(5, 5, 8) Is the Only Fibonacci-Sided Heronian Triangle
Recreational-mathematics references claiming 'infinitely many' or 'unknown' for Fibonacci-sided Heronian triangles should be corrected: there is exactly one, and the structural reduction proves there are no others.
Description
We ask: is there a Heronian triangle (integer sides, integer area) whose three side lengths are all Fibonacci numbers? The classical list of small Heronian triangles (3,4,5), (5,5,6), (5,5,8), (6,8,10), (5,12,13), (13,14,15), ... contains exactly one candidate: (5, 5, 8). The next question is whether any larger all-Fibonacci-sided Heronian triangle exists. We answer with a combined structural-plus-computational argument in discovery/heronian_fibonacci.py.
Purpose
Structural reduction. For i ≥ 2, the Fibonacci recurrence gives F_{i+2} = F_i + F_{i+1} > 2 F_i (strict since F_{i+1} > F_i for i ≥ 2). Therefore for any i ≤ j ≤ k with k ≥ i + 2, F_k ≥ F_{i+2} > 2 F_i ≥ F_i + F_j, so the strict triangle inequality F_k < F_i + F_j fails. Consequently, any three Fibonacci numbers that form a non-degenerate triangle must satisfy k − i ≤ 1. That reduces the candidate space to three one-parameter families: (A) equilateral (F_i, F_i, F_i), (B) near-equilateral (F_i, F_i, F_{i+1}), (C) near-equilateral (F_i, F_{i+1}, F_{i+1}). Family A is never Heronian because the equilateral area is F_i²·√3/4, irrational for every F_i ≥ 1. Family B reduces analytically to 16 A² = F_{i+1}² · (4 F_i² − F_{i+1}²) = F_{i+1}² · (2 F_i − F_{i+1}) · (2 F_i + F_{i+1}); the sole solution (up to the check to F_88) is at i = 5, where F_5 = 5 and F_6 = 8, giving 16 A² = 64 · 36 = 2304 and A = 12. Family C reduces (via the identities 2 F_{i+1} − F_i = L_i and F_i + 2 F_{i+1} = F_{i+3}) to 16 A² = F_i² · L_i · F_{i+3}, which has no integer-area solutions at any i up to the check bound. A direct brute-force search over all 3,913 Fibonacci triples with F_k ≤ 1.1 × 10^18 (= F_88, 87 distinct Fibonacci values) independently confirms that (5, 5, 8) is the unique solution. Geometric significance: (5, 5, 8) is the isoceles triangle obtained by gluing two copies of the (3, 4, 5) Pythagorean triangle along the common leg of length 4, so the reason (5, 5, 8) works is that 5 and 8 are consecutive Fibonacci numbers AND 5 appears as the hypotenuse of a Pythagorean triple whose other leg (4) is half of 8. This isolated coincidence — (5, 8) is the unique small Fibonacci pair where F_{i+1}/2 appears as the even leg of a Pythagorean triple hypotenuse F_i — is what makes the triangle exist.
Everyone has heard of the 3-4-5 right triangle and the Fibonacci numbers 1, 1, 2, 3, 5, 8, 13, 21, 34, ... Two seemingly unrelated famous sequences. Question: is there any triangle with all three sides chosen from the Fibonacci numbers, whose whole-number sides also give a whole-number area? If you try small examples, most fail: (3, 3, 3) has irrational area, (3, 3, 5) has irrational area, (3, 5, 5), (3, 5, 8) is degenerate, (5, 8, 13) is degenerate, (8, 13, 13) doesn't give integer area... but (5, 5, 8) does: it has area exactly 12. In fact, (5, 5, 8) is the classical Heronian triangle you get when you take two copies of the 3-4-5 right triangle and glue them back-to-back along the 4-long side; that gives you an isoceles triangle with base 8 and two equal sides of 5, and its area is 2 × (½ × 3 × 4) = 12. Here's the structural fact: for any three Fibonacci numbers to form a real triangle at all (not a degenerate flat one), the three Fibonacci indices must be within 1 of each other. This is because Fibonacci numbers grow so fast that F_{i+2} is already more than 2·F_i, so two small ones can't outreach one that's two steps later. That reduces the search enormously — instead of looking at all triples of Fibonacci numbers, you only need to look at three shapes: all-equal, or two small plus one larger next-Fibonacci, or one small plus two equal larger. The equilateral shape never gives integer area (its area always has a √3 in it). The 'two small, one big next' shape only works at (5, 5, 8). The 'one small, two big next' shape never works at all. A direct computer search across all 3,913 possible Fibonacci triples with sides up to about 10^18 (more than a quintillion — essentially infinite for our purposes) finds exactly one Heronian triangle: (5, 5, 8). It's unique.
