---

Subject and independence

Role: independent reviewer in a fresh context, commissioned for refutation only and given nothing but the assignment. The reviewer took no part in writing the page or any page of its folder and had read none of them before this review. Standing, status and acceptance text were outside the commission.

Subject: path wiki/research/erdos_501/ch_counterexample_reconstruction.md as it stood at 2026-09-28T05:03:27Z, read whole as of that time.

Artifacts, read from the folder-name PDFs held beside the two cards as of the same time:

• Glazer, draft rev10, held by Glazer (2026), 8 physical pages numbered 1--8. Physical p. 8 (Section 6: the proof of Corollary 1.2 and the written-out counterexample) read clause by clause in the text layer and on a rendered page image; physical p. 1 (abstract, the definitions of $\mathrm{Free}_\omega$ and $P$, Theorem 1.1, Corollary 1.2) read the same way; pp. 2--7 read once in the text layer only to confirm that no other definition of $\prec$, $P$ or independence intervenes (the draft uses $\prec$ once, on p. 8, and never defines it). Page images rendered: pp. 1, 2 and 8; pp. 1 and 8 were viewed.
• Lee, second version, date line "June 1, 2026", held by Lee (2026), 6 physical pages numbered 1--6. Physical pp. 5--6 (Appendix A) read clause by clause in the text layer and on rendered page images; physical p. 1 (the definition of independence and $P$, Theorem 1.1, Corollary 1.2) read the same way; pp. 2--4 read once in the text layer only. Page images rendered: pp. 1, 2, 5 and 6; pp. 1, 5 and 6 were viewed. The retained first version was not read.

Allowed material actually read: the page; the two library cards (see the exposure below); the Statement paragraph of the problem page Problem 501; docs/verification.md "Whole-claim report" and "Audit checklist" (the Erdos-specific subsections and the shared canonical-failure-modes section); docs/evidence.md "Source fidelity"; docs/math_authoring.md. The two theorem pages that the page names as consumers were not read; only the presence of a wikilink to this page in each was counted. No web search and no evidence folder.

Exposure: the two library cards were printed whole rather than to their provenance paragraphs, so their "Bears on", "Read status", "Overview", formalization and "Relation to E501" sections, which carry standing and acceptance text, reached the reviewer; the problem page's frontmatter desc, which summarizes the page-level label, was printed with the header lines. None of that text was used: every verdict below rests on the page, the two PDFs and the problem page's Statement paragraph. The working-tree status listing showed file names of untracked material in other research folders; none was opened.

Restatement

Convention. $\lambda^*$ is Lebesgue outer measure on $\mathbb R$; a set is bounded when it lies in some bounded interval; a set $X\subseteq\mathbb R$ is independent for a family $(A_y)_{y\in\mathbb R}$ when $x\notin A_y$ for every ordered pair of distinct $x,y\in X$, so both $x\notin A_y$ and $y\notin A_x$ are required of every unordered pair; CH is $2^{\aleph_0}=\aleph_1$; and $P$ is the assertion that every family $(A_y)_{y\in\mathbb R}$ of bounded sets with $\lambda^*(A_y)<1$ for every $y$ has an infinite independent set, the positive answer to the first question of Problem 501.

Claim. Assume CH. Then there is one family $(A_y)_{y\in\mathbb R}$, indexed by all reals, such that for every $y$ the set $A_y$ is countable, hence $\lambda^*(A_y)=0<1$, and $A_y\subseteq[-(|y|+1),|y|+1]$, and such that no infinite $X\subseteq\mathbb R$ is independent for it. Consequently CH refutes $P$. Nothing is claimed without CH, and nothing is claimed about the second question of the problem.

Checklist

• Quantifiers and scope: one boundary slip. The proof asserts $|x_n|<|x_0|-n$ "for every $n$", which is false at $n=0$; the chain proves it for $n\ge1$, and the contradiction uses only $n>|x_0|\ge0$ (F1). Every other quantifier matches the sources: every real $y$ gets a set; every infinite $X$ is excluded; independence ranges over all distinct pairs.
• Circularity: none. The family is built first; the nonexistence of an infinite independent set is derived by contradiction from an arbitrary such set, and $\neg P$ is not assumed anywhere.
• Model and convention changes: none. The objects are the sources' own (subsets of $\mathbb R$, Lebesgue outer measure, boundedness, the well-ordering induced by an enumeration of type $\omega_1$), and the independence convention is the sources' and the problem page's. The words "without repetition" make explicit an injectivity the sources use tacitly (F2).
• Finite and statistical overreach: inapplicable; no finite check or heuristic appears.
• Uniformity: the only constant is the $+1$ in the definition of $A_y$ and in the inequality $|x_i|>|x_j|+1$, both exactly the sources'; the iterated bound $|x_n|<|x_0|-n$ has no hidden dependence beyond $n$ and $x_0$.
• Extremal conclusions: inapplicable; no infimum, supremum or sharpness is claimed. The one existence used, the $\prec$-first $\omega$ elements of an infinite $X$, is re-derived under Weakest steps.
• Consequences and composition: "Hence CH implies $\neg P$" checked separately: the family is bounded with outer measure $0<1$ and has no infinite independent set, so the universal statement $P$ fails. The Boundary paragraph's stronger consequence (the variant with "null" in place of "outer measure below one" also fails) follows from the same family. The two consumer pages do link to this page; their use of it was not examined.
• Computation: inapplicable; the page runs no computation.
• Reproduction: inapplicable; the page states no rerun command or coverage claim.
• Source and verdict fidelity: the statement, the definition of $A_y$ and the argument match Glazer p. 8 and Lee pp. 5--6 clause by clause; the locators (Section 6, physical p. 8; Appendix A, physical pp. 5--6; the second version's date line) are correct; the standing sentence claims only author-recorded standing. Two small characterizations are noted: the desc attributes the construction, rather than the result, to Hechler (F3), and the Hechler citation drops the series name (F4).

