Wait before you dismiss this as a crank question :)

A friend of mine teaches school kids, and the book she uses states something to the following effect:

If you divide the circumference of any circle by its diameter, you get the same number, and this number is an irrational number which starts off as $3.14159... .$

One of the smarter kids in class now has the following doubt:

Why is this number equal to $3.14159....$? Why is it not some other irrational number?

My friend is in a fix as to how to answer this in a sensible manner. Could you help us with this?

I have the following idea about how to answer this: Show that the ratio must be greater than $3$. Now show that it must be less than $3.5$. Then show that it must be greater than $3.1$. And so on ... .

The trouble with this is that I don't know of any easy way of doing this, which would also be accessible to a school student.

Could you point me to some approximation argument of this form?

  • 15,566
  • 8
  • 58
  • 114
  • 807
  • 1
  • 7
  • 8
  • 15
    You reminded me of the time when $\pi = 22/7$ and $\pi$ is irrational bothered the hell out of me. – Aryabhata Jan 13 '11 at 07:51
  • 3
    A quick-and-dirty example: Inscribe a hexagon and circumscribe a square around a unit circle, showing that 3 < $\pi$ < 4. Of course, these are just initial steps of Archimedes' method mentioned in Moron's answer. –  Jan 13 '11 at 08:00
  • 13
    The generalization of the problem sounds absurd. "Why is a = a? Why is it not some other number b, such that b != a?" I must have generalized it in the wrong way. – Chao Xu Jan 13 '11 at 11:56
  • Thank you for all the answers! Three of the answers ( @user3123 , @Moron , @Derek Jennings ) are equally helpful. I'll leave it to my friend to decide which (all) to use (by the way, she was touched by the earnest way in which folks here responded to the question :) ). Since I have to choose one answer to accept, I am choosing @Moron's answer. Many thanks to all of you again. – gphilip Jan 14 '11 at 05:10
  • 1
    uhm it appears that the answers assume the kid means why Pi starts of with 3.141... ! to me the question rather implies that this fact is already obvious to him/her. rather that pupil is interested in how Pi can be a fixed constant if the terminal is not know. not sure if you get what i mean ... but i would rather assume this interpretation of the question because that student obviously already knows the meaning of irrational numbers and therefore it is very likely that the concept of proportionality is also known. –  Jan 13 '13 at 21:38
  • http://math.stackexchange.com/questions/395848/is-the-value-of-pi-in-2d-the-same-in-3d – Jawad May 18 '13 at 21:10
  • 1
    There are several issues here: (a) The ratio is the same for all circles. (b) How can we find the value of this ratio? (c) Why is this value such a strange number? – Christian Blatter May 23 '13 at 18:34
  • I got taken to task by Zev Chonoles for making similar speculation about the value of e, here: http://math.stackexchange.com/questions/26037/intuitive-understanding-of-the-constant-e/54373#54373 – Mike Jones Aug 31 '13 at 13:25
  • For the same reason $\sqrt{2} = 1.41421356237309...$ and not some other irrational number. All other numbers are either to big or too small. We experiment and fuss and get more and more precise. We know $\pi$ is some number so we measure it with more and more precise models or better yet by inscribing polygons inside and outside a circle to get a limit. larger polygons more precise the measurements. – fleablood May 24 '16 at 19:15

9 Answers9


If the kids are not too old you could visually try the following which is very straight forward. Build a few models of a circle of paperboard and then take a wire and put it straigt around it. Mark the point of the line where it is meets itself again and then measure the length of it. You will get something like 3.14..

Pi unrolled

Now let them measure themselves the diameter and circumference of different circles and let them plot them into a graph. Tadaa they will see the that its proportional and this is something they (hopefully) already know.

Or use the approximation using the archimedes algorithm. Still its not really great as they will have to handle big numbers and the result is rather disappointing as it doesn't reveal the irrationality of pi and just gives them a more accurate number of $\pi$.

