What does the fundamental theorem of calculus actually say?

That the total change in a quantity equals the accumulation of its rate of change. If F is a quantity and f is how fast it changes, then the integral of f from a to b is F(b) minus F(a). In the tank picture: the water the syringe delivered between two moments is exactly the final level minus the starting level.

Why are differentiation and integration inverse operations?

Because they read the same machine in opposite directions. Differentiating a level gives the rate it is rising; integrating a rate gives the level it has built up. Start from a level, take its rate, add the rate back up, and you are back to the level you started from, up to where it began.

What is the +C, the constant of integration?

It is the water that was already in the tank before you started measuring. A rate tells you how the level changes, never where it began, so an indefinite integral recovers the level only up to an unknown starting amount. That unknown amount is the +C, and in a definite integral it cancels, because you subtract the starting level from the final one.

Why is a definite integral the area under a curve?

Plot the rate against time. Over a thin slice of time the water added is rate times width, which is the area of a thin rectangle under the rate curve. Add every slice and the total water delivered is the area under the whole curve. So the area is not a separate meaning of the integral; it is the accumulated rate, drawn.

Essay · Integrals

The fundamental theorem is one apparatus, read twice

why differentiating and integrating undo each other — the syringe, the tank, and the water in between

Updated September 2026

A syringe filling a tank is one machine. The water standing in the tank is the integral; how fast the level is rising at any instant is the derivative. The fundamental theorem is not a bridge built between two subjects — it is the flat observation that the final level minus the starting level equals the water the syringe delivered, and everything else is that sentence in symbols.

The theorem is usually presented as the punchline of a first calculus course, a near-miracle that ties the derivative to the integral after each has been developed on its own for weeks. That framing is what makes it feel deep and arbitrary at once. Put both on the same bench and the mystery leaves: they were never two things.

The machine

Picture a tank and a syringe pushing water into it. Two numbers describe what is happening. One is the level — how much water is in the tank right now. The other is the rate — how fast that level is climbing, which is set by the cross-section of the piston at that instant. Push a fat piston and the level jumps; push a thin one and it creeps.

Call the level F(t) and the rate f(t). These are not independent quantities you happen to be tracking together. The rate is the derivative of the level: f = F′. And the level is the accumulated rate: to know how much water is in the tank, add up everything the piston has delivered. That is the whole relationship, and it is already symmetric.

Reading it forward: differentiation

Start with the level and ask how fast it is rising. That is the nudge ratio from chapter 2: let a sliver of time dt pass, watch the water rise by dF, and the rate is dF/dt. Differentiating the level hands you the piston's cross-section at that moment. Nothing about area, nothing about sums — just the slope of the level over time.

Reading it backward: integration

Now start with the rate and ask for the level. Over a thin slice of time the water added is the rate times the width of the slice, f(t) · dt. Line the slices up and add them:

total water = ∫ f(t) dt

That sum is the integral, and it is also, drawn on a graph of the rate against time, the area under the rate curve — because each term f(t)·dt is the area of a thin rectangle, height times width. The area is not a second, unrelated definition of the integral you have to reconcile later. It is the accumulated rate, seen sideways.

The plus C is the water that was already there

A rate tells you how the level is changing, and says nothing about where it started. Two tanks fed by the identical syringe, one begun half full and one begun empty, have the same rate at every instant and different levels forever. So recovering the level from the rate can only ever get you there up to an unknown starting amount:

F(t) = ∫ f(t) dt + C

The C is the water already in the tank at time zero. It looks like a technicality bolted onto an answer; it is the one piece of information a rate physically cannot carry. This is also why it disappears the moment you ask a question about change rather than about level.

Why the two readings meet

Ask for the water delivered between time a and time b. Read backward, it is the integral of the rate over that stretch. Read forward, it is simply the level at the end minus the level at the start, because that difference is the water that arrived. Those two are the same quantity, so

∫_a^b f(t) dt = F(b) − F(a)

That is the fundamental theorem. The starting water C is gone because it stood in both F(b) and F(a) and cancelled when you subtracted. Read the left side and you are adding up rates; read the right side and you are taking one difference of levels. The theorem only says: these are two names for the water in the tank.

It feels like a coincidence in the usual telling because the two sides are computed so differently — one an infinite sum of thin areas, the other a single subtraction. The machine shows why they must land on the same number without computing either: the water that flowed in is the water that flowed in.

Where the signs come from

Draw the piston back instead of pushing it, and the level falls: the rate is negative, and the integral over that stretch is negative too, subtracting water rather than adding it. A definite integral is a signed total for exactly this reason. On the area picture, the stretches where the rate dips below zero count as area below the axis, which is the same fact wearing its other costume.

The average rate of change

One more part falls out for free. Divide the water delivered by the time it took:

[F(b) − F(a)] / (b − a)

That is the single steady cross-section — one unchanging piston — that would have carried the same water in the same time. It is the average of the rate over the interval, and it is why the average rate of change and the slope of the secant line are the same number: both are the total climb spread evenly across the run.

The same machine in economics

This chapter closes where the economics series opens, because the machine is already standing there under different labels. Let output Q play the role of time. Then the marginal cost MC(Q) is the rate — the cost of the next unit, the piston's cross-section — and the total cost TC(Q) is the level, the water in the tank:

TC(Q) = ∫₀^Q MC(q) dq + FC

The fixed cost is the plus C: the cost already on the books at zero output, the water in the tank before the first unit is made. Producing nothing still costs it. So "total cost is the area under the marginal cost curve, plus fixed cost" is not a formula to memorise; it is the fundamental theorem with MC written where f was. The same reading turns total revenue into the area under marginal revenue, and total utility into the area under marginal utility — every one of them a level recovered from its rate.

Watch it fill

The apparatus is easier to trust in motion than on paper. The chapter animates the tank and the syringe filling it, so you can see the level be the area and the cross-section be the slope at the same instant.