Why this matters
The area of a triangle is base times height over two; the area of a circle is the radius squared times pi. For familiar shapes like these, you just memorize a formula. But the bumpy area carved out by a curve winding this way and that has no memorized formula at all.
So how do you measure the area under a curve? The answer people found long ago was surprisingly humble — if you cannot measure it, fill it with something you can. Pack the space under the curve densely with many very thin rectangles, and add up all their areas. A rectangle's area is width times height, which you can always measure.
This one humble idea grew into the definite integral. This article follows its birth in three steps: approximating the area with finely sliced rectangles, slicing the pieces infinitely fine to drive out the error, and the moment that infinite sum hardens into a long S-shaped symbol.
Step 1 — Fill the space under the curve with rectangles
A single rectangle cannot exactly fill the space under a curved line. Its flat top pokes above or falls short of the curve. But increasing the number of pieces changes the story.
Divide the interval into \(n\) equal pieces, and on each piece stand a rectangle as wide as that piece. Here \(n\) is the number of pieces, meaning "how many did we slice it into." Each rectangle's width is the interval width divided by \(n\), a very short length that we write as \(\Delta x\). The symbol \(\Delta\) (delta) means "a small change in an amount," so read \(\Delta x\) as "a small width in the \(x\) direction." Its height is the function value at that spot, that is, the height of the curve \(f(x)\).
Then one rectangle's area is width times height, that is, \(f(x)\,\Delta x\). Adding up all \(n\) of them gives an approximation of the area under the curve.
$$S_n = f(x_1)\,\Delta x + f(x_2)\,\Delta x + \cdots + f(x_n)\,\Delta x$$This method of slicing finely to measure is called the method of exhaustion by rectangles — literally, "divide the interval and find the area."
Step 2 — Checking on an easy example whose answer we know
Does this method really approach the true area? Let us check with an example whose answer we already know. Instead of a curve, take the straight line \(y = x\) from \(x = 0\) to \(x = 1\). The space under this line is a right triangle with base \(1\) and height \(1\), so we know its area with our eyes closed.
$$\text{true area} = \frac{1}{2}\times 1 \times 1 = \frac{1}{2}$$Now let us pretend not to know about the triangle and measure it by filling with rectangles. Dividing \([0,1]\) into \(n\) pieces makes each rectangle's width \(\Delta x = \dfrac{1}{n}\). The right end of the \(k\)-th piece is \(x_k = \dfrac{k}{n}\), and the line's height there is just \(f(x_k) = \dfrac{k}{n}\). So the sum of rectangle areas is
$$S_n = \sum_{k=1}^{n} \frac{k}{n}\cdot\frac{1}{n} = \frac{1}{n^{2}}\left(1 + 2 + \cdots + n\right).$$Here \(\sum\) (sigma) is a Greek letter meaning "add up several things"; right now it means, add up as \(k\) runs from \(1\) to \(n\). And the sum from \(1\) to \(n\) is given by the well-known formula \(1 + 2 + \cdots + n = \dfrac{n(n+1)}{2}\). Plugging it in,
$$S_n = \frac{1}{n^{2}}\cdot\frac{n(n+1)}{2} = \frac{n+1}{2n} = \frac{1}{2} + \frac{1}{2n}.$$The expression cleaned up beautifully. The rectangle sum is the true area \(\dfrac{1}{2}\) with a leftover \(\dfrac{1}{2n}\) attached. That leftover is exactly the error — the amount the rectangles poke up above the triangle. Plugging in actual numbers makes the meaning sharp.
| pieces \(n\) | rectangle sum \(S_n\) | error from true area |
|---|---|---|
| \(4\) | \(0.625\) | \(0.125\) |
| \(10\) | \(0.55\) | \(0.05\) |
| \(100\) | \(0.505\) | \(0.005\) |
| \(1000\) | \(0.5005\) | \(0.0005\) |
The finer we slice, the more steadily the error \(\dfrac{1}{2n}\) shrinks. As the pieces get thinner, the pointed teeth by which the rectangles poke above the curve flatten out more and more.
Step 3 — Drive the error to zero and it becomes exact
The error does not merely shrink — slice the pieces infinitely fine and you can push the error all the way to \(0\). Handling this "approaching without limit" is the job of the limit.
$$\lim_{n\to\infty} S_n = \lim_{n\to\infty}\left(\frac{1}{2} + \frac{1}{2n}\right) = \frac{1}{2}.$$Here \(\lim\) (limit) means "gets endlessly close to this value," and \(n\to\infty\) means "increase the number of pieces without end." As the pieces grow in number, \(\dfrac{1}{2n}\) approaches \(0\), so the rectangle sum converges to exactly \(\dfrac{1}{2}\). Perfectly the same answer the triangle formula gave us.
The key point is this. Whenever an approximation's error can be driven to \(0\), its limit is no longer an approximation but the exact true value. The rectangle method always carries an error, but the limit that wipes that error out completely through infinite division is precisely the true area under the curve.