  • 13,719
  • 3
  • 45
  • 72
  • 4
    That graphic is perfect for this question. Nothing needs be added to make it clearer, nothing can be subtracted without diminishing the clarity. Perfect. – cwallenpoole Jan 13 '11 at 20:01
  • 8
    The excellent animation appears on the Wikipedia entry for $\pi$, which includes other information relevant to the question (such as Archimedes method of approximating $\pi$). http://en.wikipedia.org/wiki/Pi – Jonas Meyer Jan 13 '11 at 20:02
  • 1
    Yes I indeed took this image from the wikipedia article as I liked it a lot too. Shame on me for not mentioning it earlier in the answer. – Listing Jan 13 '11 at 20:11
  • @CarlMummert Well like in the comments to the main question I think the kid wanted to ask how one can obtain (approximately) the value of pi and not the reason why it has that certain value. But that is also an interesting question. – Listing Jan 14 '13 at 23:42
  • but why is it a rational number at all? why is it not a whole number like 3? – Jawad May 18 '13 at 21:31
  • @Jawad: $\pi$ is not a rational number at all. It is not a whole number because the sum $$\sum_{k=0}^\infty(-1)^k\frac{4}{2k+1}$$ is not a whole number. – robjohn May 19 '13 at 00:58
  • 1
    This is just an interpretation of the fact. Does not answer anything. It is like answering: "why is a = b?". With: "because if you take a and compare to b they will be equal! Actually, it is more like saying: "because a is different from c, which is different from b"!!!??? – André Caldas May 19 '13 at 01:00
  • @CarlMummert The best way to show why it's between 3 and 4 instead of some other numbers is to construct a regular hexagon, divided into six equilateral triangles. As the "diameter" of the hexagon is twice the length of the sides of the triangles and the hexagon, it is trivial to note that the perimeter of the hexagon is exactly three times that "diameter". Now construct a circle with that diameter, centered where the six triangles meet. We can clearly see that the arcs that comprise the circle are just a bit longer than the six sides of the hexagon. Therefore pi is just a bit over 3. – Monty Harder May 25 '16 at 18:32
  • But "why" in this case basically "because it just is that way." Some facts are truly atomic, and the value of pi seems to be one of those. – Monty Harder May 25 '16 at 18:34

You can try doing what Archimedes did: using polygons inside and outside the circle.

Here is a webpage which seems to have a good explanation.

An other method one can try is to use the fact that the area of the circle is $\displaystyle \pi r^2$. Take a square, inscribe a circle. Now randomly select points in the square (perhaps by having many students throw pebbles etc at the figure or using rain or a computer etc). Compute the ratio of the points which are within the circle to the total. This ratio should be approximately the ratio of the areas $ = \displaystyle \frac{\pi}{4}$. Of course, this is susceptible to experimental errors :-)

Or maybe just have them compute the approximate area of circle using graph paper, or the approximate perimeter using a piece of string.

Not sure how convincing the physical experiments might be, though.

  • 126
  • 5
  • 79,315
  • 8
  • 179
  • 262
  • One interesting physical experiment used a shotgun loaded with bird shot, although I would think that the density of the shot on a piece of paper would vary by how far it is from the center of the blast. – kleineg Sep 23 '15 at 18:31
  • For reference, the random point experiment is called the Monte Carlo Method, more here (including the specific example of estimating pi): https://en.wikipedia.org/wiki/Monte_Carlo_method – Bitwise May 25 '16 at 18:23
  • Here too: [Liu Hui's Pi Algorithm](https://en.wikipedia.org/wiki/Liu_Hui%27s_%CF%80_algorithm). – Pacerier Jul 04 '17 at 10:59

Given the age of the children, I think that wheeling a bicycle along the ground and measuring the distance traveled for one wheel revolution would be a good idea. This exercise can be continued by asking for suggestions on how to improve the accuracy (e.g. wheel the bike two revolutions etc)

As you obviously have children in the class who pose thoughtful questions, one can ask what do you think would happen to the distance measured on the cycle milometer if pi was suddenly smaller or larger (if one let down or pumped up the tyres). What would happen to the speed registered in such cases. I'm sure there is an interesting exercise for your pupils here. Good luck.

