Skip to content

Erdős Problem #375

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

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 375 is Grimm's conjecture about a run of consecutive composite numbers. Given integers \(n,k ≥ 1\), suppose every number from \(n+1\) through \(n+k\) is composite. The question asks whether one can assign to each position \(i\) a prime \(p_i\) so that \(p_i ∣ n+i\), with no prime used twice. Thus the assignment is a system of distinct prime divisors for the whole block, not merely a statement that each number has some prime divisor.

The distinction matters. A composite number can have several prime divisors, but choices made for one position can collide with choices for another. The conjecture says that a collision-free choice always exists. The live Erdős Problems page calls the problem “FALSIFIABLE”: it is treated as open, while a single finite counterexample would disprove it. That label is not a claim of decidability, independence, or truth in a specified formal theory. No source read for this article reports an independence theorem. The natural mathematical reading is the ordinary arithmetic statement above, with the usual meanings of prime, divisibility, and consecutive integer.

The wording is preserved from the live page and from the primary Erdős–Selfridge source. In particular, the divisibility target is \(n+i\), while the composite block is \(n+1, …, n+k\).

A small example

Take \(n=48\) and \(k=4\). The block is \(49,50,51,52\), and an exact factor check gives:

factor 49 50 51 52
49: 7 7
50: 2 5 5
51: 3 17
52: 2 2 13
49%7=0, 50%5=0, 51%17=0, 52%13=0

The witnesses \((p_1,p_2,p_3,p_4)=(7,5,17,13)\) are distinct primes and divide the corresponding entries. The same calculation also makes compositeness visible: each entry has at least two prime factors, counting multiplicity. This is a reproducible finite check, not a proof of the universal statement. It tests one block only; even a very large bounded computation would leave all larger or unsearched blocks untouched.

The first two lengths are elementary. For \(k=1\), a composite number has a prime divisor. For \(k=2\), if the two consecutive composites shared their only possible prime witness, that prime would divide their difference \(1\), which is impossible. Larger blocks require controlling many possible collisions.

What is known

The primary historical trail begins with C. A. Grimm's 1969 paper, “A conjecture on consecutive composite numbers,” and the 1971 paper by Paul Erdős and J. L. Selfridge, “Some problems on the prime factors of consecutive integers II,” pages 13–21. The latter source states the distinct-prime-divisor conjecture and studies related functions measuring how long the required separation can be maintained. Erdős and Graham list the problem in their 1980 monograph, with the catalogue citation pointing to page 71.

The current live page reports several partial results, each weaker than the full universal conjecture. It records Grimm's range \(k ≪ log n / loglog n\), an improvement attributed to Erdős and Selfridge, and the Ramachandra–Shorey–Tijdeman result \(k ≪ (log n / loglog n)^3\). The cited paper is “On Grimm's problem relating to factorisation of a block of consecutive integers,” Journal für die reine und angewandte Mathematik 273 (1975), 109–124, DOI 10.1515/crll.1975.273. The same page reports that Laishram and Shorey verified the full assignment for all \(k ≥ 1\) when \(n ≤ 1.9 × 10¹⁰\); their paper is “Grimm's conjecture on consecutive integers,” International Journal of Number Theory 2 (2006), 207–211, DOI 10.1142/S1793042106000498.

These are bounded-range or conditional-length results, not a resolution of every \(n\) and \(k\). This article treats the displayed bounds as reported literature claims tied to their references; it does not certify the proofs or independently reproduce the \(1.9 × 10¹⁰\) computation.

GCL's current state

The protected source packet for issue #1645 fixes the catalogue at teorth/erdosproblems commit 536408716724a2402bdfa7c1d764bf83d64a8c89, with entry blob 4e60263a80af6207aca494d2e1388e44fa372889. That entry says falsifiable, last updated 2025-08-31; it records formal_status: unformalized and formalized: yes dated 2026-01-18. The catalogue is metadata, not the mathematical-status authority by itself, so the live page was checked as well.

The pinned FormalConjectures file is 375.lean, commit 85f863718beeec7b58a3a1926ee92e3472bc2020, blob e68e24b79463a32b5c61c101fef737bd86d0cc97. Its Erdos375Prop quantifies over \(n ≥ 1\) and all natural \(k\), assumes every \(n+i+1\) for \(i<k\) is non-prime, and asks for an injective map from Fin k to primes dividing \(n+i+1\). The unrestricted \(k=0\) case is vacuous, so the content for positive k matches the source statement.

The theorem named erdos_375 is marked open and proved only by sorry; the listed bounded-gap, Legendre, and logarithmic variants also contain sorry placeholders. File presence is therefore statement-formalization metadata, not a checked proof or a semantic-equivalence certificate. GCL has not supplied mathematical certification here. The V2 ledger, rather than this article, decides accepted coverage.

What remains

The most useful bounded successor is a statement-and-evidence audit, not an attempt to turn the special status into an ordinary solved/open claim. Its input should be the pinned proposition, the live source text as accessed 2026-10-13, and the cited primary-source records. Its target should be a line-by-line normalization table: source variables \(n,k,i\), formal variables, positive-domain conventions, composite hypotheses, injectivity, and divisibility. Acceptance requires exact agreement modulo the harmless vacuous \(k=0\) extension, explicit recording of every source-reported partial result, and a finite checker that reproduces small instances such as \((48,4)\). Falsification is either a genuine quantifier mismatch, a counterexample within the selected finite bound, or a primary source showing that a reported result has been misstated. Stop when the normalization table, source hashes, and bounded checker are complete; do not infer a universal proof from finite success.

If a later worker studies the logical label itself, it must name the base theory and the exact metamathematical claim. “Falsifiable” here means that a finite counterexample could refute the arithmetic sentence. It does not mean that a proof or counterexample is known, nor that the sentence is independent of Peano arithmetic or set theory.

Sources and claim boundaries

Primary and source documents are Grimm (1969), Erdős–Selfridge (1971), Erdős–Graham (1980), Ramachandra–Shorey–Tijdeman (1975), and Laishram–Shorey (2006). The live status and wording are Erdős Problems #375, last edited 24 January 2026 and accessed 2026-10-13. The governing assignment and its zero-comment state are the original GCL issue.

The exact protected source identities are the catalogue commit/blob and FormalConjectures commit/blob recorded above. The original SOURCE RESULT/1 citation is authenticated original RESULT/1. No future comment, proof, independence result, publication decision, priority claim, or GCL mathematical certification is being invented here. The article describes the evidence closure available to this bounded audit; it does not decide accepted coverage.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1645 — 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