mail (private/semi-private disclosure to selected individuals) versus mainstream trade-press publication a week later. Worth its own claim-note distinguishing "Nicely's disclosure" from "when it became fully public."
- Steve Young (CNN), 21 November 1994, per Nicely's account "the first mainstream media journalist to break the story" for a general audience — a third, later disclosure milestone distinct from both the 30 October e-mail and the 7 November trade-press article.
- Single-source concentration cap: this capture is the first claim-bearing use of Nicely's own site (`faculty.lynchburg.edu/~nicely/pentbug/`) as a primary in this vault; 0 of 3 slots used before independent corroboration would be required for a fourth.

## Entity candidates

- Unnamed Intel internal test engineer(s), P6 FPU test team, May–June 1994 — concept/persons — per Nicely's own account, Intel found a similar flaw internally *before* his October discovery; this is the earlier, foundational fact his priority-of-public-disclosure claim is implicitly measured against ("Nicely surfaced the bug publicly" is a claim about who told the world, not about who found it first). A secondary tradition names this person "Tom Kraljevic," unconfirmed in either primary read here — flagging the gap, not the name.
- Thomas R. Nicely — person — mathematician, Lynchburg College; no entity page yet exists in this vault for him despite three claim-notes now resting on accounts of his work.
- Richard M. Smith — person — President, Phar Lap Software; reposted Nicely's inquiry to a CompuServe forum on 1 November 1994, the specific act that moved it from a private e-mail list onto a wider public forum.
- Alexander Wolfe — person — Electronic Engineering Times journalist; per Nicely, his 7 November 1994 article is the point the story became "fully public," distinct from Nicely's own 30 October e-mail.
- A. B. Copsey (Babcock & Wilcox Nuclear Technologies) — person — sent Nicely the first e-mail confirmation of the bug on another Pentium system (28 October 1994), the first independent replication in Nicely's own chronology.

> [!note] Seek's commentary:
> The existing vault note hedged correctly — it flagged the dates as unverified rather than inventing false confidence — and the hedge turned out to be exactly right to place: the dates were correct, just uncited. What's more interesting than "the dates check out" is that they check out against the strongest possible version of a primary: not a professor's decades-later memory of what he did, but the actual document he sent that day, complete with an Intel call-reference number a fabricator would have no reason to include. The 2011 FAQ is worth less than the 1994 e-mail even though it's more detailed, because detail and evidentiary weight point in opposite directions once you're relying on memory. — Seek