Weakest steps

W1, the $\omega$-sequence. Suppose $X\subseteq\mathbb R$ is infinite. Since $\prec$ is a strict well-ordering of $\mathbb R$ (the enumeration is a bijection $\omega_1\to\mathbb R$, and $\prec$ is the pullback of $<$ on $\omega_1$), the restriction of $\prec$ to $X$ well-orders an infinite set, so its order type is an ordinal $\gamma\ge\omega$, and for each $n<\omega$ the element $x_n$ of $X$ of rank $n$ exists; these satisfy $x_0\prec x_1\prec\cdots$. Lee p. 5 says "choose a strictly increasing sequence"; Glazer p. 8 supposes one given. The page's parenthetical justification is correct and composes with W2 by supplying, for every $i<j$, distinct points $x_i\prec x_j$ of $X$.

W2, the inequality. Write $x_i=r_\alpha$ and $x_j=r_\beta$ with $i<j$, so $\alpha<\beta$. By definition, $A_{x_j}=\{r_\gamma:\gamma<\beta,\ |r_\gamma|\le|x_j|+1\}$. If $|r_\alpha|\le|x_j|+1$ held, then $r_\alpha$ would satisfy both membership conditions with $\gamma=\alpha$, so $x_i\in A_{x_j}$; this direction needs no injectivity. Independence for the ordered pair $(x_i,x_j)$ of distinct points gives $x_i\notin A_{x_j}$, hence $|x_i|>|x_j|+1$. The other half of independence, $x_j\notin A_{x_i}$, is automatic, since every member of $A_{x_i}$ is $\prec$-below $x_i$ and $x_i\prec x_j$; the argument therefore uses independence only in the increasing direction, which is all the sources use. Injectivity of the enumeration is used earlier, to make $y=r_\beta$ determine $\beta$ (so that $A_y$ is well defined) and to make $\prec$ antisymmetric.

W3, the iteration. From W2 at $(i,i+1)$: $|x_i|>|x_{i+1}|+1$ for every $i<\omega$. Adding $i$ to both sides, $|x_i|+i>|x_{i+1}|+(i+1)$, so the sequence $|x_n|+n$ is strictly decreasing, and for $n\ge1$, $|x_n|+n<|x_0|$, that is, $|x_n|<|x_0|-n$. At $n=0$ this reads $|x_0|<|x_0|$ and is false; the correct universal statement is $|x_n|\le|x_0|-n$ for every $n$, strict for $n\ge1$. Choose an integer $n>|x_0|$ (Archimedean property); then $n\ge1$ and $|x_n|<|x_0|-n<0$, contradicting $|x_n|\ge0$. So no infinite independent set exists. The argument's reliance on $n\ge1$ is what confines F1 to the sentence and not the conclusion.

Strongest attack

The attack sought an infinite independent set that escapes the argument. Independence is symmetric, but $A_y$ only ever contains points $\prec$-below $y$, so the constraint on a pair is one-sided: for $x\prec y$ in $X$, the only condition is $x\notin A_y$, that is, $|x|>|y|+1$. An infinite $X$ would therefore need $|\cdot|$ to drop by more than $1$ along every $\prec$-increasing pair. The attack tried to make $X$ avoid long $\prec$-increasing chains, or to place its $\prec$-first elements so that the drops do not accumulate: both fail, because every subset of $\mathbb R$ is well-ordered by $\prec$, so an infinite $X$ has $\prec$-first elements $x_0\prec x_1\prec\cdots$, and along them $|x_n|+n$ strictly decreases from $|x_0|$, which $[0,\infty)$ cannot sustain past $n=\lfloor|x_0|\rfloor+1$. A second attack asked whether the page uses more than CH: it uses a bijective enumeration of $\mathbb R$ in type $\omega_1$ (CH with choice), the countability of every $\beta<\omega_1$, the nullness of countable sets and the Archimedean property, nothing else. A third checked whether some real lacks a set (no: the enumeration is onto) or some $A_y$ escapes $[-(|y|+1),|y|+1]$ (no: membership requires $|r_\alpha|\le|y|+1$). The only successful attack is on a sentence, not the argument: the universal "for every $n$" in the iteration step is false at $n=0$, with witness $|x_0|<|x_0|$; the step that reaches the contradiction uses only $n>|x_0|$, so the reconstruction survives with F1 corrected.

