Imagine a woodchipper that throws a chip each time it passes a number, and each chip is smaller than the last. Feed it, and it spits: a chip, a smaller chip, a smaller one still, the chips fading toward dust. Now picture three of these machines, side by side, looking exactly alike. The only difference is hidden inside — which numbers each one chips on. That single hidden difference decides whether the machine ever stops, and, for the two that never stop, whether stopping was ever even close.
The first machine fills a bag
The first chipper throws chips of one-half, then one-quarter, then one-eighth — each exactly half the one before. It runs, the chips shrink, and the pile climbs: a half, then three-quarters, then seven-eighths… creeping toward a single full bag and no further. ½ + ¼ + ⅛ + … = 1. The chipper finishes. The total is bounded, tidy, done.
This is the ordinary resolution of an old worry. Zeno said: to cross a room you must first cross half, then half the rest, then half of that — infinitely many steps, so you can never arrive. The answer is this first machine. Infinitely many shrinking steps can add to a finite total. The bag fills. You arrive. The shrinking saves you.
So a rule suggests itself, and it feels safe: if the chips shrink toward nothing, the pile must be finite. Hold onto that rule. The next two machines break it — and only one of them will surprise you.
The second machine is fed everything
The second chipper is fed every number, in order — a chip of one at 1, one-half at 2, one-third at 3, one-quarter at 4, and on forever: 1 + ½ + ⅓ + ¼ + … The chips shrink, just like the first machine. By the safe rule, the pile should be finite.
It is not. This pile grows without end. It is the harmonic series, and its divergence was shown by Nicole Oresme around the year 1350 — the chips fade, but the total climbs past every ceiling you name. The safe rule is already broken. But notice why it breaks here, because it will matter: this machine is fed everything. Every number on the line, no gaps, an unbroken flood of wood. Of course a machine fed all of arithmetic never runs dry. Its endlessness is real — and it is not a surprise. It runs forever because it is never, for a single moment, starved.
The third machine is fed almost nothing
The third chipper is fed only the primes — a chip of one-half at 2, one-third at 3, one-fifth at 5, one-seventh at 7, one-eleventh at 11, and nothing in between. It skips 4, 6, 8, 9, 10; and it skips more and more as it climbs. The primes thin out: by the time the machine is working in the large numbers it passes ten composites, then fifty, then a hundred, to find a single prime to chip. It is being fed a sliver of a sliver — a diet approaching starvation, the wood arriving rarer and rarer the longer it runs.
It never stops either. ½ + ⅓ + ⅕ + ⅐ + … climbs without end — and this is the astonishment, proved by Euler in 1737. The second machine ran forever on a feast; anyone would expect that. The third runs forever on famine. You threw away nearly every number, kept only the thinning primes, fed the machine less and less — and the pile still never fills. The same fate as the glutton, reached on a starvation diet.
| machine | fed | the pile | why |
|---|---|---|---|
| halves | ½, ¼, ⅛ … | fills the bag (= 1) | shrinks fast — converges |
| every number | 1, ½, ⅓, ¼ … | runs forever | fed everything · Oresme, ~1350 |
| the primes | ½, ⅓, ⅕, ⅐ … | runs forever | fed almost nothing, and still · Euler, 1737 |
Why the famine is the miracle
The contrast is the whole story. Feeding a machine every number and watching it never stop tells you little — it is gorged, and endlessness is the obvious outcome of an endless meal. That is why Oresme's harmonic result, true and clever as it is, sits closer to a footnote. The primes are a different matter. They are a vanishing fraction of the numbers: their density falls toward zero, so the wood the third machine receives dwindles toward nothing as it climbs. By every intuition it should be the one to starve, fill its small bag, and stop. It is starving — and it does not stop. That refusal, on so little, is Euler's miracle, and it is why his is the name people remember.
So the safe rule — shrinking chips mean a finite pile — fails twice, but the two failures are not equal. One machine breaks it the easy way, by being handed everything. The other breaks it the hard way, by being handed almost nothing and persisting anyway. The primes are spaced at exactly the knife-edge: thin enough that you would bet on them stopping, dense enough that they never do.
One machine never stops because it is fed everything. The other never stops on a starvation diet. Only the second is a miracle — and the primes are the second.
Underneath all three, the same old guarantee keeps every machine turning: the wood never runs out, because the primes never run out — Euclid proved that around 300 BC. Hand the third machine every prime you can name, and a larger one is always waiting past the pile. So it runs on, fainter and fainter, the chips approaching dust, the diet approaching nothing — and the pile climbing, slowest of all the slow infinities, forever. The glutton's endlessness you can explain. The starving machine's endlessness you can only admire.