---

Subject and independence

The reviewer worked in a fresh context from the commissioning assignment alone, took no part in writing the page or any page in its folder, and was charged with refutation. The subject is path wiki/research/erdos_501/glazer_lemma_4_3_reconstruction.md as it stood at 2026-09-28T05:03:27Z, read in full as of that time.

The artifact is the PDF held under the library card Glazer (2026), eight pages whose physical and printed numbers agree. Physical p. 5 was read in full, in the text layer and as page images rendered at 110 and 160 dots per inch; the 160 dpi image was read for every displayed formula of Lemma 4.3 and of Proposition 4.4 (display (4.1), the cardinality display, the intersection display, and (4.2)). Physical pp. 1, 4 and 8 were read in the text layer for the title, the Section 4 preamble that defines supports, and the reference list; the whole text layer was searched for the source's uses of the $\Delta$-system and cofinality notations (the source defines neither).

Allowed material read: the Statement section and the Source paragraph of the Proposition 4.4 reconstruction as of the same time, because the page's Boundary paragraph makes a claim about it; the provenance paragraph of the Glazer card; the "Whole-claim report" and "Audit checklist" sections of docs/verification.md (the Erdos-specific sections and the shared "canonical failure modes" list); the "Source fidelity" section of docs/evidence.md; docs/math_authoring.md in full; and the Statement paragraph of wiki/problems/set_theory/E0501/_index.md. No other reconstruction page was read, since the page cites none as an input.

Exposures: three, all incidental and none bearing on the mathematics checked. The extraction of the card's provenance paragraph also printed the card's "Bears on" and "Read status" paragraphs and the opening lines of its Overview; the "Read status" paragraph records what the card's author did and did not verify. The extraction of the problem page's Statement paragraph also printed its Status and Source paragraphs, because that page uses bold run-in labels rather than headings. The Proposition 4.4 page's Source paragraph was read together with its Statement. No evidence-folder content, no other review, nothing among the private working files, and no web search reached the reviewer; the untracked evidence/verify directory was only listed by name before this report was written into it.

Restatement

Work in ZFC plus CH. Let $\langle S_\alpha:\alpha<\omega_2\rangle$ be any sequence of countable sets; "countable" includes finite and empty, and the sequence may repeat values. Then there exist a set $T'\subseteq\omega_2$ with $|T'|=\omega_2$, an injective map $\xi\mapsto\alpha_\xi$ from $T'$ into $\omega_2$, and one countable set $R$ such that $S_{\alpha_\xi}\cap S_{\alpha_\zeta}=R$ for every pair of distinct $\xi,\zeta\in T'$. No hypothesis beyond CH is used; no large cardinal, no forcing.

Conventions on the page: an indexed family $(A_\xi)_{\xi\in T}$ is a $\Delta$-system with root $R$ when every two distinctly indexed members meet exactly in $R$; $S^{\omega_2}_{\omega_1}$ is the set of ordinals below $\omega_2$ of cofinality $\omega_1$; $[M]^{\aleph_0}$ is the set of all countable subsets of $M$, finite ones included.

Relation to the source's (4.1), "every family of $\omega_2$ countable sets has a $\Delta$-subsystem of size $\omega_2$": a set of $\omega_2$ countable sets, enumerated injectively, is the special case in which the sequence is injective, and then the $\omega_2$ distinct indices $\alpha_\xi$ name $\omega_2$ distinct members. Conversely the sequence form follows from the set form in ZFC: if the sequence takes $\omega_2$ distinct values apply the set form to them and pull the indices back; otherwise it takes at most $\omega_1$ values, so by the regularity of $\omega_2$ one value is taken $\omega_2$ times, and that constant subsequence is a $\Delta$-system whose root is the value. The page's precise statement is therefore equivalent to the source's, not stronger.