Premises

• CH, as $2^{\aleph_0}=\aleph_1$, used once: with choice it gives a bijection $\omega_1\to\mathbb R$, hence the enumeration $\{r_\alpha:\alpha<\omega_1\}$ without repetition and the induced strict well-ordering $\prec$. Explicit hypothesis of the statement.
• ZFC facts, unlabeled on the page and elementary: every ordinal $\beta<\omega_1$ is countable, so $\{r_\alpha:\alpha<\beta\}$ is countable; a countable subset of $\mathbb R$ is Lebesgue null; an infinite subset of a well-ordered set has an initial segment of order type $\omega$; for every real $t$ there is an integer $n>t$.
• The definition of independence and of $P$: Glazer p. 1 (abstract and the display defining $\mathrm{Free}_\omega$) and Lee p. 1 (Section 1), both held and read clause by clause; the problem page's Statement paragraph states the same first question. Interface used: $P$ is the universal statement over all families of bounded sets of outer measure below one; one family with no infinite independent set refutes it.
• No theorem is imported. The two held write-ups are the sources of the construction and both are read at the cited pages; Hechler's 1972 note, to which both attribute the result, is not held, and the page says so and proves the claim directly rather than citing it. No local claim page is consumed.
• Explicit assumptions on the page beyond CH: none. The page assumes nothing about the second question of the problem or about the theorem pages that consume it.

Findings

F1. Severity: required. Location: "so $|x_n|<|x_0|-n$ for every $n$". The universal statement is false at $n=0$, where it reads $|x_0|<|x_0|$; the chain $|x_0|>|x_1|+1>|x_2|+2>\cdots$ proves the strict inequality only for $n\ge1$. Witness: Lee p. 6 writes "Iterating, we get $|x_n|<|x_0|-n$, which is impossible for $n>|x_0|$" with no universal quantifier, and Glazer p. 8 does not spell out the step, so the quantifier is the page's own. The argument is unaffected because the next sentence uses only $n>|x_0|\ge0$. Proposed replacement: "so $|x_n|<|x_0|-n$ for every $n\ge1$. For an integer $n>|x_0|$ this gives $|x_n|<0$, which is impossible."

F2. Severity: suggested. Location: "without repetition" and "(its first $\omega$ elements in the order $\prec$)". Both are supplied by the page and are not marked as such: Lee p. 5 says "fix an enumeration" and "Let $\preceq$ be the induced well-ordering", leaving injectivity tacit, and says "Choose a strictly increasing sequence" without justification; Glazer p. 8 says "Enumerate" and supposes the sequence given, and never defines $\prec$. Both supplements are correct (W1 above). Proposed replacement: mark them, for example "without repetition (the sources' enumeration is tacitly injective; stated here so that $\prec$ is a well-ordering)" and "(supplied: its first $\omega$ elements in the order $\prec$, which exist because an infinite well-ordered set has order type at least $\omega$)".

F3. Severity: note. Location: the desc, "the construction, written out in both 2026 notes and attributed by them to Hechler". The sources attribute the result to Hechler ("Hechler proved the negative answer from CH", Glazer p. 1; "Hechler [4] proved $\neg P$ under the assumption CH", Lee p. 1; both abstracts say "Hechler's counterexample") and then write out a family "for completeness" (Glazer p. 8, Lee p. 5) without saying that this family is Hechler's. The body's Source paragraph, "Both attribute the result to S. H. Hechler", is exact. Proposed replacement for the desc: "of a family of countable bounded sets under CH with no infinite independent set, written out in both 2026 notes for the result they attribute to Hechler."

F4. Severity: note. Location: "Bull. Acad. Polon. Sci. 20 (1972), 429--431". Both reference lists give the series: Glazer [3], p. 8, and Lee [4], p. 6, read "Bull. Acad. Polon. Sci. Sér. Sci. Math. Astronom. Phys. 20 (1972), 429--431". The volume, year and pages identify the note, so nothing is misdirected. Proposed replacement: "Bull. Acad. Polon. Sci. Sér. Sci. Math. Astronom. Phys. 20 (1972), 429--431".

Verdict

Source fidelity: faithful with corrections. The statement, the family and the argument are those of Glazer Section 6 (physical p. 8) and Lee Appendix A (physical pp. 5--6), with correct locators and the sources' conventions; the one required correction (F1) is to a quantifier the page added in the proof, and F2--F4 are marking and citation refinements.

The argument as reconstructed: sound. Each deduction was re-derived above; the false instance of the iteration sentence at $n=0$ is not used by the step that reaches the contradiction.

Limitations: the review covers only this page against the two held write-ups. Hechler's 1972 note is not held, so whether the written-out family is his is not decided here. The two pointers into the problem page, that it records the open attribution question and that it records the same construction along a well-ordering of order type $\mathfrak c$ under Martin's axiom, lie outside the commissioned read set and were not checked; the mathematics of the second pointer is standard (under Martin's axiom every set of fewer than $\mathfrak c$ reals is null) but was not audited against the problem page. The consumer pages were not examined beyond the presence of their links.

This focused review assigns no tier and changes no status.