Derek Jennings
  • 13,233
  • 34
  • 59
  • 1
    +1: Like the bike idea. Perhaps wet the tyres, and have a piece of paper making a sound, to help count the number of revolutions, allowing for longer rides and better approximations. – Aryabhata Jan 13 '11 at 08:35
  • @Moron Ooh. Good idea for wetting the tires. But then you'd have the excess tire drip... Maybe lay a towel down to prevent any excess water coming off? – cwallenpoole Jan 13 '11 at 20:03
  • @cwa: Perhaps, but you can try measuring along the center of the tyre track I suppose. – Aryabhata Jan 13 '11 at 21:47

I got this idea from a colleague in grad school. It's one of my favorite lessons.

Since my original posting of this answer, I've created an animation that appears on my trigonography site. The animation itself is too big to post here, but a frame of it appears below.

Take a rubber chicken ---yes, a rubber chicken--- to a chalkboard.

  • Press its feet together firmly to the board, hold a piece of chalk in its beak, and use the chicken like a compass to draw a nice, big circle.

  • Mark a point on the circle you've drawn.

  • Press the chicken's feet together at that point, bend the chicken around the circle, and mark where its beak lands.

  • Do that step again. And again. And again. And again. And again. And ag--- oh, wait ... we can't do it again: we've run out of circle! Hmmmm ...

  • Fold the chicken in half. Then fold it again. Voila! The double-folded chicken just fits into the left-over space on the circle.

enter image description here

  • Consequently, a circle with a radius of $1$ rubber chicken has a circumference of about $6.25$ rubber chickens ... an estimate that's within half a percent of the actual value (namely, $2\pi$)!

  • Now note: There's nothing special about the rubber chicken. We could've used a rubber cat, or a rubber python, or a rubber earthworm, or a rubber Godzilla, or a rubber band that extends from one end of the galaxy to the other. No matter what we object we use for a radius, we can fit just-about six-and-a-quarter of them around the circumference.

This doesn't explain anything about the irrationality of $\pi$ or that kind of thing, but it's a lesson the kids will never forget. :)

  • 69,412
  • 11
  • 108
  • 217
  • 18
    Let me guess, it only works with homogeneous spherical chickens in a vacuum[?](http://en.wikipedia.org/wiki/Spherical_cow) – Myself May 18 '13 at 23:04
  • 3
    @Myself: I think you're thinking of cows. :) – Blue May 18 '13 at 23:05
  • 13
    In what world is there nothing special about rubber chickens? This is why mathematicians are looked down on so strongly... you guys are so abstract and theoretical you can't see the significance of rubber chickens in the real world. – corsiKa Sep 24 '15 at 01:49

Update : When Pi is Not 3.14 | Infinite Series | PBS Digital Studios can be used for teaching.

Why is this number equal to 3.14159....? Why is it not some other (ir)rational number?

Answer is that with the usual Euclidean metric that is the number that one gets, the value of $\pi$ is dependent on the geometry that is being used, so on a sphere the $\pi$ used to obtain the area will be different.

Another question to ask is that if $S=\pi_1 r^2$ is the area of the circle and $C=\pi_2 2 r$ is the circumference why $\pi_1=\pi_2 = \pi$ ? What geometries or metrics will result in $\pi_1 \neq \pi_2$ ?

