Skip to content

Erdős Problem #363

Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: disproved (not automatically the live result).
Subject areas: number theory.

Read this status correctly

The status above is reproduced from GCL's immutable external-catalogue snapshot. It is not a new claim of current mathematical resolution, a source-semantic review, an independent proof check, or GCL certification. A published proof and a GCL replay are separate events.

The question in plain language

Erdős Problem 363 asks whether only finitely many finite collections of pairwise disjoint intervals of integers can have a square as the product of all their entries, when every interval contains at least four integers. For positive starting points, write a block of length \(k\) as \(f(n,k)=n(n+1)\cdots(n+k-1)\). The separated version orders the blocks so that \(n_i+k_i\leq n_{i+1}\); this prevents overlap while allowing the lengths \(k_i\) to vary.

The answer to the broad finiteness question is no. The point is not that every choice of the number of blocks or their lengths has infinitely many examples. Rather, one explicit fixed-length, fixed-number family is enough to refute universal finiteness. The original issue packet is MATH-PROGRAMME issue #1633, and the maintained Erdős Problems entry records the same resolution.

A small example

The construction of Bauer and Bennett uses the Pell-type equation \(u^2-3v^2=-2\). The recurrence they give begins \(u=1,5,19,71,265\); the corresponding value at \(u=265\) is \(v=153\). Their first family then gives starting points

\[ (x_1,x_2,x_3)=((u-5)/4,(v-3)/2,(u+1)/2)=(65,75,133). \]

Thus the three blocks are \([65,68]\), \([75,78]\), and \([133,136]\). Direct integer arithmetic gives

\[ (65\cdot66\cdot67\cdot68)(75\cdot76\cdot77\cdot78)(133\cdot134\cdot135\cdot136)=218942618242667679360000=467913045600^2. \]

The blocks are disjoint and each has length four, so this is an exact finite witness. It does not by itself establish infinitude; infinitude comes from the Pell recurrence and the identity proved in Bauer–Bennett, not from extrapolating this one calculation. It also does not settle the separate two-block length-four problem.

What is known

Maciej Ulas, “On products of disjoint blocks of consecutive integers,” L’Enseignement Mathématique 51 (2005), 331–334, supplied the first relevant infinite constructions: length-four blocks with four blocks, and with any fixed number at least six. The live entry attributes these cases to Ulas and states that he also conjectured a sufficiently-large-number-of-blocks result for each fixed length.

Bauer and Bennett, “On a question of Erdős and Graham,” L’Enseignement Mathématique 53 (2007), 259–264, prove in Theorem 2.2 that for every number of blocks \(j\geq3\), when all lengths are four, the separated square equation has infinitely many positive-integer solutions. Their proof includes the Pell family used above and extends the earlier cases \(j=4\) and \(j\geq6\) to \(j=3\) and \(j=5\). This is a primary proof of the counterexample needed for Problem 363’s general finiteness assertion.

Bennett and Van Luijk, “Squares from blocks of consecutive integers: a problem of Erdős and Graham,” Indagationes Mathematicae 23 (2012), 123–127, prove in Theorem 1.1 that for every \(r\geq5\), there are infinitely many separated collections of \(r\) blocks of length five whose product is a square. These results concern specific uniform lengths; they should not be inflated into a theorem covering every length and every block count.

GCL's current state

At the frozen catalogue revision teorth/erdosproblems commit 536408716724a2402bdfa7c1d764bf83d64a8c89, the entry’s Git blob is 4e60263a80af6207aca494d2e1388e44fa372889. It records informal status “disproved,” formal status “Lean,” and statement formalized as “yes,” with those status fields dated 10 March 2026. The live page was last edited 2 December 2025, so its displayed editorial date and the frozen catalogue dates are not the same evidence surface.

The pinned Formal Conjectures file is at commit 85f863718beeec7b58a3a1926ee92e3472bc2020, blob 2c12f2693a188afa9e3fe2c64837df9b4264960a. It defines interval finiteness, cardinality at least four, pairwise disjointness, and square-valued total product, then states the equivalence with answer(False). Its proof is explicitly by sorry; the file is therefore a formal statement and a reported formalization, not a GCL proof or certification.

There are semantic details to repair before treating that statement as a faithful interface. It uses Finset ℕ, so zero is available, and Icc a b can be empty when \(a>b\). The cardinality condition is vacuous for an empty list or empty intervals, and lists can repeat empty sets while remaining pairwise disjoint. The mathematical source problem instead concerns positive, nonempty consecutive blocks and a specified finite collection. These differences do not weaken the published counterexample, but they matter for any reusable formal lemma. V2 ledger state, not this article, decides accepted coverage.

What remains

The bounded successor task is to reconcile the formal interface with the primary statement, then formalize only the already-published three-block length-four counterexample or its polynomial/Pell identity. Its input should be the pinned Lean file, the exact Theorem 2.2 family, and a normalization requiring a nonempty finite list of positive starts, strict separation, and explicit block lengths. The target is one theorem saying that the normalized predicate has an infinite family of witnesses, with the parameter recurrence and square identity stated separately.

Acceptance should require a source-linked semantic comparison, a checked identity over integers, and a Lean theorem whose hypotheses exclude empty and zero-containing artefacts. Falsification is any mismatch in the interval endpoints, separation inequality, Pell congruence, or square-root expression. Stop after one faithful family is specified and its finite-index arithmetic has been checked; do not launch an exhaustive build or claim that all length/count cases are resolved. If the Ulas original remains inaccessible, retain that as an explicit documentary gap rather than silently treating secondary citations as a read primary source.

Sources and claim boundaries

The source lock consists of the original issue, the live problem page, the frozen catalogue commit and blob above, the frozen Formal Conjectures commit and blob above, and the directly readable primary Bauer–Bennett PDF. The Bennett–Van Luijk PDF supplies the length-five extension. Ulas’s bibliographic record is DOI 10.5169/seals-3601; the scan endpoint required bot verification during this inquiry, so its full primary text was not independently read here.

Accordingly, this article claims a published Bauer–Bennett resolution of the broad finiteness question, reports the catalogue and live-page statuses with their dates, and records one independently recomputed exact witness. It does not claim Lean replay, mathematical certification, publication authority, or a complete classification of lengths and block counts. Original SOURCE RESULT/1 citation: authenticated original RESULT/1.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1633 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
  • GCL triage: OUTSIDE_STRICT_OPEN_TRIAGE.
  • Formal-statement coverage: A provider statement was included in the pinned snapshot. Presence does not mean the statement or proof has been checked.
  • External mathematical status: consult current problem page and its dated bibliography.

Back to catalogue · Compendium introduction