You draw ten triangles. The same pattern appears every time.
Then you draw a hundred more. Still the same.
Is that enough to say the pattern is always true?
Mathematics asks for something stronger.
Quick Answer
A proof explains why a conclusion must follow from what we already accept.
It does not check every case one by one. It finds a reason that covers every case allowed by the statement.
Examples show that an idea works somewhere. A proof shows why it must work everywhere the claim says it should.
A Proof Is a Path of Reasons
In the previous article, we saw how Euclid built geometry from definitions, postulates, common notions, and earlier propositions.
A proof is the path connecting those starting points to a new conclusion.
what we accept → justified step → justified step → conclusion
Each step must have a reason.
That reason may come from a definition. It may come from a starting assumption. It may come from a theorem already proved. Or it may come from a valid rule of logic.
David Joyce's guide to Euclid describes the propositions in Book I in this way: each statement in a proof is justified by a definition, postulate, common notion, or an earlier proposition.[1]
Why Are Examples Not Enough?
Consider this statement:
The sum of two even integers is even.
We can test examples:
2 + 4 = 6
10 + 14 = 24
100 + 8 = 108
Every example works.
But the claim is about all even integers. There are infinitely many of them.
Testing more cases can make us confident that we have found a pattern. It does not explain why no allowed case can fail.
Now Prove It Once
An even integer is an integer that can be written as two times another integer.
So let two even integers be:
a = 2m
b = 2n
Here, m and n can be any integers.
Add the two numbers:
a + b = 2m + 2n = 2(m + n)
The sum m + n is also an integer.
So a + b is two times an integer.
By the definition of even, the sum is even.
We did not test every pair. We used a description that represents every pair of even integers at once.
A good proof does not win by checking more examples. It finds the reason the examples all obey the same rule.
Figure 1. Examples, proofs, and counterexamples answer different questions.
One Counterexample Can Do the Opposite
Now consider a different claim:
All prime numbers are odd.
Try a few:
3, 5, 7, 11, 13...
The pattern looks convincing.
Then we meet:
2
Two is prime. Two is even.
The universal claim is false.
A single valid counterexample can refute a statement that claims something is true for every case. Confirming examples cannot do the reverse for an infinite collection of cases.[2]
Proof Is Not the Same as Evidence
This distinction matters.
Measurements can provide evidence. Experiments can provide evidence. Computer calculations can provide enormous amounts of evidence.
But a deductive proof has a different job.
If its accepted starting statements are true and every logical step is valid, the conclusion is forced by those starting statements. That is the basic idea of deductive validity.[3]
This does not make experiments less valuable. Mathematics and science simply ask different kinds of questions at different moments.
A Proof Can Also Explain
A proof is not only a certificate that says “true.”
A good proof often tells us why something is true.
The Berkeley notes on mathematical arguments describe a proof as an explanation that convinces other mathematicians and also helps them understand why the statement is true.[4]
In our even-number proof, the key reason was visible:
2m + 2n = 2(m + n)
The common factor of 2 survives addition.
That is the idea behind every example we tested.
What Does a Proof Actually Prove?
This is an important boundary.
A proof does not magically prove that its starting assumptions describe the physical world.
It proves that:
if we accept the starting assumptions, then the conclusion follows from them.
In Euclidean geometry, theorems follow from the Euclidean framework.
If we change a postulate, we may get a different geometry.
That story becomes important much later. For now, the key point is simpler: a proof makes the dependency visible.
Try This Mental Test
When you see a mathematical claim, ask three questions:
- What exactly is being claimed?
- What am I allowed to assume?
- Why must the conclusion follow?
If the claim says “for every,” also ask:
Can I find one counterexample?
This small habit changes how mathematics feels.
You stop asking only: “What is the answer?”
You begin asking: “What makes the answer unavoidable?”
Words to Keep
proof
A chain of justified reasoning that shows why a mathematical conclusion follows.
universal statement
A statement that claims something is true for every allowed member of a set.
counterexample
One allowed case that makes a universal claim false.
deduction
Reasoning in which the conclusion follows logically from the premises.
premise
A statement used as a starting point in an argument.
One Sentence to Keep
A mathematical proof shows why a conclusion must follow for every case covered by the claim, not just for the examples we happened to test.
What Should We Ask Next?
Proof depends on clear starting words.
Euclid's first definition begins with something that seems almost too simple:
a point
But what is a mathematical point if it has no size?
That is our next question:
What Is a Point If It Has No Size?
Previous: Why Did Euclid Build Geometry from Definitions and Proofs?
Sources & Further Reading
- David E. Joyce, Euclid's Elements, Book I — explains how Euclid's propositions are justified from definitions, postulates, common notions, and earlier propositions.
- University of Texas at Austin, “Proof by Counterexample” — explains why examples do not prove a universal claim and why one counterexample can refute one.
- OpenStax, “Types of Inferences” — introduction to deductive validity and counterexamples to invalid deductive arguments.
- UC Berkeley Mathematics, “Introduction to Mathematical Arguments” — describes proof as an explanation that convinces and helps explain why a statement is true.