This limit value — the infinite sum of infinitely thin slices — is precisely the definite integral.
The moment sigma hardens into a long S
Now it is time to see the symbol's birth. The approximation we worked with had this form.
$$S_n = \sum_{k=1}^{n} f(x_k)\,\Delta x.$$Applying the limit of infinite division changes two things at once.
- The number of things added becomes infinite. The \(\sum\) that added a finite count cannot hold this infinite sum, so a new symbol becomes necessary.
- The width of each piece becomes infinitely small. The finite width \(\Delta x\) shrinks to an infinitely thin width, which we write as \(dx\).
So the mathematician Leibniz stretched out the letter S — the first letter of the Latin summa, meaning "sum" — into a new symbol. That is the integral sign \(\int\). The Greek letter sigma, which once meant addition, hardened like a fossil into a long S meaning the sum of infinitely thin slices.
$$\sum_{k=1}^{n} f(x_k)\,\Delta x \quad\xrightarrow{\;n\to\infty\;}\quad \int_{a}^{b} f(x)\,dx$$Each symbol still carries the trace of its approximation days.
| approximation (finite) | definite integral (infinite) | meaning |
|---|---|---|
| \(\sum\) | \(\int\) | add up (sum) |
| \(\Delta x\) | \(dx\) | piece width — finite → infinitely small |
| \(f(x_k)\) | \(f(x)\) | the height at that spot |
| \(k=1 \sim n\) | \(a \sim b\) | the interval added over |
So the expression \(\displaystyle\int_{0}^{1} x\,dx = \dfrac{1}{2}\) is nothing but our calculation compressed into symbols: "fill the space under the line \(y=x\) from \(0\) to \(1\) with infinitely thin slices and add them all, and you get \(\dfrac{1}{2}\)."
In one picture: finer pieces converge to the true area
The interactive below is a screen for filling the \([0,1]\) triangle under the line \(y = x\) with rectangles.
- As you increase the pieces slider, the rectangles thin out and the red teeth (error) poking above the curve flatten more and more.
- Check in the value at the top that the rectangle sum gets closer and closer to the true area \(0.5\). The displayed error is exactly \(\dfrac{1}{2n}\).
- Push the slider to the end and a symbol transformation appears alongside — \(\sum\) into \(\int\), \(\Delta x\) into \(dx\) — the moment infinite division gives birth to the definite integral.
Raising the piece count \(4 \to 10 \to 100\) and watching the sum narrow \(0.625 \to 0.55 \to 0.505\) makes "slice it infinitely and it becomes exact" tangible. The name and symbol we gave that infinite sum is the definite integral.
The key summary
| Question | Answer |
|---|---|
| Why has the area under a curve no formula? | Unlike a triangle or circle, a curved area has no memorized formula, so we fill it with measurable rectangles |
| Why is the rectangle sum an approximation? | Its flat top pokes above or falls short of the curve, creating an error — in our example the error is \(\dfrac{1}{2n}\) |
| How does it become exact? | Slicing the pieces infinitely (\(n\to\infty\)) drives the error to \(0\), and that limit is the true area |
| Where did \(\int\) come from? | The S of the Latin summa ("sum"), stretched long — the infinite version of sigma |
| What is \(dx\)? | The finite piece width \(\Delta x\), shrunk infinitely small |
Threaded into one line — filling the space under a curve with thin rectangles and adding them gives an approximation; the limit that drives the error to \(0\) by slicing the pieces infinitely is the true area; and the symbol given to that infinite sum is the long S-shaped definite integral.
When you study this with AI
Memorizing the definite integral as just "the formula for finding area" blurs why it is that particular symbol and why a limit is needed. When asking an AI, asking it to walk through "approximate with rectangles, then slice infinitely" makes the meaning of the symbols follow on its own.
Useful example questions:
- "Approximate the area under the line \(y=x\) on \([0,1]\) with \(n\) rectangles, set up the sum, and show step by step how the \(n\to\infty\) limit becomes \(\dfrac{1}{2}\)."
- "Explain how the integral sign \(\int\) and \(dx\) each derive from \(\sum\) and \(\Delta x\)."
- "Approximating \(y=x^{2}\) on \([0,1]\) with rectangles, compute what the sum becomes and whether the limit goes to \(\dfrac{1}{3}\)."
Closing
The integral sign looks like a strange glyph when you first meet it. But that long S is really a fossil of a very humble action: "add up." We could not measure the space under a curve, so we filled it with measurable rectangles; we wiped out the approximation's error through infinite division; and we needed a new symbol to hold the now-infinite sum, so we stretched sigma into a long S. These three steps are the whole birth of the definite integral.
Grasp the single picture of the rectangle sum narrowing to the true area as the pieces are sliced, and \(\int\) comes across not as a symbol to memorize but as a picture whose meaning — "add up all the infinitely thin slices" — is visible right in its shape.
Related reading — good to read first: Why differentiation is the instantaneous slope → next: Why integration is the reverse of differentiation · Why integrals carry a + C · draw it yourself: Desmos · Wolfram Alpha