Checklist

• Quantifiers and scope: pass. Every quantifier of (4.1) is preserved; the page's precise form adds only the trivially equivalent sequence reading shown above. The boundary cases of an empty or finite root are covered by the page's stated convention for $[M]^{\aleph_0}$ (see F2), and the case $R_\xi=\emptyset$ still yields a regressive value $\eta(\xi)<\xi$ because every $\xi\in S^{\omega_2}_{\omega_1}$ is positive.
• Circularity: pass. Nothing equivalent to the conclusion is assumed; the argument runs from the chain construction, Fodor's theorem, and CH to the root.
• Model and convention changes: pass with a note. The two conventions the page introduces (indexed $\Delta$-systems and all-countable-subsets $[M]^{\aleph_0}$) are stated before use and match what the proof needs; neither substitutes a different object for the source's. F2 asks that the second be marked as a reading of the source's symbol.
• Finite and statistical overreach: inapplicable. No finite case, sample, or heuristic average appears.
• Uniformity: pass. The one uniform object, the stage $\eta$, is supplied by Fodor's theorem on a stationary set, and the bound $|[M_\eta]^{\aleph_0}|=\aleph_1$ depends on nothing but CH.
• Extremal conclusions: inapplicable. The only extremal-flavored claim is the cardinality $|T'|=\omega_2$, rederived under Weakest steps.
• Consequences and composition: one failure outside the proof. Every "so" and "thus" inside the proof was rederived and holds. The Boundary paragraph's consequence sentence, that the supports in Proposition 4.4 are pairwise distinct because each contains its own block, does not follow and is not guaranteed by the source (F1). The page consumes no local claim; Proposition 4.4 is its consumer, not a premise.
• Computation: inapplicable. The page carries no computation or evidence program.
• Reproduction: inapplicable. The page states no rerun command and no coverage claim.
• Source and verdict fidelity: pass for the lemma, fail for the Boundary characterization. The displayed statement is (4.1) word for word, the label "Lemma 4.3 (Generalized $\Delta$-system)" and physical p. 5 are correct, "draft rev10" and "eight-page" agree with the card's provenance, and the Standing paragraph claims only an author-recorded reconstruction. The Boundary paragraph attributes to the source's Proposition 4.4 setup a distinctness the source neither states nor arranges (F1).

Weakest steps

Bounding the root below its stage. Fix $\xi\in S^{\omega_2}_{\omega_1}$. Since $\mathrm{cf}(\xi)=\omega_1$, $\xi$ is a limit ordinal, so by continuity $M_\xi=\bigcup_{\eta<\xi}M_\eta$. The root $R_\xi=A_\xi\cap M_\xi$ is a subset of the countable set $A_\xi$, hence countable. For each $x\in R_\xi$ let $\eta_x<\xi$ be the least stage with $x\in M_{\eta_x}$. The set $\{\eta_x:x\in R_\xi\}$ is a countable subset of $\xi$, and a countable subset of an ordinal of uncountable cofinality is bounded in it; let $\eta(\xi)<\xi$ be a bound (any ordinal below $\xi$ when $R_\xi$ is empty). The chain is increasing, so $x\in M_{\eta_x}\subseteq M_{\eta(\xi)}$ for every $x\in R_\xi$, that is $R_\xi\subseteq M_{\eta(\xi)}$. The map $\xi\mapsto\eta(\xi)$ is thus regressive on $S^{\omega_2}_{\omega_1}$, which is stationary in the regular cardinal $\omega_2$; Fodor's theorem gives a stationary $T\subseteq S^{\omega_2}_{\omega_1}$ and a single $\eta$ with $\eta(\xi)=\eta$ on $T$. A stationary subset of $\omega_2$ is unbounded, and an unbounded subset of a regular cardinal has that cardinal's size, so $|T|=\omega_2$. This composes with the next step by placing every root $R_\xi$, $\xi\in T$, inside the one set $[M_\eta]^{\aleph_0}$.

The pigeonhole under CH. Every countable subset of $M_\eta$ is the range of a function $\omega\to M_\eta$ or is empty, so $|[M_\eta]^{\aleph_0}|\le|M_\eta|^{\aleph_0}=\aleph_1^{\aleph_0}$. In ZFC,

$$2^{\aleph_0}\le\aleph_1^{\aleph_0}\le(2^{\aleph_0})^{\aleph_0} =2^{\aleph_0\cdot\aleph_0}=2^{\aleph_0},$$

and CH makes this $\aleph_1$. The map $\xi\mapsto R_\xi$ sends $T$, of size $\aleph_2$, into a set of size at most $\aleph_1$. If every fiber had size at most $\aleph_1$ the domain would have size at most $\aleph_1\cdot\aleph_1=\aleph_1$, so some fiber $T'$ has size $\aleph_2$, and its common value $R$ is countable as a subset of $A_\xi$. This composes with the root computation by fixing $R$ and $T'$ with $R_\xi=R$ for all $\xi\in T'$.

