Arc Length and Surface Area

Measuring a curve without straightening it

Distance along a curved path is not usually the horizontal change or the vertical change. A hiking trail, a bent wire, and the rim of a shaped vessel each demand the distance traveled along the curve. Calculus finds that distance by replacing a short curved piece with a straight segment, measuring the segment, and then shrinking the pieces. The formula is therefore a consequence of the Pythagorean theorem applied locally. It is not a new rule to memorize separately from the picture that creates it.

Let a smooth graph be given by $y=f(x)$ from $x=a$ to $x=b$. Over a small interval, its horizontal change is $dx$ and its vertical change is approximately $dy=f'(x)\,dx$. The short straight segment joining the nearby points has length $ds$. The Pythagorean theorem gives

$$ds=\sqrt{(dx)^2+(dy)^2} =\sqrt{1+\left[f'(x)\right]^2}\,dx.$$

The square root is essential. It combines the horizontal and vertical changes as perpendicular legs rather than adding them directly. A steep graph has a large $|f'(x)|$, so its arc-length contribution is greater than $dx$ alone; a horizontal graph has derivative zero, so its contribution reduces to $dx$. These limiting cases are quick checks on the formula before a calculation starts.

From local segments to an integral

Adding the segment lengths from $a$ to $b$ produces the arc-length formula

$$L=\int_a^b\sqrt{1+\left[f'(x)\right]^2}\,dx.$$

Here $L$ is a total length, while the integrand is length gained per horizontal unit. The derivative belongs in the formula because it tells how quickly the height is changing as the input moves. If the curve is instead described as $x=g(y)$, it can be more natural to use horizontal changes in $x$ and integrate with respect to $y$:

$$L=\int_c^d\sqrt{1+\left[g'(y)\right]^2}\,dy.$$

Neither form is more fundamental. Choose the one that makes the curve, derivative, and bounds easier to describe. In either form, check that the variable in the derivative matches the differential at the end. Writing $dx$ after differentiating with respect to $y$ is not a harmless typo; it mixes two different local descriptions.

For a concrete example, take $f(x)=\frac{x^2}{2}$ from $x=0$ to $x=1$. Then $f'(x)=x$, and the curve length is

$$L=\int_0^1\sqrt{1+x^2}\,dx.$$

This integral does have an antiderivative involving a logarithm, but the geometric setup is more important than the particular evaluation. The graph’s horizontal displacement is $1$ unit, while its curved length must be greater than $1$ unit because the path rises as well as moves right. A numerical estimate near $1.15$ units is therefore plausible. An answer smaller than $1$ would contradict the drawing.

Rotating a curve creates bands

Surface area of revolution uses the same short segment $ds$, but each segment sweeps out a narrow band. When the graph $y=f(x)$ is rotated about the $x$-axis, the band has approximate circumference $2\pi f(x)$ and slant width $ds$. Its small area is therefore

$$dS=2\pi f(x)\sqrt{1+\left[f'(x)\right]^2}\,dx.$$

Adding the bands gives

$$S=2\pi\int_a^b f(x)\sqrt{1+\left[f'(x)\right]^2}\,dx,$$

provided $f(x)$ is the nonnegative distance to the axis. When an axis is shifted, replace $f(x)$ with the appropriate distance. For rotation about $y=k$, the radius is $|f(x)-k|$, not simply $f(x)$. This is the same distance discipline used in washer and shell problems, now applied to the circumference of a moving band.

The phrase “surface area” can conceal an important difference from volume. A volume slice has a cross-sectional area times thickness. A surface band has a circumference times slant length. Replacing $ds$ by $dx$ would treat a tilted band as though it were flat and will underestimate its area unless the curve is horizontal. The square-root factor is the record of that tilt.

A reliable setup routine

Start every problem by identifying the object to be measured: a path length or a surface created by rotation. Sketch the curve and mark the interval. Differentiate before substituting into the square root so that the role of the slope stays visible. For surface area, separately identify the distance to the axis; that distance supplies the circumference factor.

Then run three checks. The integrand for arc length must be nonnegative, because lengths do not cancel. Its units must be length after multiplication by the differential. Finally, compare with a simple lower bound: the arc is at least as long as the straight-line distance between its endpoints, and a nonflat surface band must have more area than its horizontal projection suggests. These checks do not replace computation, but they expose a wrong model early.

For retrieval, the curve $y=\frac{x^2}{2}$ on $0\le x\le1$ is rotated about the $y$-axis. Decide whether to express the surface-area integral using $x$ or solve for $x$ in terms of $y$. State the radius, the relevant arc-length element, and why your choice keeps the setup simple. Do not evaluate until the small band can be explained in words. The next lesson turns the same “small contribution plus accumulation” idea toward work, fluid force, and variable-density models.