Papex

Open problem · Mathematics · Open

Legendre’s conjecture

Is there always a prime number between two consecutive perfect squares?

Summary

Legendre conjectured that a prime lies between n² and (n + 1)² for every positive integer n. It has been checked by computer to very large n and would follow from strong enough bounds on gaps between primes, but no proof covers every n.

Why it is open

In 1912 Edmund Landau listed four problems about primes that he considered "unattackable at the present state of science"; one asks whether there is at least one prime between n² and (n + 1)² for every positive integer n [2]. Pintz numbers it fourth, in Landau's original order, and the Wikipedia article on Landau's problems third; the others are Goldbach's conjecture, the twin prime conjecture and the question of primes of the form n² + 1 [2, 10]. The conjecture is attributed to Adrien-Marie Legendre [6, 7].

It is a question about large gaps between primes. Writing dₙ for the gap after the nth prime pₙ, Pintz notes that a bound dₙ < 2√pₙ would imply the conjecture and is of about the same strength [2]. The best unconditional bound is weaker. In 2001 Roger Baker, Glyn Harman and János Pintz proved that for every x beyond some point x₀ there is a prime between x − x^0.525 and x [1]. Marc Chamberland and Armin Straub note that this gives a prime between the consecutive powers n^α and (n + 1)^α when α ≥ 2.106 and n is large enough [7]; Legendre's conjecture is the case α = 2.

The Riemann hypothesis (RH), that all non-trivial zeros of the zeta function have real part 1/2, would not settle it either. Harald Cramér showed in 1920 that RH bounds the gaps by a constant times √p log p, which, Pintz writes, "still falls short of answering Landau's question No. 4 positively, even assuming the RH" [2]. Ethan Lee and Peter Campbell, writing in 2026, likewise describe the conjecture as out of reach or open even under RH [8, 9].

No prize or body sets terms for a resolution; a proof for every n, or one interval without a prime, would settle it. The status here follows the Wikipedia article on Landau's problems, which says that as of 2026 all four problems are unresolved [10].

What is known

Gaps between primes (1930–2001). Guido Hoheisel showed in 1930 that dₙ is at most a constant times pₙ^θ for some θ below 1, with θ = 1 − 1/33,000 [2]. Albert Ingham reached θ = 5/8 + ε in 1937, which gives a prime between n³ and (n + 1)³ for every sufficiently large n [2]. Henryk Iwaniec and Matti Jutila obtained θ = 13/23 in 1979 by combining analytic and sieve methods, and later improvements used both [2]. The best exponent known in 2026, 21/40 = 0.525, is the 2001 result of Baker, Harman and Pintz, who add that with enough effort their threshold x₀ "could be determined effectively" [1, 7].

Explicit powers (2023–2026). A result is explicit when the range of n it covers is stated. Michaela Cully-Hugill made Ingham's cube result explicit for every n ≥ exp(exp(32.892)); Lee writes that the remaining range "is largely impervious to modern techniques" [8]. In 2023 Cully-Hugill proved that there is a prime between n¹⁵⁵ and (n + 1)¹⁵⁵ for every n ≥ 1 [5]. Chamberland and Straub report that Cully-Hugill and Daniel Johnston lowered the power to 90 in 2025 [7], and a 2026 preprint by Lee, not peer reviewed, proves it for 86 [8].

Under the Riemann hypothesis (1920–2026). Emanuel Carneiro, Micah Milinovich and Kannan Soundararajan proved in 2019 that, on RH, there is a prime between x and x + (22/25)√x log x for every x ≥ 4 [4]. Under RH any gap exponent θ above 1/2 is admissible, while a conjecture of Cramér, that gaps are at most of order (log p)², would allow any θ above 0 [7]. In a 2026 preprint, not peer reviewed, Chamberland and Straub combine the bound of Carneiro and colleagues with the computation below: assuming RH, there is a prime between x^(2+δ) and (x + 1)^(2+δ) for every real x ≥ 1 when δ ≥ 1/4, and with more work δ can be lowered to 0.2253 [7].

Almost primes (1920–2026). An almost prime is an integer with few prime factors. Viggo Brun showed in 1920 that for large x every interval (x, x + √x), and so every interval between large consecutive squares, contains an integer with at most 11 prime factors; Jing-Run Chen lowered this to two in 1975, again for large n [2]. For every n ≥ 1, Adrian Dudek and Johnston found an integer with at most four prime factors, counted with multiplicity, in (n², (n + 1)²), and Campbell, in the Journal of Number Theory, at most three [9]. His proof gives at most two whenever n² ≤ 10³¹ [9].

Computation (2013–2024). Tomás Oliveira e Silva, Siegfried Herzog and Silvio Pardi confirmed the even Goldbach conjecture up to 4 × 10¹⁸ and computed the gaps between primes up to that bound [3]. Up to 4 × 10¹⁸ every gap after a prime p ≥ 11 is smaller than (log p)², Carneiro and colleagues report [4], and Jonathan Sorenson and Jonathan Webster give n ≤ 2 × 10⁹ as the resulting record for Legendre's conjecture [6]. Their own algorithm, run for about half a year on each of two platforms, verified the stronger conjecture of Ludvig Oppermann, with primes between n² and n(n + 1) and between n(n + 1) and (n + 1)², for all n ≤ 7.05 × 10¹³, finding 27-digit primes [6].

Main approaches

Zeros of the zeta function. The explicit formula of prime number theory ties the primes in a short interval to the zeros of the zeta function, and bounds on the number of zeros far from the line of real part 1/2 (zero-density estimates) give primes in shorter intervals [2]. Strongest evidence: Ingham's theorem for cubes and the explicit powers 155 and 90 [2, 5, 7]. Main difficulty: even with every zero on the line, as RH asserts, the method gives gaps of order √p log p [2, 6]; for explicit versions, Lee writes that further gains in one key error term "are limited, as the present results are quite sharp" [8].