The root identity. Let $\xi<\zeta$ in $T'$. As $\zeta$ is a limit ordinal, $\xi+1<\zeta$, so $M_{\xi+1}\subseteq M_\zeta$. The ordinal $\alpha_\xi$ lies in $M_{\xi+1}$, so $A_\xi=S_{\alpha_\xi}\in M_{\xi+1}$, being the value at $\alpha_\xi$ of the sequence, which lies in $M_0\subseteq M_{\xi+1}$. A countable member of $M_{\xi+1}$ is a subset of it: $H(\theta)$ contains a surjection $f\colon\omega\to A_\xi$ when $A_\xi$ is nonempty (its transitive closure is countable), so by elementarity $M_{\xi+1}$ contains such an $f$, and for each $n\in\omega\subseteq\omega_1\subseteq M_{\xi+1}$ the value $f(n)$ is the unique $y$ with $(n,y)\in f$, which elementarity places in $M_{\xi+1}$. Hence $A_\xi\subseteq M_{\xi+1}\subseteq M_\zeta$ and

$$A_\xi\cap A_\zeta=A_\xi\cap A_\zeta\cap M_\zeta=A_\xi\cap R_\zeta =A_\xi\cap R=R,$$

the last equality because $R=R_\xi=A_\xi\cap M_\xi\subseteq A_\xi$. The injectivity of $\xi\mapsto\alpha_\xi$ on $T'$ holds because $\alpha_\xi\in M_{\xi+1}\subseteq M_\zeta$ while $\alpha_\zeta\notin M_\zeta$ by choice. Together the three steps give the restated conclusion.

Strongest attack

The strongest attack on the proof aimed at collapsing the $\Delta$-system: either two indices $\xi<\zeta$ of $T'$ with $\alpha_\xi=\alpha_\zeta$, so that the "system" repeats one set, or a pair with $A_\xi\not\subseteq M_\zeta$, which would break the first equality of the root identity. Both fail for the same reason: $\zeta$ has cofinality $\omega_1$, so $\xi+1<\zeta$ and $M_{\xi+1}\subseteq M_\zeta$, whereas $\alpha_\zeta$ was chosen outside $M_\zeta$. A second attack tried to make the chain construction fail at limit stages of cofinality $\omega$, where a union of $\omega$ models of size $\omega_1$ might be feared to grow or to lose elementarity; it has size $\omega_1$ and is elementary by the union-of-chains lemma (a witness to an existential statement with parameters in the union already lies in some link, which is elementary in $H(\theta)$). A third attack tried an empty or finite root, where the source's symbol $[M_\eta]^{\aleph_0}$ under its common reading would not count $R_\xi$; the page's stated convention counts all countable subsets, and the cardinal $\aleph_1^{\aleph_0}$ bounds those as well, so the count survives (F2 asks only that the reading be marked).

The attack that landed is on the Boundary paragraph. The source's proof of Proposition 4.4 (physical p. 5) chooses "a countable support $S_\alpha$ for $\dot w_\alpha$" and enlarges it "to contain $R_0\cup D_\alpha$"; nothing prevents two names from sharing a support. Witness: take $\dot w_\alpha=\dot w_\beta$ for some $\alpha\ne\beta$, both read from a common countable support $S\supseteq R_0\cup D_\alpha\cup D_\beta$; then $S_\alpha=S_\beta=S$ is a legitimate choice, each contains its own block, and the sets are not pairwise distinct. The page's sentence asserts the distinctness as a fact and gives a reason that does not entail it. The lemma's own statement and proof are untouched, because the page's precise form is indexed and tolerates repetition; only the reconciliation offered in the Boundary paragraph is wrong.

Premises

• Downward Löwenheim--Skolem for $H(\theta)$ and the union-of-chains lemma. Interface used: for every $X\subseteq H(\theta)$ with $\omega_1\subseteq X$ and $|X|=\omega_1$ there is $M\prec H(\theta)$ with $X\subseteq M$ and $|M|=\omega_1$; and the union of an increasing chain of elementary submodels of $H(\theta)$ is an elementary submodel of $H(\theta)$. Cited to T. Jech, Set Theory, third millennium edition, Chapter 12. Not held in the library; checked against the reviewer's knowledge of the standard statements, and the union lemma rederived above. Also used, without being named among the imports: $H(\theta)$ is transitive, so membership and "$f$ is a function from $\omega$ onto $A$" are absolute, and the surjection lies in $H(\theta)$ for any uncountable $\theta$ (F3).
• Fodor's theorem. Interface used: a function $f$ on a stationary subset $S$ of a regular uncountable cardinal $\kappa$ with $f(\xi)<\xi$ for all $\xi\in S$ is constant on a stationary subset of $S$. Cited to Jech, Chapter 8. Not held; standard statement.
• Stationarity of $S^{\omega_2}_{\omega_1}$ in $\omega_2$. Cited to Jech, Chapter 8. Not held; standard statement (for regular $\lambda<\kappa$ the ordinals below $\kappa$ of cofinality $\lambda$ form a stationary set).
• The identity $\aleph_1^{\aleph_0}=2^{\aleph_0}$. A ZFC theorem, rederived above in one line; the page names it as imported without a locator (F4).
• CH, in the form $2^{\aleph_0}=\aleph_1$: the lemma's hypothesis.
• The axiom of choice is used to pick $\alpha_\xi$ and the elementary submodels; the page works in ZFC, as the source does.
• No local claim is consumed. The page has no depends_on and cites no L-claim; the Proposition 4.4 page is a consumer.