  • 9,517
  • 6
  • 34
  • 81
  • 2
    Nice take, but you still have to answer why it is $3.14159...$ in the Euclidean case. In other words, you have to define $\pi$ in a more fundamental way, independent of geometry. I think that is beyond what a kid in elementary school can grasp. Therefore, the other answers, particularly that of user3123 and Derek, are better adapted to kids. – Raskolnikov Jan 13 '11 at 13:15
  • For a kid that can use scissors, get them to cut a circles into triangular slices where two sides are the raidus of circle and measure the area of slices by arranging them in a way that would almost cover a square with sides equal to radius of the circle. They can repeat that process until they are satisfied that the ratio of circle covering the square is close enough to 3. Next They can repeat the same thing, using aluminum foil and covering a hemisphere of a sphere and then trying to cover a square with sides equal to great arc, they can measure the greate arc by a string. – jimjim Jan 13 '11 at 14:30
  • 2
    @Raskolnikov : $\pi$ is dependent on geometry and the metric used, but slicing a circle and a hemisphere and getting them to rearrange them to cover squares, physically they can see the ratio being close to $\pi$. As to getting real understanding of why that ratio in Euclidean space is $3.14159\dots$ , there must be a way for them to compare it to non-euclidean space, so the question becomes how do you teach students non-euclidean geometry? – jimjim Jan 13 '11 at 14:47
  • 2
    @Arjang: $\pi$ is independent of the geometry. Wether the universe is hyperbolic or euclidean, the series $\sum_{n=0}^{\infty} \frac{(-1)^n}{2n+1}$ still converges to the same number. What depends on the geometry is the outcome of "circumference divided by diameter". – Raskolnikov Jan 13 '11 at 14:50
  • 2
    @Raskolnikov : "What depends on the geometry is the outcome of "circumference divided by diameter" , but isn't that the definition of $\pi$ students are taught? the series definition also depends on the norm being used, there are number of restriction that need to be met just for a definition of $\pi$ to coincide with the value $3.1415\dots$ , as matter of fact it might even be more beneficial to think of $\pi$ as a variable that only under certain metrics and geometries it's values coincides with the value of pi of this physical reality. – jimjim Jan 13 '11 at 15:03
  • 1
    Moron's answer is more suited for the question (to bound the value above and below). – jimjim Jan 13 '11 at 15:07
  • This is an interesting answer, which I think is related to Alex Basson's comment on the question. This does partially address the question as written, "Why is the ratio of circumference to diameter 3.14159...?", rather than the intended question, "How do we know that (in the Euclidean plane) the ratio of circumference to diameter is 3.14159...?" – Jonas Meyer Jan 13 '11 at 19:52
  • @Arjang : Your answer is a bit above the level of the students in question, and I am reasonably sure that the kid didn't mean it this way :) . Thanks for the answer, all the same. – gphilip Jan 14 '11 at 05:12
  • @gphilip : Just let them know that there are more things to what they are learning, like something they can lookup for fun themselves. I know a 4 year old that plays guitar hero like I never could, I wouldn't be suprised if some of the kids grab the idea and run with it, all they need is just a hint of there is more to what they are learning and they can follow it up for themselves. – jimjim Jan 14 '11 at 05:26
  • @Arjang : Thanks, and you are are right in that one never knows what may ignite a spark. – gphilip Jan 14 '11 at 05:34
  • This is exactly how I read the question, +1 – Carl Mummert Jan 13 '13 at 21:55
  • @Jawad : Howabout 42?http://math.stackexchange.com/questions/254620/pi-in-arbitrary-metric-spaces – jimjim May 18 '13 at 21:44
  • too advanced for me mate but thank you and +1 – Jawad May 18 '13 at 21:47
  • http://math.stackexchange.com/questions/395848/is-the-value-of-pi-in-2d-the-same-in-3d – Jawad May 18 '13 at 21:48
  • @Raskolnikov : https://www.youtube.com/watch?v=ineO1tIyPfM – jimjim Jan 24 '17 at 21:52
  • @gphilip : Added a link to a video for your student – jimjim Jan 24 '17 at 21:56

I wrote a blog article about this a few years ago in response to a friend who asked why $\pi$ came out to be approximately 3, rather than some completely different number. I didn't know, but tried to make a list of potentially relevant lines of investigation. I came up with 8 items, some obviously more pertinent than others. In approximately descending order of quality, they were:

  1. Inscribe a regular hexagon in a circle. The hexagon obviously has a perimeter of 6. The circle passes through the same six points but instead of going straight from one point to the next it takes a circuitous route. So it must be a bit longer. Therefore $\pi$ must be a bit more than 3.

    circle with inscribed hexagon

  2. The shortest curve that can enclose a unit area has length $2\sqrt\pi$.

  3. A penny on the table can be touched by at most six other pennies at the same time; this is closely related to the fact that 6 is the largest integer less than $2\pi$, the circumference of the penny.

    two-dimensional kissing number

  4. $\pi$ is the smallest positive root of the equation $$x-\frac{x^3}6 +\frac{x^5}{120} - \frac{x^7}{5040} +\cdots = 0$$

  5. I mentioned Buffon's needle, but I should have left it out, since it's closely related to $\pi$'s appearance as the perimeter of a unit circle and so is a repeat of earlier items.

