Can a Polygon Ever Become a Circle?

A hexagon is obviously not a circle.

A 96-sided regular polygon is harder to tell apart from one. A million-sided polygon would look smoother still.

So here is the question left by the previous article:

If a polygon can get closer and closer to a circle, does it ever actually become the circle?

Quick Answer

No polygon with a finite number of sides is a circle.

A polygon has straight sides joined at vertices, or corners. A circle has a smooth curved boundary whose points are all the same distance from its center.

But a sequence of regular polygons can approach a circle as closely as we want. The important idea is not that one last polygon finally changes into a circle. It is that a measurable difference can be made smaller than any tolerance we choose.

“Arbitrarily close” and “exactly equal” are different ideas. A limit lets mathematics talk carefully about that difference.

The Deeper Question — What Does “Approach” Mean?

In everyday language, we can say that a 96-sided polygon “looks almost circular.” Mathematics asks for something more precise:

What are we measuring when we say two shapes are getting closer?

There are several formal ways to compare shapes. For this article, we will use one simple geometric measure that we can see and calculate: the largest radial gap between an inscribed regular polygon and a unit circle.

That gives us one mental model for the whole article:

finite polygon → measurable gap → smaller gap → arbitrarily small gap → limit

Zoom In: The Straight Side Is Still There

Suppose a regular polygon is drawn inside a circle.

Its vertices touch the circle, but between two neighboring vertices the polygon follows a straight chord while the circle keeps curving outward.

When the polygon gains more sides, each chord becomes shorter. The visual mismatch becomes harder to see. But for every finite number of sides, each edge is still straight.

Figure 1. The mismatch becomes smaller as the polygon gains sides, but every finite polygon still has straight edges.

Turn “Closer” into a Number

Now take a circle of radius 1 and place a regular n-sided polygon inside it.

Look at the midpoint of one polygon side. That midpoint is the place on that side that lies farthest inward from the circle.

Draw a line from the center of the circle to that midpoint. The distance from the center to the midpoint is the polygon's apothem—the perpendicular distance from the center of a regular polygon to one of its sides.

For a unit circle, simple right-triangle geometry gives:

apothem = cos(π / n)

The circle's radius is 1, so the largest radial gap along that side is:

gap = 1 − cos(π / n)

Figure 2. The shape comparison becomes a measurable error: radius minus apothem.

Watch the Gap Shrink

Sides	Maximum radial gap
6	0.13397460
12	0.03407417
24	0.00855514
48	0.00214108
96	0.00053541
192	0.00013386
1,000	0.00000493

Every number is still positive. Every finite polygon still misses the circle somewhere between neighboring vertices.

But the gap can be made extremely small.

This lets us say something stronger than “it looks more circular”:

For any tolerance greater than zero, we can choose enough sides to make this radial gap smaller than that tolerance.

That sentence is the bridge from visual approximation to the modern idea of a limit.

What a Limit Adds

A limit describes what a quantity approaches as a process continues. It does not require one finite step to equal the limiting value.

Here, as n grows without bound:

• the side length becomes smaller;
• the radial gap approaches zero;
• the polygon's perimeter approaches the circle's circumference;
• the polygonal boundary approaches the circular boundary.

There is no “last finite polygon” that suddenly becomes a circle. The sequence has a limiting shape.

Is a Circle an “Infinite-Sided Polygon”?

You may hear this phrase because it captures a useful intuition. But as a literal definition, it can create confusion.

A safer statement is:

A circle can be treated as the limiting shape of regular polygons whose side counts keep increasing.

That wording keeps the two ideas separate: every polygon in the sequence has finitely many straight sides, while the circle is the smooth shape the sequence approaches.

Try It — Make the Error Smaller Than a Target

Suppose your tolerance is 0.001. How many sides do you need before the radial gap falls below that value?

Predict first. The table tells us 48 sides are not enough, while 96 sides are.

Now let Python search automatically:

Python — find the first side count below a tolerance

import math

tolerance = 0.001
n = 3

while True:
    gap = 1 - math.cos(math.pi / n)

    if gap < tolerance:
        print("first n:", n)
        print("gap:", gap)
        break

    n += 1

Now change tolerance to 0.0001. Before running the code, predict whether the required side count will increase or decrease.

The exercise changes the question from “Does it look circular?” to “Is the error below the amount I am willing to accept?”

Modern Descendant — Computer Graphics Uses the Same Question

Computer graphics often represents curves and curved surfaces with many straight line segments or triangles.

The screen does not need an abstract infinity of segments. It needs enough segments that the geometric error is smaller than what matters at the current scale, resolution, or engineering tolerance.

That is the same structural question we have just learned:

How fine must the simpler model become before the remaining error is small enough?

The tools are modern, but the habit of refining an approximation and checking its error is much older.

Where Archimedes Fits

Archimedes did not use modern limit notation. In Measurement of a Circle, he worked geometrically with regular polygons inside and outside a circle and pushed the construction to 96 sides to bound π.^[1]

Modern calculus can describe related polygon-refinement processes using limits, but we should not retroactively turn Archimedes' geometry into the later formal theory of limits.^[2]

Words to Keep

finite
Having a count that ends at some number.

apothem
The perpendicular distance from the center of a regular polygon to one of its sides.

tolerance
The largest error we are willing to accept for a particular purpose.

limit
A value or object that a changing sequence approaches as the process continues.

One Sentence to Keep

No finite regular polygon is a circle, but the measurable gap can be made as small as we choose, which is why the circle appears as the limiting shape of the sequence.

What Should We Ask Next?

We now know that the gap can become as small as we want.

But that immediately raises a better question:

How fast does the error shrink when we add more sides?

If we double the side count, does the error become half as large? One quarter? Something else?

Next: How Fast Does the Gap Around Pi Shrink?

Previous: What Happens When a Polygon Gets More and More Sides?

Foundation: How Did Archimedes Calculate Pi Without a Calculator?

Sources & Further Reading

1. MacTutor History of Mathematics, “Archimedes: Numerical Analyst” — Archimedes' inscribed and circumscribed 96-gons and his bounds for π.
2. OpenStax, Calculus Volume 1, “The Limit Laws” — includes a project using inscribed polygons to connect circle geometry with the modern limit idea.
3. David Weisbart, “Modernizing Archimedes’ Construction of π” — a modern treatment of the polygon construction.

Mathematical note: the radial-gap formula, tolerance test, limit language, and Python code are modern tools used to examine the geometry. They are not presented as Archimedes' original notation or proof.