48
$\begingroup$

The question of "What is the oldest open problem in mathematics?" comes up from time to time, and there seems to be consensus that the answer is "Are there any odd perfect numbers?".

There are many other old open problems in number theory. See the prior mathoverflow question on that question for some examples.

What I'm interested to know here is:

What is the oldest (or some candidates for the oldest) open math problem clearly outside of number theory?

EDIT: The same question was asked two years ago on the History of Science and Mathematics StackExchange, but did not elicit an accepted answer.

This is a soft question, so I do not want to apply too rigid of 'rules'. But a few guidelines are:

  • I can't formally define what is number theory. Certainly I want to exclude anything to do with primes, factorizations, irrationality, transcendentality, constructibility, rational/integral solutions to equations, etc.
  • There should be a clear record of the problem being formulated as a conjecture or question (as opposed to "so-and-so surely would have considered X after studying y".)
Improve this question
$\endgroup$
18
  • 8
    $\begingroup$ see hsm.stackexchange.com/q/14748/1697 $\endgroup$ Commented Sep 13, 2024 at 6:30
  • 6
    $\begingroup$ I’m voting to close this question because it has already been asked on hsm.se, where it is more on-topic $\endgroup$ Commented Sep 13, 2024 at 9:12
  • 18
    $\begingroup$ @FedericoPoloni I am not sure. I expect that there exist many mathematicians who can answer questions like this and who are active rather on MO than on HSM $\endgroup$ Commented Sep 13, 2024 at 11:18
  • 16
    $\begingroup$ @FedericoPoloni if mathematicians are not against to see such questions, then why not? History of mathematics is intimately related to mathematical research: every research paper has historical part; the high interest to history of mathematics is much more natural for mathematicians then to anybody else, and it can teach us a lot. The same concerns philosophy of mathematics, teaching mathematics and other things mathematicians have a natural interest in. $\endgroup$ Commented Sep 13, 2024 at 12:29
  • 12
    $\begingroup$ @TimothyChow If it is properly answered on HSM, then it can be closed as a duplicate, but I have doubts about this. Many questions from Math.se are duplicated here if they do not get an answer there, why should it be different with HSM? $\endgroup$ Commented Sep 13, 2024 at 14:11

9 Answers 9

Reset to default
49
$\begingroup$

I don't have a 200 year old candidate but I can offer a 150 year old one (from Maxwell's "Treatise on Electricity and Magnetism" written in 1873).

You are allowed to put $n$ electric point charges in $\mathbb R^3$ in any way you want. What is the maximal number of equilibrium points of the resulting electric field as a function of $n$? (there are several variation of that: both positive and negative charges are allowed, only positive charges are allowed, all charges must be unit, you place them in $\mathbb R^d$ with arbitrary $d$, etc., etc.).

As far as I know, in full generality the question still remains open. I learned it from Alexandre Eremenko, who can probably comment more on it.

This was just to set the "trivial lower bound" :-)

Improve this answer
$\endgroup$
5
  • 5
    $\begingroup$ It seems (from the paper of Gabrielov, Novikov and Shapiro: arxiv.org/pdf/math-ph/0409009) that Maxwell asserted without proof an upper bound (in fact $(n-1)^2$) on the number of equilibrium points in his treatise, rather than posing it as a question/conjecture. That said, it is a good example particularly given that it is precisely formulated in the original source. $\endgroup$ Commented Sep 14, 2024 at 16:58
  • 5
    $\begingroup$ @Mark Lewko: more precisely, they did not prove finiteness, but assuming finiteness, they proved un upper bound. As far as I know, the problem is unsolved even for 3 charges. I've a computer assisted for 3 equal charges, but did not verify that it is correct, and it is not published. $\endgroup$ Commented Sep 15, 2024 at 13:10
  • 3
    $\begingroup$ I still have flashbacks about this problem - Boris Shapiro gave me this problem to work on, first year as an undergraduate student. I did not manage to get far :) $\endgroup$ Commented Oct 3, 2024 at 18:10
  • 6
    $\begingroup$ In principle this problem is algorithmically decidable (as a function of $n$ and $d$) using quantifier elimination for RCF, since a definable set in an o-minimal theory is infinite if and only if it has a non-isolated point (which can be expressed in first-order logic in RCF). I have no idea whether this results in a feasible computation even in the $n=3$, $d=3$ case though. $\endgroup$ Commented Jul 18, 2025 at 10:02
  • 4
    $\begingroup$ Also something that this tells us is that if there is a counterexample, then there's a counterexample with algebraic parameters. $\endgroup$ Commented Jul 18, 2025 at 12:35
31
$\begingroup$

Stability of the Solar System ?

