Why the Order of the Bounds Matters
A definite integral is not determined only by the interval between its endpoints. The order of the endpoints matters: reversing them changes the sign of the integral. This is essential when a change of variable transforms the original endpoints into bounds written in the opposite order.
For example, the interval from \(1\) to \(3\) and the interval from \(3\) to \(1\) cover the same points, but their definite integrals have opposite values. This sign is not an extra feature of substitution. It follows from the signed-integral convention and is what makes the definite substitution formula work with transformed endpoints.
The Signed-Integral Identity
For an integrable function \(f\), the definite integral is oriented: moving from the lower endpoint to the upper endpoint gives one sign, and moving in the reverse direction gives the opposite sign. In particular, if \(a\) and \(b\) are in the domain of \(f\),
If the endpoints are equal, the integral is zero:
One way to see the meaning of the sign is through the Fundamental Theorem of Calculus. If \(F\) is an antiderivative of \(f\), then the integral from \(a\) to \(b\) is \(F(b)-F(a)\). Reversing the endpoints gives \(F(a)-F(b)\), which is the negative of the original difference.
Why the Identity Works
Suppose \(f\) is continuous on an interval containing \(a\) and \(b\), and let \(F\) be an antiderivative of \(f\) there. The Fundamental Theorem of Calculus gives
Applying the same theorem with the endpoints in the other order gives
Negating the second expression yields
This proves the signed-integral identity under the stated continuity condition. The same endpoint-reversal rule holds whenever the relevant definite integrals exist.
Transformed Bounds Keep Their Orientation
In definite substitution, the endpoints in the new variable come from applying the substitution to the original endpoints. If \(u=g(x)\), then the original lower endpoint \(x=a\) gives \(u=g(a)\), and the original upper endpoint \(x=b\) gives \(u=g(b)\). Thus the transformed bounds are written from \(g(a)\) to \(g(b)\), in that order.
For a continuous integrand \(h\) and a differentiable substitution \(u=g(x)\), the definite substitution formula is
The bounds on the right do not have to be numerically increasing. If \(g(b)<g(a)\), the transformed integral is written with its lower endpoint numerically larger than its upper endpoint. That notation is valid: it records the orientation. If you prefer to write the bounds in increasing order, reverse them and negate the integral.
The substitution formula can also be justified directly by the Fundamental Theorem of Calculus. Let \(H\) be an antiderivative of \(h\). By the chain rule, \(H(g(x))\) has derivative \(h(g(x))g'(x)\). Therefore,
This argument explains why the transformed bounds preserve their order: the endpoint difference is \(H(g(b))-H(g(a))\). Swapping the bounds would reverse that difference, so the sign must also be reversed.
Worked Example: Reversing Bounds in a Simple Integral
Worked Example: Reversing Bounds in a Simple Integral
Evaluate \(\int_3^1 2x\,dx\), and compare it with the integral whose bounds are reversed.
An antiderivative of \(2x\) is \(x^2\). Evaluating with the stated bounds gives
Now reverse the bounds and evaluate:
The values are negatives of each other, as required:
Check: differentiating the antiderivative \(x^2\) gives \(\frac{d}{dx}(x^2)=2x\), the original integrand. The endpoint evaluations therefore use a valid antiderivative.
Worked Example: A Substitution Produces Descending Bounds
Worked Example: A Substitution Produces Descending Bounds
Evaluate \(\int_0^2 (3-x)^2\,dx\) using \(u=3-x\). The original lower endpoint \(x=0\) gives \(u=3\), while the original upper endpoint \(x=2\) gives \(u=1\). The transformed bounds are therefore \(3\) to \(1\).
Since \(du=-dx\), the transformed integral is
Use the signed-integral identity to write this with bounds in increasing order:
Now evaluate:
Check in the original variable. An antiderivative of \((3-x)^2\) is \(-\frac{(3-x)^3}{3}\), because
Evaluating at the original endpoints confirms the result:
Worked Example: Reversed Original Bounds and Substitution
Worked Example: Reversed Original Bounds and Substitution
Evaluate \(\int_2^0 \frac{2x}{x^2+1}\,dx\) using \(u=x^2+1\). The original lower endpoint \(x=2\) gives \(u=5\), and the original upper endpoint \(x=0\) gives \(u=1\). Also, \(du=2x\,dx\).
Substitution gives
An antiderivative of \(1/u\) is \(\ln u\) for \(u>0\). Thus
Equivalently, reversing the transformed bounds requires a minus sign:
Check directly in \(x\): an antiderivative is \(\ln(x^2+1)\), since its derivative is \(\frac{2x}{x^2+1}\). Therefore,
A Common Mistake: Swapping Bounds Without Changing the Sign
A frequent error is to obtain transformed bounds in descending order, silently swap them to make them ascending, and leave the integral's sign unchanged. For the substitution in the previous example, that mistake would turn
into the incorrect expression \(\int_1^5 \frac{1}{u}\,du=\ln 5\). The correct relationship includes a minus sign:
The wrong answer can also be detected by checking the original integral: its bounds run from \(2\) down to \(0\), and its value is \(\ln 1-\ln 5=-\ln 5\), not \(+\ln 5\). The correction is simple: either keep the transformed bounds in their original order, or swap them and explicitly negate the integral.
A Reliable Way to Handle Bound Order
When using substitution in a definite integral, keep track of the endpoints as part of the calculation, rather than treating them as a final formatting choice.
Write \(u=g(x)\) and determine the differential factor needed to rewrite the integrand.
Calculate \(u=g(a)\) from the original lower endpoint and \(u=g(b)\) from the original upper endpoint.
Use \(g(a)\) to \(g(b)\) as obtained, or reverse them and include a minus sign.
Use an antiderivative in the new variable, then check the sign or verify the value in the original variable when practical.
Why Integral Orientation Matters
The orientation rule lets the substitution formula work without requiring transformed bounds to be written in increasing order. It also prevents sign errors when the original bounds themselves are reversed. The reliable principle is that the order of endpoints carries mathematical information: reversing that order reverses the sign.
There are two equivalent ways to express the same definite integral. You may preserve the transformed order, even if it is descending, or you may write the bounds in the opposite order and attach a minus sign. You may not change the order and keep the sign unchanged.
Check Your Understanding
Use the signed-integral identity and transformed-bound order to answer these questions.
- Write \(\int_4^1 f(x)\,dx\) using bounds \(1\) to \(4\).
- For \(u=2x+1\) on an integral with \(x\)-bounds \(0\) to \(3\), what are the transformed bounds, in order?
- If a substitution produces \(\int_7^2 h(u)\,du\), how can you write the same integral with bounds \(2\) to \(7\)?
- What sign error occurs if bounds are swapped but the minus sign is omitted?
- Why does the Fundamental Theorem of Calculus imply that reversing the endpoints changes the sign?
