Octonionic Layer

S^7 is where the number systems run out of room. Real numbers, complex numbers, and quaternions each kept enough good behavior to build the lower spheres. The octonions — eight-dimensional numbers — are the last step, and they pay for it by giving up a rule we have used without thinking the whole way up: associativity.

Lesson Goal

Understand what the octonions keep and what they lose, and why losing associativity means quaternion proofs cannot simply be copied up to S^7.

The Idea

Each step doubles the dimension and drops a privilege (this doubling is the Cayley–Dickson construction):

ℝ  (1D):  ordered
ℂ  (2D):  loses ordering
ℍ  (4D):  loses commutativity   (a·b ≠ b·a)
𝕆  (8D):  loses associativity   ((a·b)·c ≠ a·(b·c))

For quaternions, you could still regroup a product freely: (a·b)·c always equalled a·(b·c). For octonions that fails — the parentheses genuinely matter. The unit octonions still sit on S^7, and they are still closed under multiplication and have inverses, but they are not a group, precisely because associativity is gone.

The Euler side stays calm, though: S^7 is odd-dimensional, so by the same parity rule as before, χ(S^7) = 0.

Worked Example

Two checks, one safe and one cautionary.

The safe, numeric one — Euler characteristic by parity:

S^6 :  2
S^7 :  2 − 2 = 0     (odd sphere)

The cautionary one — do not assume regrouping is legal. In quaternions you may write (i·j)·k = i·(j·k). In octonions there exist basis triples where the two sides differ by a sign, so a step that silently drops parentheses can flip an answer. The bounded Cayley–Dickson model in the cards below is exactly where that non-associativity is pinned down and checked.

Common Mistake

The biggest trap on this page is importing a quaternion or group argument by habit. Only the theorem-card claims — the bounded Cayley–Dickson coordinate facts, including the explicit non-associativity — are checked; the fuller octonionic topology is future work. See What “Proved” Means Here.

Checkpoint

A proof for S^3 rewrites (a·b)·c as a·(b·c) in a key step. Can you reuse it unchanged on S^7? Why or why not? (Answer: no — octonion multiplication is non-associative, so that rewrite is not valid.)

Source Trail

Paper source: Octonionic Units And Nonassociative Coils