(Question often attributed to Newton in Opticks, 1717 or 1730.)

To further specify as requested by Timothy Chow, make it a few ($3\leqslant N\leqslant 8$) planets under pure Newtonian attraction, as in Suzuki, A history of the stability problem in celestial mechanics, from Newton to Laplace (1642–1787) (p. 24):

The other form of the stability question was that the planets themselves, by mutual gravitation alone, might disturb their orbits “until the system wants a reformation,” a point most famously raised by Newton in Optics. Because of the uncertainty of the validity of universal gravitation, this problem could not even be asked until after mid-century, but thereafter, progress was rapid and by 1760, a preliminary answer was obtained by Charles Euler, followed, over the next two decades, by the works of Lagrange and Laplace.

A “clear record of the problem being formulated as a conjecture or question” is in e.g. Laplace (1784):

si l’on n’a égard qu’aux lois de la gravitation universelle, les moyennes distances des corps célestes aux foyers de leurs forces principales sont immuables (…). Mais les excentricités et les inclinaisons sont-elles renfermées constamment dans d’étroites limites? C’est un point important du système du monde qui reste encore à éclaircir.

Improve this answer
$\endgroup$
10
  • 11
    $\begingroup$ I was about to suggest this one, but it's a little murky. First, is it a physics question about the actual solar system, or a math question about the equations of Newtonian mechanics? If the latter, what is the question exactly? E.g., Arnold's work on the KAM theorem tells us that there are some solar-system-like systems that are stable. So maybe the question is determining which systems are stable and which aren't. In addition to the article you linked to, this Quanta article is also interesting. $\endgroup$ Commented Sep 14, 2024 at 12:36
  • 2
    $\begingroup$ That question, or some version of it, is still open, but it isn't the question Newton asked. This article by Scott Tremaine gives some more information about what the early discussions of the problem looked like. $\endgroup$ Commented Sep 14, 2024 at 12:41
  • 1
    $\begingroup$ Suzuki (1996, §1.5 “The Stability Question Raised”) discusses the two versions: “non-gravitational causes”, versus analysis of the purely gravitational problem (p. 24). The latter is the mathematical problem I have in mind, discussed by Laskar. $\endgroup$ Commented Sep 14, 2024 at 12:47
  • 2
    $\begingroup$ I see you have edited the question to clarify, but the number of planets is not the only critical parameter; the mass-ratio is also important. See for example this abstract. $\endgroup$ Commented Sep 14, 2024 at 15:37
  • 4
    $\begingroup$ There is actually a bizarre story surrounding the "Charles Euler" paper. Apparently Euler entered a Paris Academy prize competition in his son's name (Carl/Carolus/Charles). Jacobi speculates that this was because he had been teased for winning these competitions so many times. For proof of the true author Jacobi cites the inscription: "The father decided that all the stars in heaven should move." $\endgroup$ Commented Sep 16, 2024 at 0:33
17
$\begingroup$

I suggested this candidate in a comment: For which $d$ and $g$ does there exist a curve in $\mathbb{P}^3$ of degree $d$ and genus $g$? In Hartshorne's Algebraic Geometry, Chapter VI, Section 6, it is indicated that the Steiner Prize was awarded in 1882 to Noether and Halphen for work on this problem and related problems (although if one wants to be pedantic about the earliest date on which the question was formulated explicitly, I don't know when that would be).According to another MO answer, Hartshorne's claim that this problem is open remains true today.

EDIT: Looking more carefully at the references in that other MO answer, I think the paper by Gruson and Peskin does actually solve this problem. (Mori is strengthening their result in certain special cases.) If so, then this problem is no longer open.

If this particular problem doesn't answer Mark Lewko's question, then there may be similar questions in classical algebraic geometry that do. For comparison, the interpolation problem, of determining when there is a (Brill–Noether) curve of degree $d$ and genus $g$ passing through $n$ general points in $\mathbb{P}^r$, was only recently solved by Larson and Vogt.

