Setting the Wrong Boundaries: Why Definite Integral Substitution Goes Wrong

Suppose you're solving:

$$\int_{x=1}^{x=2} 2x\sqrt{x^2 + 1}\,dx$$

You recognize this as an ideal candidate for u-substitution. Here's what a strong student might write, aiming to be careful:

Let $u = x^2 + 1$ so that $du = 2x\,dx$. Substitute all parts:

$$\int_{x=1}^{x=2} 2x\sqrt{x^2 + 1}\,dx = \int_{x=1}^{x=2} \sqrt{u}\,du$$

Now integrate:

$$= \int_{x=1}^{x=2} u^{1/2} du = \left. \frac{2}{3} u^{3/2} \right|_{u=1}^{u=2} = \frac{2}{3}[(2)^{3/2} - (1)^{3/2}]$$

Calculate numerically:

• $2^{3/2} = 2 \cdot \sqrt{2} \approx 2.828$
• $1^{3/2} = 1$

So:

$$\frac{2}{3} (2.828 - 1) \approx \frac{2}{3} (1.828) \approx 1.218$$

But here's the problem: those bounds (u=1 and u=2) do not match the x bounds. The definite integral result is now numerically wrong.

The substitution step was correct. The error appears in this crucial line:

$$\int_{x=1}^{x=2} 2x\sqrt{x^2 + 1}\,dx = \int_{x=1}^{x=2} \sqrt{u}\,du$$

If you use the same numbers (1 and 2) as limits for the u-integral, they no longer correspond to the original interval on the x-axis. The limits in a definite integral must always match the variable of integration, as required by the Fundamental Theorem of Calculus: $\int_{a}^{b} f(x) dx$ is the net area as x goes from a to b. When you change variables, the endpoints must be in the new variable's units.

It's tempting to keep the original numbers because the substitution replaces everything else so cleanly. This is why the mistake feels so easy to make: nothing looks off until you check the result against the graph or the answer key.

Instead, after letting $u = x^2 + 1$, you must change the bounds:

• When $x = 1$, $u = 1^2 + 1 = 2$
• When $x = 2$, $u = 2^2 + 1 = 5$

So:

$$\int_{x=1}^{x=2} 2x\sqrt{x^2 + 1}\,dx = \int_{u=2}^{u=5} \sqrt{u}\,du$$

Now integrate:

$$= \left. \frac{2}{3} u^{3/2} \right|_{u=2}^{u=5} = \frac{2}{3}[(5)^{3/2} - (2)^{3/2}]$$

Calculate:

• $5^{3/2} = 5 \cdot \sqrt{5} \approx 11.180$
• $2^{3/2} = 2 \cdot \sqrt{2} \approx 2.828$

So:

$$\frac{2}{3}(11.180 - 2.828) = \frac{2}{3}(8.352) \approx 5.568$$

This is the true value of the definite integral. The earlier answer (about 1.218) is nowhere close. The limits matter.

The u-bounds must come from plugging each x-bound into your substitution.

This step follows directly from the Fundamental Theorem of Calculus: the definite integral $\int_{a}^{b} f(x) dx$ computes the net area under the curve from x = a to x = b. When you substitute u = g(x), du = g'(x) dx, the interval you integrate over in u must be u = g(a) to u = g(b) because your new variable measures something different along the axis.

You can only keep the original x-limits if, after integrating in u, you back-substitute for x before evaluating. Mixing u with x-limits (or vice versa) is an error.

If you have practiced indefinite integrals, you’re used to plugging x back in at the end. That trick works because you never use the bounds until after you’ve returned to x. But with definite integrals, when you swap to u, you’re literally swapping the tape measure. The two endpoints must stay consistent with your new variable.

This is not a mindless bookkeeping rule: the numbers themselves change meaning. A bound of 2 in x-space is a bound of 5 in u-space when u = x^2 + 1.

The invisible-bounds trap: It feels natural to just copy the limits down after substitution, but they mean something entirely different in the new variable. This is the main way definite integrals with substitution go off the rails.

Suppose you have:

$$\int_{x=0}^{x=1} 3(x+1)^2 dx$$

Let $u = x + 1$, so $du = dx$. When $x = 0, u = 1$; when $x = 1, u = 2$.

Substitute:

$$\int_{x=0}^{x=1} 3(x+1)^2 dx = \int_{u=1}^{u=2} 3u^2 du$$

But look what happens if you accidentally keep the original x-bounds as u-bounds:

$$\int_{u=0}^{u=1} 3u^2 du$$

Integrate:

• Correct: $\left. u^3 \right|_{u=1}^{u=2} = 8 - 1 = 7$
• Wrong: $\left. u^3 \right|_{u=0}^{u=1} = 1 - 0 = 1$

So the error is still visible. However, if your substitution were u = x, both bounds would match—so re-indexing blindly “works” but only in trivial cases. Except when your substitution function is the identity (u = x), this mistake burns you.

Try:

$$\int_{x=-2}^{x=1} (x+2) dx$$

Let $u = x + 2$, so when x = -2, u = 0; when x = 1, u = 3. The correct setup is:

$$\int_{u=0}^{u=3} u du = \left. \frac{1}{2}u^2 \right|_{0}^{3} = \frac{9}{2} - 0 = 4.5$$

If you forget and use u = -2 to u = 1:

$$\int_{u=-2}^{u=1} u du = \left. \frac{1}{2}u^2 \right|_{-2}^{1} = 0.5 - 2 = -1.5$$

Completely wrong. Check in the original x-coordinates:

$$\int_{x=-2}^{x=1} (x+2) dx = \int_{x=-2}^{x=1} x dx + \int_{x=-2}^{x=1} 2 dx = \left. \frac{1}{2} x^2 \right|_{-2}^{1} + \left. 2x \right|_{-2}^{1} = (0.5 - 2) + (2 - (-4)) = (-1.5) + 6 = 4.5$$

Only changing the bounds matches the true value.

Evaluate:

$$\int_{x=0}^{x=3} 6x (x^2 + 1)^2 dx$$

Try:

• Correct substitution: u = x^2 + 1
• Change both limits and variable
• Integrate and evaluate

Correct answer:

$$\int_{u=1}^{u=10} 3u^2 du = [u^3]_{1}^{10} = 1000 - 1 = 999$$

If you kept the x-limits (0 and 3) in u, you’d end up with $[u^3]_0^3 = 27 - 0 = 27$, which is wildly off. Units and intervals matter.

If you find yourself hesitating on this step, you are not alone—this is the single easiest way to sabotage a correct substitution. Fast check: ask yourself if the bounds match the new variable. If not, recalculate before evaluating.

Lean on this check as long as you need it. If you want feedback on your process, Learn4Less tutors can spot these boundary mismatches faster than any calculator, but you do not need a tutor to master this step if you make it part of your mental checklist.