I hadn't noticed until now tennis balls were symmetric about more than one axis. Which lead me to think there could be an elegant way of expressing the seam as a parametric curve, but I haven't found any yet.

Tennis ball seam

I've tried using products of sines and cosines, but none of them even approximate the curve and it's getting rather ugly.

This is not homework, by the way.

  • 2
    $\begingroup$ See "Designing a Baseball Cover" here: geofhagopian.net/MAM/DesigningBaseballCover.pdf which investigates a related question for baseballs. $\endgroup$
    – Steve Kass
    Feb 28, 2013 at 4:21
  • $\begingroup$ See the description of this youtube video. $\endgroup$
    – Daryl
    Feb 28, 2013 at 4:23
  • 3
    $\begingroup$ The implicit curve $x^2-z^2+y^3/3=0$ on the unit sphere looks pretty good to me: i.stack.imgur.com/5qcd3.png. Perhaps someone can find a nice parametrization. $\endgroup$
    – user856
    Feb 28, 2013 at 6:32
  • $\begingroup$ The two "halves" of the ball are identical, and they have to fit together nicely. So, it seems to me that the dividing curve must have some specific equation. Just the fact that it "goes up and down twice" or "looks pretty good" probably isn't enough. $\endgroup$
    – bubba
    Feb 28, 2013 at 7:55
  • $\begingroup$ I take back (some of) what I said. From the baseball paper, it seems that there is some freedom in the design of the seam, even though the two halves are identical. $\endgroup$
    – bubba
    Feb 28, 2013 at 8:02

2 Answers 2


Let's aim for a simple approximation. Define the curve by $r = r_0$, $\theta = (\pi/3)\sin 2\phi$.

OK, what is that and how did we get it?

We want a curve on the surface of a sphere $r = r_0$. We want the curve to go up and down twice as it revolves around the $z$-axis. We want the curve to be symmetrical in that we could turn it over, shift it halfway around, or mirror-reflect it and it would fit onto itself.

We know that $f(\phi)=\sin 2\phi$ goes up and down twice as $\phi$ goes from $-\pi$ to $+\pi$. We know that $f(\phi)=\sin 2\phi$ has several symmetries such as $f(\phi)=-f(-\phi)$, $f(\phi)=-f(\phi-\pi/2)$, $f(\phi)=f(\pi/2-\phi)$. So we're getting there. Now we want to wrap this $f$ around a sphere somehow. One way is if we imagine the sphere has an equator like Earth's equator and $f$ is proportional to the elevation angle $\theta$ above or below the equator, like latitude on Earth. Let's say that we want $\phi$ to be at maximum, two thirds of the way to the "North Pole", and at minimum, two thirds of the way to the "South Pole". That's one sixth of the circumference of the sphere, which is an angle in radians of $\pi/3$. That will be the proportion: $\theta = (\pi/3)f$.

We have a plan. To be quite clear, it's appropriate to explain how we relate the radius $r$, the elevation angle $\theta$, and the parameter $\phi$ of $f$ (thus also of $\theta$) to Cartesian coordinates.

In $xyz$-space, we can define spherical coordinates in the following way, which is one of the usual ways:

Let $r = \sqrt{x^2 + y^2 + z^2}$.

  • Then $r(x_0,y_0,z_0)$ measures the distance from the point $(0,0,0)$ to the point $(x_0,y_0,z_0)$.

Let $\phi = \arctan(y/x)$.

  • Then $\phi(x_0,y_0,z_0)$ measures the azimuth angle $(1,0,z_0)$, $(0,0,z_0)$, $(x_0,y_0,z_0)$. It's like measuring the longitude at which you're standing on Earth's surface. From (literally) another point of view, it's like measuring the compass direction in which you're facing, if you're standing on Earth's surface (not at a pole).
    The number $1$ is arbitrary. If we would write any other positive number instead of $1$, the angle measure would be the same.
  • This is an angle in the plane where $z=z_0$.
  • The azimuth angle measures zero when $x_0$ is positive and $y_0=0$.
  • Assuming that $x_0$ is positive, the azimuth angle begins to increase as $y_0$ begins to increase from $0$.

Let $\theta = \arctan(z/r)$.

  • Then $\theta(x_0,y_0,z_0)$ measures the elevation angle $(x_0,y_0,0)$, $(0,0,0)$, $(x_0,y_0,z_0)$.
  • This is an angle in the plane where $(x,y)$ is in any proportion $k$ to $(x_0,y_0)$, that is, $(x,y)=k(x_0,y_0)$.
  • The elevation angle measures zero when $z=0$.
  • The elevation angle begins to increase as $z$ begins to increase from $0$.
  • 2
    $\begingroup$ Does this curve split the surface of the ball into two identical pieces? $\endgroup$
    – bubba
    Feb 28, 2013 at 8:02
  • $\begingroup$ This is pretty cool. $\endgroup$ Feb 28, 2013 at 14:25
  • 1
    $\begingroup$ @bubba Thanks, I went on to explain some of the symmetries. So, yes, yes it does. $\endgroup$
    – minopret
    Feb 28, 2013 at 14:56
  • $\begingroup$ @minopret Space curve $ \theta = (\pi/3)\sin \phi $ looks even better to me. $\endgroup$
    – Narasimham
    May 13, 2019 at 4:56
  • $\begingroup$ I'm using mac grapher and unfortunately (even if I switch θ and φ) none of these looks right. Could somebody share a screenshot or link of these equations setup properly on Mac Grapher or Geogebra or something? $\endgroup$
    – cmarangu
    Oct 15, 2019 at 5:07

You might be interested in an article entitled "Generalized Baseball Curves: Three Symmetries and You're In!" published in Loci/JOMA. The article is available online at http://www.maa.org/publications/periodicals/loci/generalized-baseball-curves-three-symmetries-and-youre-in.

We look at the class of spherical curves that have the same symmetries as the curve that forms the seam of a baseball. Also some of the references in this article try to address your question. While we don't directly answer your question by finding the explicit parametric equations of a baseball curve, we do propose a curve (see Figure 7 in the article) that does fit the seam of a real-life baseball very well.

  • $\begingroup$ This is really interesting, thank you! $\endgroup$ Jun 3, 2015 at 1:09

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.