Improve this answer
$\endgroup$
3
  • $\begingroup$ Look at the reference I gave hsm.stackexchange.com/questions/14748/… for examples "outside physics" and older than those in this list. $\endgroup$ Commented Sep 15, 2024 at 13:13
  • 1
    $\begingroup$ @TimothyChow: do you have any information about the original source of this problem? As an aside, I had never heard of the Steiner Prize. Apparently for some of its existence (which came to an end with WWI) eminent mathematicians of the era (including Dirichlet and Weierstrass) would pose problems, with the idea the prize would go to the best work on the problem. Reportedly these problems frequently went unsolved and the prizes unclaimed. It would be interesting to know if the full statements of these problems survive. See: ams.org/bookstore/pspdf/mprize-prev.pdf $\endgroup$ Commented Sep 15, 2024 at 22:44
  • $\begingroup$ @MarkLewko I don't think that tracking down the text of the people who offered the prize is going to lead anywhere. A more promising route may be to look at the collected works of Halphen, to see what problems he stated explicitly. $\endgroup$ Commented Sep 15, 2024 at 22:49
15
$\begingroup$

The Navier–Stokes equation was mentioned in a comment so I thought I would give a link to Sylvio R. Bistafa's essay, 200 Years of the Navier-Stokes Equation, which gives some historical information. Briefly, although Navier's work appeared in 1822, the Clay Prize version of the problem is probably not older than the problem fedja suggested, because the early work on the equation implicitly or explicitly assumed laminar flow.

Improve this answer
$\endgroup$
1
  • 10
    $\begingroup$ I originally voted to close this MO question as a duplicate of the HSM question, but since the community has decided to reopen the MO question, I edited the HSM question to point here to MO. $\endgroup$ Commented Sep 14, 2024 at 13:11
12
$\begingroup$

The inverse Galois problem is another possible candidate, though it may not be easy to track down the earliest explicit statement of it. Wikipedia claims that it was posed in the early 19th century, but Galois died in 1832 and there does not seem to be any concrete evidence that he explicitly posed the problem. I kind of doubt it was explicitly posed any earlier than 1853, when Kronecker first stated what we now call the Kronecker–Weber theorem. In the other direction, the problem should be dated no later than Hilbert's 1892 Crelle paper, "Uber die Irreducibilitat ganzer rationaler Functionen mit ganzzahligen Coefficienten".

Improve this answer
$\endgroup$
7
  • 12
    $\begingroup$ This problem would almost always be considered part of number theory. $\endgroup$ Commented Sep 15, 2024 at 23:14
  • 8
    $\begingroup$ @WillSawin Could be, though in my anecdotal experience, number theorists don't seem to be particularly interested in it, while (for example) group theorists like John Thompson are. $\endgroup$ Commented Sep 15, 2024 at 23:20
  • 1
    $\begingroup$ Well for example David Zywina has done a lot of work on it. $\endgroup$ Commented Sep 16, 2024 at 12:33
  • 1
    $\begingroup$ ... and, after all, who gets to decide what is or isn't "number theory"? :) $\endgroup$ Commented Sep 16, 2024 at 17:27
  • 3
    $\begingroup$ Number theorists are very much interested in the quantitative variant (how many number fields of given $G$ and discriminant at most $X$ in absolute value), cf Malle's conjecture and work on it. $\endgroup$ Commented Sep 17, 2024 at 18:43
12
$\begingroup$

According to Wikipedia, it was already discussed in the 18th century if chess is a draw. Hence this question is a good candidate for being the oldest open problem in mathematics outside number theory. (Note that chess is mathematics, even if we don't think of it that way.)

Improve this answer
$\endgroup$
6
  • 3
    $\begingroup$ +1, although some details of the rules of chess have changed even in the 20th century, so technically it would have been a slightly different problem $\endgroup$ Commented Jul 19, 2025 at 0:46
  • 1
    $\begingroup$ If this kind of thing counts, then perhaps the analogous question in go also counts. In go, the expert consensus on the value of the first-move advantage is explicitly codified in the game, in the form of komi. One could interpret the komi value as a mathematical conjecture (although the counter-argument is that the komi value is an estimate of the practical advantage of going first rather than the theoretical value of going first). $\endgroup$ Commented Jul 19, 2025 at 13:43
  • $\begingroup$ @TimothyChow I am not familiar with go. For any deterministic two-person board game (where the possible outcomes are "white wins", "black wins", "draw"), it is a mathematical problem whether the game is a draw. However, this does not mean this question was explicitly asked and researched (and is still unsettled). What I tried to convey is that in the case of chess, the question was explicitly asked and researched already in the 18th century. If this happened earlier with chess or another board game, it serves as a potential candidate and should be mentioned. $\endgroup$ Commented Jul 19, 2025 at 17:16
  • $\begingroup$ @GHfromMO In go, one does not only win or lose; one loses by a certain number of points. As in chess, experts believe that the first player has an advantage. Not only that, they have attempted to formally quantify the magnitude of that advantage, by giving the second player a certain number of points to compensate for the disadvantage of going second. This is "komi." The appropriate value of this compensation has been much discussed by experts of the game, and there is no uniform consensus on its "correct" value. But the point is that they explicitly confronted the issue quantitatively. $\endgroup$ Commented Jul 19, 2025 at 18:16
  • 3
    $\begingroup$ @GHfromMO In chess, if a forced win for White is scored as $+1$, a forced win for Black is scored as $-1$, and a draw is scored as $0$, then the mathematical conjecture in question is "The score of chess is 0." In go, the analogous conjecture is, "The score of go is 6 or 7," and that question has been explicitly studied. It is just as mathematical as "The score of chess is 0." The only caveat is that it may be that the chess literature is more explicit about the notion of "perfect play." I'm not familiar enough with the go literature to know whether they had the concept of perfect play. $\endgroup$ Commented Jul 19, 2025 at 18:58
8
$\begingroup$