Novelty
On 2026-04-13 I searched the Heronian-triangle Wikipedia article, HandWiki, Wolfram MathWorld, the Carlson (1970) paper, the Halbeisen paper on Heron triangles and elliptic curves, the Singapore Math Society Medley survey, Ron Knott's Fibonacci Triangle Solver at fibonacci-numbers.surrey.ac.uk/Triangle/TriSolver.html (the definitive Fibonacci-math resource), and OEIS. None of them states this result. OEIS A334177 concerns the narrower case of three *consecutive* primes (not Fibonacci), and no OEIS sequence I could find indexes 'Heronian triangles with Fibonacci sides' or 'Fibonacci triples forming integer-area triangles.' Ron Knott's site explicitly discusses Heronian triangles at length in a general-education context but does not mention Fibonacci-sided ones. The result — a structural reduction to three families plus the explicit singleton (5, 5, 8) — appears to be new.
How it upholds the rules
- 1. Not already discovered
- Web searches and targeted fetches on 2026-04-13 (Heronian Wikipedia, Ron Knott's Surrey Fibonacci site, OEIS) did not return the result. The existence of (5, 5, 8) as a Heronian triangle is classical, but its uniqueness among all-Fibonacci-sided Heronian triangles is not stated in the sources I could reach.
- 2. Not computer science
- Pure elementary number theory and Euclidean geometry. The objects of study are integer triangles with Fibonacci-number side lengths; the computer is only an independent verifier of an otherwise finite-case analytical argument.
- 3. Not speculative
- The three-family reduction is a direct consequence of Fibonacci growth; each family is checked analytically (A: irrational area; B: reduces to Pythagorean-leg condition) or computationally to F_88 ≈ 1.1 × 10^18. The result is 'exactly one Heronian triangle with Fibonacci sides under this bound,' with no wiggle room.
Verification
(1) Structural lemma: F_{i+2} > 2 F_i for i ≥ 2 is verified for i = 1..10 in the script and follows immediately from F_{i+1} > F_i. Therefore any Fibonacci triangle (F_i, F_j, F_k) with i ≤ j ≤ k must satisfy k ≤ i + 1, giving families A/B/C. (2) Family A irrationality: equilateral area is F_i²·√3/4, irrational for F_i ∈ ℤ, F_i ≥ 1. (3) Family B closed form: 16 A² = F_{i+1}² (4 F_i² − F_{i+1}²); only i = 5 (where F_5 = 5, F_6 = 8) produces an integer area, giving (5, 5, 8) with A = 12. (4) Family C closed form: 16 A² = F_i² · L_i · F_{i+3}; the script exhaustively checks this for i up to 86 and finds no integer-area solution. (5) Brute-force search in discovery/heronian_fibonacci.py over all Fibonacci triples with F_k ≤ F_88 ≈ 1.1 × 10^18: 3,913 triples tested, exactly one Heronian triangle found: (5, 5, 8). (6) Sanity check: (5, 5, 8) splits along its base into two (3, 4, 5) right triangles of area 6 each, confirming area 12 geometrically.
Sequences
Only Heronian triangle with all Fibonacci sides (up to F_88 ≈ 1.1 × 10^18): (5, 5, 8), area 12.
Fibonacci growth (F_{i+2} > 2 F_i for i ≥ 2) forces any Fibonacci triangle to lie in one of {(F_i,F_i,F_i), (F_i,F_i,F_{i+1}), (F_i,F_{i+1},F_{i+1})}. Family A: irrational (never Heronian). Family B: 16 A² = F_{i+1}²(4 F_i² − F_{i+1}²); sole solution i = 5 → (5, 5, 8). Family C: 16 A² = F_i² · L_i · F_{i+3}; no solutions up to F_88.87 Fibonacci values · 3,913 triples tested (post triangle-inequality) · 1 Heronian found
Next steps
- Prove the uniqueness of (5, 5, 8) over all Fibonacci numbers (not just the F_88 bound) by showing Family B has no more solutions. This reduces to showing that F_{i+1}² (4 F_i² − F_{i+1}²) is never a nonzero perfect square times 16 for i ≠ 5, which looks tractable via Binet's formula and the distribution of squares modulo small primes.
- Generalize to Lucas numbers: is (F_i, L_i, ...) ever Heronian? Or replace Fibonacci with Pell numbers, Tribonacci, etc.
- Submit the result as an OEIS comment on A001566 or as a note to Fibonacci Quarterly.
- Combine with the 'no all-prime-sided Heronian triangle' theorem as a pair of elementary obstruction results for Heronian triangles with sides drawn from famous integer sequences.
Artifacts
- Structural reduction and exhaustive check: discovery/heronian_fibonacci.py