Findings

F1. Severity: required. Location: Boundary, "there the sets $S_\alpha$ are pairwise distinct because each contains its own block $\{\alpha\}\times\omega$". Defect: the deduction does not follow, and the conclusion is not guaranteed by the source. Containing a private block does not stop two supports from coinciding; a support may contain several blocks. Witness: source physical p. 5, proof of Proposition 4.4, "choose a countable support $S_\alpha$ for $\dot w_\alpha$ and enlarge it to contain $R_0\cup D_\alpha$", with two names $\dot w_\alpha=\dot w_\beta$ read from one support $S\supseteq R_0\cup D_\alpha\cup D_\beta$, giving $S_\alpha=S_\beta$. Proposed replacement text: "The lemma is applied in [Proposition 4.4] to the countable supports of $\omega_2$ names. Those supports need not be pairwise distinct, since two names may share a support; this is why the statement above is given for an indexed sequence and concludes with an injection on indices rather than with $\omega_2$ distinct sets. The sequence form is equivalent to the source's family form: a family of $\omega_2$ sets is the injective case, and a sequence with fewer than $\omega_2$ distinct values repeats one value $\omega_2$ times, a $\Delta$-system with that value as root."

F2. Severity: suggested. Location: Definitions, "$[M]^{\aleph_0}$ is the set of its countable subsets". Defect: the convention departs from the common reading of the symbol (subsets of size exactly $\aleph_0$) and the page does not say that it is a reading of the source's display, nor that the roots $R_\xi=A_\xi\cap M_\xi$ may be finite or empty, which is the case the wider convention exists to cover. Witness: source physical p. 5, display "$|[M_\eta]^{\aleph_0}|=(\aleph_1)^{\aleph_0}=\aleph_1$", and the definition $R_\xi=A_\xi\cap M_\xi$ two lines above it. Proposed replacement text: "$[M]^{\aleph_0}$ is the set of all countable subsets of $M$, finite and empty ones included (the roots $R_\xi$ below may be finite); the source's display is read with this convention, under which its count is unchanged."

F3. Severity: note. Location: Proof, "Let $\theta$ be a regular cardinal" and "Two consequences of the setup are used". Defect: the regularity of $\theta$ is a qualification the source does not state ("a sufficiently large $H(\theta)$"), and the two consequences are proofs the page supplies for facts the source uses without proof; neither is marked as supplied, and the second relies on the transitivity of $H(\theta)$ and on the surjection $\omega\to A$ belonging to $H(\theta)$, which the Standing paragraph does not list. Witness: source physical p. 5, "Take a continuous increasing chain ... of elementary submodels of a sufficiently large $H(\theta)$" and "Since $A_\xi$ is countable and belongs to $M_{\xi+1}$, it is a subset of $M_{\xi+1}$". Proposed replacement text: "Let $\theta$ be an uncountable regular cardinal (regularity is a convenience the source leaves implicit) large enough that the sequence lies in $H(\theta)$" and "Two consequences of the setup, stated without proof in the source, are used; both rest on the transitivity of $H(\theta)$."

F4. Severity: note. Location: Standing, "the cardinal arithmetic $\aleph_1^{\aleph_0}=2^{\aleph_0}$". Defect: the only import without a locator; the identity is a ZFC theorem. Witness: the Standing paragraph itself, against its two other imports, which name chapters. Proposed replacement text: "the ZFC identity $\aleph_1^{\aleph_0}=2^{\aleph_0}$ (from $2^{\aleph_0}\le\aleph_1^{\aleph_0}\le(2^{\aleph_0})^{\aleph_0}$; Jech, Chapter 5)".

Verdict

Source fidelity: faithful with corrections. The statement, the label, the physical page, the revision and the proof outline match the artifact; the one required correction, F1, concerns the Boundary paragraph's characterization of the source's Proposition 4.4 setup, not the lemma.

The argument as reconstructed: sound. Every step of the proof was rederived above; no hypothesis is used that the page does not make available, and no imported theorem is applied outside its hypotheses.

Limitations: the imported results are cited to a textbook the library does not hold, so their interfaces were checked against the reviewer's knowledge of the standard statements rather than against a held copy; the Proposition 4.4 page was read only at its Source and Statement sections, so F1 is a finding about this page's sentence and the source's text, not a review of that page. No computation was involved.

This focused review assigns no tier and changes no status.