Quanta magazine https://www.quantamagazine.org/new-math-revives-geometrys-oldest-problems-20250926/ just reported the following:

In the third century BCE, Apollonius of Perga asked how many circles one could draw that would touch three given circles at exactly one point each. It would take 1,800 years to prove the answer: eight.

Quanta Eight

(Image from Quanta article.)

If you believe Quanta magazine that this question was indeed asked 1800 years ago and solved only now, then:

(a) This problem was open at the time this Mathoverflow question was asked

(b) It was open for about 1800 years, while other answers talk about problems about 200 years old, and

(c) It is a geometry problem. It is, like many other geometry problems, can be written in coordinates and reduced to some equations, but this is just one of several possible solution tools - by its nature it is still a geometry problem!

Improve this answer
$\endgroup$
3
  • 6
    $\begingroup$ 1800 years after the 3rd century BCE is the 16th century CE, or the 1500s. I don't know if it's true that the question of Apollonius was answered in the 1500s, but anyway that question is not the one that has been answered now. $\endgroup$ Commented Sep 26, 2025 at 23:27
  • 5
    $\begingroup$ I agree with @ZachTeitler, I think this answer is based on a misunderstanding of the Quanta article, which was just using the question of Apollonius of Perga as an example to motivate the general area of enumerative geometry. $\endgroup$ Commented Sep 27, 2025 at 0:45
  • 5
    $\begingroup$ en.wikipedia.org/wiki/Problem_of_Apollonius says that Apollonius himself solved it. The 16th century seems to refer to Adriaan van Roomen, whose solution appears to be the oldest surviving in full detail. $\endgroup$ Commented Sep 27, 2025 at 1:56
6
$\begingroup$

A long-standing unresolved problem in celestial mechanics is to determine the number / finiteness of relative equilibria in the $n$-body problem, for any choice of positive masses $m_1, \ldots, m_n$. (#6 of Smale's problems.) The $n$-body problem was introduced at least as far back as Newton, but does anyone have a reference for the earliest reference to this specific question?

Improve this answer
$\endgroup$
6
$\begingroup$

I just learned from Quanta Magazine about a geometry problem of the 17th century which was solved in August 2025.

Given a convex polytope in $R^3$, can you drill a hole in it through which another copy of the same polytope can pass?

Prince Rupert showed in 1693 experimentally that cubes have this property. Wallis proved this rigorously. Since then, many convex polytopes were shown to have this "Rupert property". Only recently the first example of a convex polytope without Rupert property was constructed: https://arxiv.org/abs/2508.18475

Improve this answer
$\endgroup$
7
  • 1
    $\begingroup$ This is a series of questions, one for each convex polyhedron, rather than a single question. I don't think any one that is still open was asked in the 17th century. (The question about whether all convex polyehdra have this property is no longer open, and was asked in 2017.) $\endgroup$ Commented Oct 25, 2025 at 14:16
  • 2
    $\begingroup$ @Will Sawin: According to V. I. Arnold, a good problem is not a yes/no question, but a direction of research. Every mathematician, once s/he learns about Prince Rupert result, will immediately ask herself "which convex polytopes have this property?". $\endgroup$ Commented Oct 25, 2025 at 17:37
  • 1
    $\begingroup$ I would be happy with such logic if I ask the question, so I voted up. However, the given question explicitly ask that "There should be a clear record of the problem being formulated as a conjecture or question (as opposed to "so-and-so surely would have considered X after studying y".)" $\endgroup$ Commented Oct 26, 2025 at 7:44
  • 1
    $\begingroup$ @Bogdan Grechuk: So it remains to find Wallis' paper and check whether he asks the general question there:-) $\endgroup$ Commented Oct 26, 2025 at 13:08
  • $\begingroup$ I like this problem, although it's ironic that it wasn't posted here until arguably the biggest open problem in the area (does there exist a convex polytope without the Rupert property) was solved... $\endgroup$ Commented Oct 27, 2025 at 14:34

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.