  6. The probability that two randomly-selected integers are relatively prime is $\frac6{\pi^2}$, but I think on closer examination this turns out to be related to the circles again.

  7. $$1+\frac14+\frac19 + \cdots = \frac{\pi^2}6$$ but this seems closely related to the previous item.

  8. $\pi$ appears not just in circles but also in spheres, but this is obviously scraping the bottom of the barrel.

People have since written to me pointing out that $\pi$ appears in certain statements of physical law, but these all seem clearly and simply related to its appearance as a circle and sphere constant. For example in Coulomb's law: the electric field E at distance $r$ from a charge $q$ is $$\frac1{4\pi\epsilon_0}\frac{q}{r^2}.$$ But the appearance of $4\pi$ in the denominator here is solely because it is the surface area of a unit sphere, and so we divide by it to get the electric flux per unit area.

I had a bit more to say in the original article. I hope this is some help.

  • 62,206
  • 36
  • 276
  • 489

First off, notice that the kernel of your question is obscured by focusing on the decimal digits of an irrational (indeed, transcendental) number. We could ask the same kind of question about a plain vanilla positive integer. For example, why is it that 1729 ("Hardy's taxicab number") is the smallest positive integer representable as the sum of two cubes in two different ways? Why isn't it some other number instead? Why is it this specific number? However, this new perspective puts us on the road to the answer to your question. Consider this question: Why is it that, for convex polyhedra, the number of vertices minus the number of edges plus the number of faces always equals 2? However, by stereographic projection, this is equivalent to asking why it is that, for connected planar graphs, the number of vertices minus the number of edges plus the number of regions (counting the "infinite" region as one of the regions) always equals 2. However, the proof for the case of connected planar graphs typically involves dismantling the graph in a canonical manner that reduces to a triangle, for which the value of 2 is "obvious". This brings us to the answer to your question: it is not a matter of mathematics, or even of metamathematics, but a question of psychology. The more steeped you are in Mathematics, the more obvious things seem to you. For example, for Euler, who had, among other things, committed many primes to memory, the value of Hardy's taxicab number would perhaps seem self-evident, as it did to Ramanujan. To an Infinite Intelligence (God), everything would be self-evident, but for finite minds there always comes a boundary where they wonder why a certain value is what it is. Nor is Mathematics unique in that regard. In Physics they still haven't figured out why the so-called "fine structure constant" is the value that it is (approximately 1/137), and the reasons for its value is the subject of on-going debate.

Anyway, perspective might be gained and retained by bearing in mind the following famous joke:

Mathematics is really Psychology.

Psychology is really Biology.

Biology is really Chemistry.

Chemistry is really Physics.

Physics is really Mathematics.

Mike Jones
  • 4,310
  • 1
  • 35
  • 39

I think it is a very nice way to explain what is a definition in math. You have got a number with the same value for every one. As long you can make the kids believe the ratio circumference/diameter does not depends on the size of the circle nor of the kid measuring it nor the paper nor of the the weather nor...

Then you leave the choice: do you want to make your computation with the symbol $\pi$ or with the value $3.14159...$ which you will have to remember. Personally, I'd rather learn a new symbol and leave it like that in the formulas rather than go the hard way of multiply with a 5 decimal digit (well and $\infty$ decimal digit in some sense).

Afterword you can think of a way to compute it. You'll go into explaining how clever where the many mathematicians that did try to design they own way to compute $\pi$.

But what really matters is: what ever the way you try, you will find the same number. And in facts, the number of things you can do without knowing the exact value is incredible.

How does know the exact value of $\pi$ by the way ?

And how can you prove that $3.14159...$ so close to $\pi$ ?

  • 716
  • 3
  • 6

"Why is this number equal to 3.14159....? Why is it not some other irrational number?"

One can answer the second question "pi can equal some other irrational number" and it will have little bearing on any sort of application of mathematics. There exist an uncountable infinity of numbers close to pi by the absolute value metric. So, any engineer, physicist, or applied mathematician seeking a certain precision could use a value within that level of precision instead of pi.

So, I don't think the approximations you've gotten here answer your first question.

Doug Spoonwood
  • 10,720
  • 1
  • 30
  • 50