Sieves combined with analysis. Sieve methods count integers left after removing multiples of small primes; Baker, Harman and Pintz combine Harman's sieve with mean-value estimates for Dirichlet polynomials in place of zero-density estimates [1]. Strongest evidence: the exponent 0.525 [1]. Main difficulty: for consecutive powers the result reaches only α = 2.106, for large n [9], and Chamberland and Straub expect both the effort needed to make its threshold explicit and the threshold itself to be enormous [7].

Almost primes. Weighted sieves find integers with few prime factors in every interval [9]. Strongest evidence: Chen's two prime factors for large n and Campbell's three for all n [2, 9]. Main difficulty: the parity obstacle, which in general prevents sieve methods from detecting primes [2]; Campbell writes that reaching two prime factors for all n "is beyond the present framework" [9].

Computation and random models. Cramér's model treats each integer n as prime with probability 1/log n; it predicts gaps of order (log p)² and the truth of all four of Landau's conjectures [2]. Strongest evidence: computed gaps stay below (log p)² up to 4 × 10¹⁸ [4], and verification reaches n ≤ 7.05 × 10¹³ [6]. Main difficulty: no computation covers all n, and Helmut Maier showed in 1985 that the model's predicted count of primes fails in some intervals of length (log x)^λ, however large the fixed λ [2].

Open sub-questions

  1. Explicit cubes. How far are the zero-free regions now proved from the minimal regions Lee computes for primes between consecutive kth powers, k ≥ 65, and which region would make Ingham's cube theorem hold for every n ≥ 1 [8]?
  2. Below the 86th power. Can the power k with a prime in every (nᵏ, (n + 1)ᵏ), n ≥ 1, be lowered below 86, for instance by extending Lee's result for k = 70 on a subsequence of integers to all n [7, 8]?
  3. An explicit Baker–Harman–Pintz theorem. Can the threshold x₀ for primes between x − x^0.525 and x be computed, and which consecutive powers would the theorem then cover for every n [1, 7]?
  4. Two prime factors for every n. Can Campbell's three prime factors be lowered to two for all n, given that his computation already gives two for n² ≤ 10³¹ and that he points to Iwaniec's bilinear form of the sieve error term [9]?
  5. Below δ = 1/4 on RH. With the bound of Carneiro and colleagues and verification beyond n = 7.05 × 10¹³, how far can Chamberland and Straub's δ be lowered for all real x ≥ 1, given their view that δ ≥ 1/5 "seems out of reach" [4, 6, 7]?

Work on this in Papex

If you have a testable answer to one of the sub-questions above, you can draft it as a private hypothesis in Papex, or prepare the draft from Claude or ChatGPT with the Papex connector. You review the exact version, sign it and decide whether to publish it; a public record that names this problem is listed on this page. A Papex record shows who signed which version and when. It is not peer review and does not mean the hypothesis is correct.

Key references

  1. Baker, R. C., Harman, G. & Pintz, J. (2001). The difference between consecutive primes, II. Proceedings of the London Mathematical Society 83(3), 532–562. 10.1112/plms/83.3.532 · Checked 16 Oct 2026
  2. Pintz, J. (2009). Landau’s problems on primes. Journal de Théorie des Nombres de Bordeaux 21(2), 357–404. 10.5802/jtnb.676 · Checked 16 Oct 2026
  3. Oliveira e Silva, T., Herzog, S. & Pardi, S. (2014). Empirical verification of the even Goldbach conjecture and computation of prime gaps up to 4·10¹⁸. Mathematics of Computation 83(288), 2033–2060 (published online 2013). 10.1090/S0025-5718-2013-02787-1 · Checked 16 Oct 2026
  4. Carneiro, E., Milinovich, M. B. & Soundararajan, K. (2019). Fourier optimization and prime gaps. Commentarii Mathematici Helvetici 94(3), 533–568. 10.4171/CMH/467 · Checked 16 Oct 2026
  5. Cully-Hugill, M. (2023). Primes between consecutive powers. Journal of Number Theory 247, 100–117. 10.1016/j.jnt.2022.12.002 · Checked 16 Oct 2026
  6. Sorenson, J. & Webster, J. (2025). An algorithm and computation to verify Legendre’s conjecture up to 7·10¹³. Research in Number Theory 11(1), 4 (published online 6 December 2024). 10.1007/s40993-024-00589-4 · Checked 16 Oct 2026
  7. Chamberland, M. & Straub, A. (2026). Weakening the Legendre conjecture. Preprint, arXiv:2602.22502 (version 1, 26 February 2026), not peer reviewed. Source · Checked 16 Oct 2026
  8. Lee, E. S. (2026). Minimal zero-free regions for results on primes between consecutive perfect kth powers. Preprint, arXiv:2602.14340 (version 2, 3 March 2026), not peer reviewed. Source · Checked 16 Oct 2026
  9. Campbell, P. J. (2027). On the existence of integers with at most 3 prime factors between every pair of consecutive squares. Journal of Number Theory 293, 488–508 (volume dated April 2027; DOI registered August 2026). 10.1016/j.jnt.2026.07.014 · Checked 16 Oct 2026
  10. Wikipedia. Landau’s problems (revision of 3 June 2026). Source · Checked 16 Oct 2026

Page reviewed 16 Oct 2026

Source

Open, as listed by Wikipedia · Landau’s problems. Checked 08 Oct 2026.

The status is the source’s. Papex does not decide whether a problem is solved.

Records on this problem

No public record addresses this problem yet.

More in Mathematics