The Kakeya needle problem asks for the minimum area planar region in which one can completely turn around a line segment through a series of translations and rotations. There is no minimum: There are "Kakeya needle sets" of arbitrarily small area.

I ask the same question but for a rigid plus-sign, two equal-length segments at $90^\circ$ sharing their midpoints, forming a $+$ shape. Because it seems difficult to achieve $360^\circ$ rotation using the type of spikey sets so effective for a single needle, I'm wondering if the answer here might be just a disk?


  • 1
    $\begingroup$ Do you know the answer for three lines at equal angles? $\endgroup$
    – Ville Salo
    Jun 23 at 10:46
  • 2
    $\begingroup$ @VilleSalo: No, I don't. Nor for a T-shape. I thought perhaps the + is the easiest to settle. $\endgroup$ Jun 23 at 11:08

1 Answer 1


I do not know if the minimal area needed to rotate the + is that of the smallest disk containing it. However, I do know that it cannot be done with arbitrarily small area.

This is a special case of a beautiful general theorem by Csörnyei, Héra, and Laczkovich, Theorem 1.2 of Closed sets with the Kakeya property.

I will try to translate the relevant piece of their theorem to have a bit less jargon, but the paper is very clear.

Theorem: Let $A$ be a non-empty, closed, connected subset of the plane. Suppose that there exists a non-trivial rigid motion $m$ of the plane such that $A$ can be continuously moved through rigid motions to $m(A)$ inside sets of arbitrarily small area. Then $A$ must be a line segment, a half-line, a full line, a circular arc, a circle, or a singleton.

There may be a simple argument for the plus sign that also gives your conjectured bound; I haven't thought about it. Maybe looking at their proof would provide it.

New contributor
anon is a new contributor to this site. Take care in asking for clarification, commenting, and answering. Check out our Code of Conduct.
  • 2
    $\begingroup$ Wow, that's an amazing theorem! It still seems possible that there is a set of area smaller than the disk area, but not arbitrarily small. $\endgroup$ 2 days ago
  • $\begingroup$ Welcome to MathOverflow! Please note that the formatting here works a bit different than you're used to, see the Help Center for details. $\endgroup$
    – Glorfindel
    2 days ago
  • 2
    $\begingroup$ I do not get this. If "full line" means "unbounded straight line", how can it be moved within a set of arbitrarily small area unless it is never rotated? And if it is never rotated, then it is simply not moved at all! $\endgroup$
    – user21820
    2 days ago
  • 4
    $\begingroup$ @user21820 I believe the point is that the motion $m$ can be a nontrivial translation of the plane parallel to the line. Then, of course, the line never changes along $m$, but $m$ is a nontrivial rigid motion of the plane. $\endgroup$ 2 days ago
  • $\begingroup$ @LukasMiaskiwskyi: I see. That makes sense, thanks! $\endgroup$
    – user21820
    2 days ago

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.