Generating Triples and Exploring Higher Powers

Enlarging the triangle with sides 3, 4, and 5 gives many new lengths, but all of those triangles belong to one scaled family. How might you discover a genuinely different starting triple, such as 5, 12, and 13? Instead of searching through every possible triangle, look at a familiar number pattern: the way square numbers grow when you add successive odd numbers.

A square can be enlarged by attaching an L-shaped border of dots. The border contains an odd number of dots. Occasionally that odd number is itself a perfect square. At that moment, the growing-square picture becomes an equation in which two squares add to another square.

We will turn this observation into a method, explain which triples it produces, and then consider a related question about cubes and higher powers. Distinguish what you have proved from what you have merely noticed in examples.

The Sum of Consecutive Odd Numbers

Begin with one dot. Adding 3 dots makes a 2-by-2 square; adding the next 5 makes a 3-by-3 square. The successive totals are 1, 4, 9, 16, and so on. These are square numbers.

To grow a square of side n − 1 into a square of side n, add a new row of n dots and a new column of n − 1 further dots. The corner dot has already been counted in the row. The total added is n + (n − 1) = 2n − 1, the nth odd number.

The existing square contains (n − 1)² dots. Adding the border of 2n − 1 dots gives n² dots. This explains the identity (n − 1)² + (2n − 1) = n². It also explains why the sum of the first n odd numbers is n².

The Difference Between Consecutive Squares

Subtract (n − 1)² from the growing-square equation. You obtain n² − (n − 1)² = 2n − 1. So the difference between consecutive squares is always an odd number.

You can check this algebraically. Expanding (n − 1)² gives n² − 2n + 1. Subtracting it from n² leaves 2n − 1. The dot picture and the algebra describe the same relationship in different forms.

Usually the added odd number is not a square: 7, for instance, is not a square. But 9, 25, 49, and 81 are both odd and square. These are the useful border sizes for generating triples.

Choosing an Odd Number That Is Also a Square

Choose the odd square 9 = 3². It is the fifth odd number, since 2 × 5 − 1 = 9. The growing-square identity with n = 5 gives 4² + 9 = 5². Replacing 9 by 3² gives 4² + 3² = 5².

Now choose 25 = 5². It is the thirteenth odd number because 2 × 13 − 1 = 25. The same identity gives 12² + 25 = 13², hence 12² + 5² = 13². We have generated a new primitive triple rather than scaled the original one.

The steps are: choose an odd square greater than 1, find its position among the odd numbers, and use that position for n. The resulting triangle has perpendicular sides equal to the chosen odd square’s root and n − 1, with hypotenuse n.

Generating New Triples

For any odd positive integer m greater than 1, choose the border m². Solving 2n − 1 = m² gives n = (m² + 1)/2. The preceding side is n − 1 = (m² − 1)/2. Both are integers because an odd square plus or minus one is even.

You may use this expression as a shortcut after understanding the growing-square reasoning. Excluding m = 1 is necessary: it would give a zero side, and a triangle cannot have a side of length zero.

Why These Triples Are Primitive

Every generated triple contains n − 1 and n. Consecutive integers have no common factor greater than 1. If a number divided both, it would divide their difference, which is 1.

Any factor common to all three entries would have to divide these two consecutive entries. That is impossible unless the factor is 1. Therefore every triple produced by this method, with an odd m greater than 1, is primitive.

The chosen odd number m need not be prime. For m = 9, the triple (9, 40, 41) is primitive even though 9 is composite. Primitiveness concerns a shared factor of the complete triple.

What This Method Misses

A method that produces only primitive triples need not produce all primitive triples. Look at (8, 15, 17). It satisfies the theorem, and its entries have greatest common factor 1. Yet neither perpendicular side is one less than the hypotenuse: 17 − 15 = 2 and 17 − 8 = 9.

Our method always produces a side n − 1 and a hypotenuse n. The triple (8, 15, 17) does not have that pattern, so it cannot come from this particular construction. This is a counterexample to the claim that the method finds every primitive triple.

Nevertheless, there are infinitely many odd integers greater than 1, and their values of (m² + 1)/2 keep increasing. Thus this construction does produce infinitely many distinct primitive triples, even though some other primitive triples are missed.

A Long-Standing Open Problem

The equation x² + y² = z² has many positive-integer solutions. Fermat asked what happens if the power 2 is replaced by 3, 4, or a higher integer. Can two positive perfect cubes add to another positive perfect cube? Can two positive fourth powers add to a fourth power?

Trying the entries 3, 4, and 5 gives 3³ + 4³ = 27 + 64 = 91, whereas 5³ = 125. So a square-power triple does not automatically work for cubes. However, the failure of one example does not establish the failure of every possible example.

Fermat proposed the much stronger statement that xⁿ + yⁿ = zⁿ has no solutions in positive integers x, y, and z when n is an integer greater than 2. This is known as Fermat’s Last Theorem. A note he left claimed a proof, but no such proof by him was found. The problem resisted attempts for more than 300 years.

Andrew Wiles encountered the problem as a ten-year-old and later devoted years to it. His proof was completed in 1994. The problem is no longer open. Its proof uses mathematics far beyond what is needed here; the lesson is to understand the question and the difference between checking examples and proving a general statement.