Calculus that gives two answers
For two centuries calculus delivered correct answers about planets, heat, and bridges, so few people worried about its foundations. But by the early 1800s the cracks were too loud to ignore. The most disturbing one: an infinite sum can change its value when you simply reorder the terms. Addition is supposed to not care about order — yet here it does. Something in our handling of infinity is being assumed without proof.
The alternating harmonic series, summed in order: S = 1 - 1/2 + 1/3 - 1/4 + 1/5 - 1/6 + ... = ln 2 (about 0.693) Now REORDER the SAME terms: one positive, then two negatives. 1 - 1/2 - 1/4 + 1/3 - 1/6 - 1/8 + 1/5 - 1/10 - 1/12 + ... Group each triple: (1 - 1/2) - 1/4 + (1/3 - 1/6) - 1/8 + (1/5 - 1/10) - 1/12 + ... = 1/2 - 1/4 + 1/6 - 1/8 + 1/10 - 1/12 + ... = (1/2)(1 - 1/2 + 1/3 - 1/4 + ...) = (1/2) ln 2 (about 0.347) SAME terms, SAME values, just a different order -> HALF the sum. No step above is illegal in finite arithmetic. So WHICH step silently used a property of infinity that we never justified?
The lesson is not that addition broke. It is that “add up infinitely many things” is not the same operation as adding finitely many things, and we had been treating them as if they were. The rearrangement theorem later explains exactly when reordering is safe (only for absolutely convergent series) and when it is catastrophic (for merely conditionally convergent ones). But to even state that theorem, we first need a precise meaning for an infinite sum.
Infinitesimals: useful, but undefined
Newton and Leibniz computed derivatives with an infinitesimal dx: a quantity small enough to drop at the end, yet nonzero enough to divide by in the middle. Bishop Berkeley mocked these as “the ghosts of departed quantities” — and he had a point. A number that is both zero and not-zero is a contradiction. The calculations gave right answers, but the justification was a sleight of hand.
Functions that refuse to behave
The final shock was that pictures lie. Everyone “knew” a continuous curve must be smooth except at a few corners. Then Weierstrass exhibited a function that is continuous everywhere yet has a corner at every single point — a curve you can draw without lifting your pen but that has no tangent anywhere. You cannot sketch it; intuition has no foothold. This is a pathological example, and its job is exactly to break a belief we never checked.
Once a single counterexample exists, a vague belief is dead and only a careful definition can survive. Analysis is the discipline that grew from cleaning up after these shocks: it gives infinity, limits, continuity, and the real numbers themselves definitions precise enough that no rearrangement, no ghost, no pathological curve can sneak past. The rest of this track builds those definitions, one honest brick at a time.