A reciprocal for nothing — and why it isn't infinity
Everyone learns, early and firmly, that you cannot divide by zero. It is stated like a law of nature. It is not one. It is a theorem — and its proof quietly leans on the very rule we threw out in the last chapter. Remove that support, and the reciprocal of zero simply falls into the hole it left.
"You can't divide by zero" is really the claim that zero has no reciprocal — no partner r with 0·r = 1. There is a short, honest proof of it, and it is worth seeing exactly once:
So no reciprocal can exist. But look at the load-bearing step: 0·r = 0. That is absorption — the rule that lets zero eat whatever it touches — and in the last chapter we found it was never a law, just an accident of bundling three jobs onto one symbol. Take absorption away and the proof has nothing to stand on. The impossibility was never in zero; it was in the axiom. Remove the axiom and the reciprocal is not added — it was implied all along, and merely forbidden.
So we give it a name.
Read those two lines with the discipline from the last chapter. The := introduces a new object — ω, the reciprocal of zero — and the = is the single rule that pins it down: multiplying it by zero must give back 1, because that is what "reciprocal" means. Nothing else is assumed about ω. Everything else about it will be forced.
Now the crucial fork, and the single most important distinction in this whole book. You have almost certainly seen "1/0 = ∞" written somewhere. That is not what ω is.
Infinity is a statement about size. It is where a quantity goes as it grows without any bound — the endpoint of a process you approach but never arrive at. It is analytic: it lives at the far end of a limit.
ω is not a size at all. It is a direction. It is the exact algebraic partner of zero, fixed in one clean line — 0·ω = 1 — with no process, no growing, no limit anywhere. You do not travel toward ω; you name it, and it is simply there. The clearest tell is what happens when you turn the crank twice: infinity, doubled, is still infinity, still running off the end of the line — but ω comes back around.
The old "1/0 = ∞" mistook one for the other. It took a zero whose direction had already been thrown away — leaving only a vanishing size — and then, quite reasonably, watched that size blow up when inverted. Invert the whole thing, direction and all, and you do not get a blow-up. You get ω: bounded, exact, and pointing the opposite way.1 This line — a symbolic direction on one side, an analytic size on the other — is the seam the rest of the book is built along.
If ω keeps the direction, where did the size go? Nowhere — we just have to give it a shelf to sit on. That shelf is the grade: a single integer attached to every value, recording how many factors of zero (or of ω) it carries.
The rule for grades is the one you would guess. Multiplying two values adds their grades; taking a reciprocal flips the sign. And now the defining law reads as simple bookkeeping: 0·ω multiplies a grade +1 by a grade −1, the two cancel to grade 0, and what's left is an ordinary 1. Zero and ω are not a small thing and a huge thing. They are the same thing at opposite grades — one step below the ordinary numbers, and one step above.
If 0·ω = 1, an honest reader immediately asks the next thing: then what is 0 times an ordinary number, like 0·5? In old arithmetic the answer was "0, obviously" — but that answer was absorption, and it is gone.
The consistent answer is that 0·5 is not a collapse to nothing. It is 5 held one grade down — a genuine value, written 0·5 and kept as such, not erased. It reads as zero only to an instrument that can't see below grade 0; the 5 is still on its shelf, recoverable by multiplying back up with ω. The full arithmetic of these graded coefficients is the business of a later chapter;2 here the point is only that nothing was destroyed — the size was filed, not burned.
One consequence is worth stating plainly, because it is where division becomes total. With no absorption in the way, cancellation is now unconditional:
That 0/0 = 1 is a real choice, not a discovery, and it is worth flagging as one.3 It follows the moment you insist that x/x = 1 hold with no exceptions, but "with no exceptions" is the thing we decided to want. Other systems make other choices at exactly this fork; we take the one that keeps every operation total and every cancellation reversible, because that is the whole premise.
Division is now total: there is no input it refuses, no gap to route around. And a number is no longer just a size on a line — it is a size together with a grade, a coefficient with a place to stand. We have one new object, ω, pinned by one rule, and a shelf to keep the bookkeeping straight. The natural next question is what this zero does when you start raising things to its powers — and the answer, surprisingly, is where the number −1 comes from.