Archimedes of Syracuse is credited with accurately estimating pi, the ratio of a circle's circumference to its diameter. Around 250 BCE, he proved the value lies between 3 10/71 and 3 1/7, or roughly 3.1408 and 3.1429. That two-sided bound was rigorous, not a guess. He squeezed the circle between inscribed and circumscribed polygons of up to 96 sides and computed their perimeters. The method anticipated integral calculus by nearly two millennia.
The Polygon Squeeze Mechanism
The core idea is a trap. Inscribe a regular polygon inside a circle, and its perimeter is slightly less than the circumference. Circumscribe one outside the circle, and its perimeter is slightly more. Because the polygon perimeters are computable with straight lines and square roots, you get lower and upper bounds on pi from pure geometry.
Archimedes started with a hexagon, whose relationship to the circle is trivial. He then doubled the side count repeatedly: 12, 24, 48, finally 96 sides. Each doubling required computing square roots of increasingly awkward numbers, done by hand in Greek numeral notation. The inscribed 96-gon gives a perimeter-to-diameter ratio of about 3.1408. The circumscribed 96-gon gives about 3.1429. Pi is trapped between them.
Here is why the bound is airtight. The inscribed polygon sits entirely inside the circle, so its perimeter cannot exceed the circumference. The circumscribed polygon fully contains the circle, so its perimeter cannot be smaller. No approximation error can hide inside that sandwich. That logical structure, not the numerical answer, is the real achievement.
A plain-language version: Archimedes built a fence just inside the circle's edge and another just outside, then measured both fences.
Why 96 Sides and What the Digits Actually Say
Doubling from 6 to 96 sides is five halvings of the arc per side. Each halving makes the polygons hug the circle more tightly, shrinking the gap between the bounds. The gap between his two limits is only about 0.002, or two parts in a thousand. That precision, from scratch, is remarkable.
His upper bound of 3 1/7 is the famous approximation 22/7. The lower bound, 3 10/71, equals 223/71. The midpoint of his interval, about 3.14185, sits within roughly 0.0004 of the true value of pi, 3.14159265. He never claimed a single number. He claimed a proven interval, which is stronger.
For scale, a circle of diameter 1,000 feet would have its circumference pinned to within about 2 feet by his bounds. For a coin-sized circle of diameter 1 inch, the uncertainty is about 0.002 inch. Both errors are far below any practical measurement he could have made with cords or rods.
Timeline of the Value
| Date | Figure or event | Contribution to pi |
|---|---|---|
| c. 1650 BCE | Egyptian Rhind Papyrus | Implied value near 3.1605 |
| c. 250 BCE | Archimedes of Syracuse | Proven bounds 3.1408 to 3.1429 |
| c. 150 CE | Ptolemy | Value 3.1416, used for centuries |
| c. 480 CE | Zu Chongzhi | Bounds equivalent to 3.1415926 to 3.1415927 |
| 1610s | Ludolph van Ceulen | 35 digits via polygon methods |
| 1706 | William Jones | Symbol pi introduced |
| 1761 | Johann Lambert | Proof that pi is irrational |
The table shows the polygon method had a long shelf life. Van Ceulen pushed the same squeeze technique to 35 digits, a computational feat that consumed much of his life.
Edge Cases: Where the Method Strains
The polygon squeeze is exact but slow. Each doubling roughly halves the error, so gaining one more correct digit demands substantial extra work. Going from 96 sides to many thousands is a grind of nested square roots. That is why pi records stalled for centuries until series methods arrived.
There is also a subtlety in Archimedes' arithmetic. He used rational approximations for the square roots at each step, choosing values that preserved the direction of the inequality. When you approximate downward for the lower bound and upward for the upper bound, the trap stays valid. This careful error accounting is a genuine precursor to rigorous numerical analysis.
A counterintuitive professional caveat: Archimedes' bound of 22/7 as the upper limit is sometimes remembered as his "value of pi." That inverts the logic. 22/7 was the looser edge of a two-sided proof, not a measurement he trusted as exact. Reporting "Archimedes said pi is 22/7" flattens one of the most rigorous arguments in ancient mathematics into a rounded guess.
Misconceptions Worth Correcting
Several myths cling to this result. Each one is easy to check against the actual text of Measurement of a Circle.
Myth: Archimedes found the exact value of pi. He bounded it, which is logically different and stronger than a guess. Myth: He used physical measurement of circles. The method is purely geometric, using no ruler or cord. Myth: 22/7 is his headline answer. It is only the upper bound of his interval. Myth: The idea was entirely original. Approximating circles with polygons predates him; his contribution was the rigorous two-sided proof. Myth: The result was quickly superseded and forgotten. It stood as the standard rigorous bound for centuries.
The most persistent error is treating ancient approximations as if they were modern decimals. Archimedes worked in fractions and geometric ratios, not a decimal system.
Practical Actions for Using This History
If you want to engage with the actual mathematics, here are concrete steps.
Verify the bounds yourself. Compute the perimeter of a regular 96-gon inscribed in and circumscribed about a circle of radius 1. Divide each perimeter by 2. You should recover values near 3.1408 and 3.1429. Watch the convergence. Repeat for 6, 12, 24, 48, and 96 sides. Plot the gap between bounds to see it shrink each doubling. Distinguish bound from estimate. When citing Archimedes, say he proved pi lies between 3 10/71 and 3 1/7, not that he "calculated pi." Use 22/7 only for rough work. Its error against true pi is about 0.04 percent, fine for carpentry, not for precision engineering. Teach the logic, not just the number. The lesson is the sandwich argument, which generalizes to bounding any unknown quantity.
A quick comparison of common approximations puts his result in context.
| Approximation | Numerical value | Error vs. true pi | Status |
|---|---|---|---|
| Archimedes lower bound | 3.1408 | about 0.0008 low | Proven lower limit |
| Archimedes upper bound (22/7) | 3.1429 | about 0.0013 high | Proven upper limit |
| Ptolemy | 3.1416 | about 0.000007 high | Later refinement |
| Zu Chongzhi (milü) | 355/113 | about 0.0000003 | Far more precise, later |
| Common school value | 3.14 | about 0.0016 low | Convenience only |
Note that 355/113, discovered over 700 years later, is dramatically tighter than anything the polygon method produced at 96 sides. Progress required new techniques, not just more sides.
Limits of the Achievement and of This Account
Archimedes did not compute pi to arbitrary precision, and he did not need to. His goal was a rigorous bound sufficient for astronomy and engineering of his era. The exact value of pi is irrational, proven only in 1761, so no finite fraction could ever capture it. His interval approach was the correct response to an unknowable exact decimal.
The precise side counts and the form of his square-root approximations come to us through later copies and translations, so minor numerical details carry some textual uncertainty. The central claim, a two-sided bound near 3.14, is not in doubt.
This article is general educational information about the history and mathematics of pi. It is not a treatment plan, legal opinion, or investment recommendation, and it promises no outcomes. For questions tied to professional engineering tolerances or standards, consult a qualified local specialist.