Why the Order of the Bounds Matters
A definite integral has two endpoints, and their order carries information. When the upper endpoint lies to the left of the lower endpoint, the integral is not interpreted as the positive area between the graph and the horizontal axis. Instead, reversing the endpoints changes the sign of the accumulated quantity.
This convention makes definite integrals consistent with signed accumulation. It also gives a simple way to evaluate an integral whose bounds appear in the less familiar order: reverse the bounds, evaluate in the increasing direction, and attach a minus sign.
The condition that \(f\) is Riemann integrable on \([a,b]\) ensures that the integrals in this identity exist. Throughout this tutorial, endpoint reversal will be applied only to such integrals. The case of equal endpoints is not needed here.
The Endpoint-Reversal Identity
For any Riemann-integrable function \(f\) on \([a,b]\), with \(a<b\), the key identity is
The interval between the endpoints has not changed; only its orientation has changed. Consequently, the magnitude of the accumulated quantity is the same, while its sign is reversed. This is the integral version of the signed-width convention used in oriented Riemann sums.
In particular, one must distinguish the signed integral from the geometric area. Geometric area is nonnegative, while a definite integral can be negative because of the function’s values, the direction of integration, or both. Reversing the bounds changes the sign of the integral, not the graph or the geometric region.
Why Reversing the Bounds Negates the Integral
Here is a justification using Riemann sums. Fix \(a<b\). An increasing partition of \([a,b]\) has points \(a=x_0<x_1<\cdots<x_n=b\). Its subinterval widths are \(\Delta x_i=x_i-x_{i-1}>0\), and a tagged Riemann sum has the form \(\sum_{i=1}^n f(\xi_i)\Delta x_i\), where \(\xi_i\in[x_{i-1},x_i]\).
For the reversed interval, write the partition in decreasing order as \(b=y_0>y_1>\cdots>y_n=a\). The oriented width of its \(i\)th subinterval is \(y_i-y_{i-1}<0\). Pair that subinterval with the same segment of \([a,b]\), traversed in the opposite direction. Its positive width is the negative of the reversed width, and the function value can be sampled at the same point. Thus corresponding terms in the two sums differ by a factor of \(-1\).
Each descending subinterval \([y_i,y_{i-1}]\) is the same segment as an ascending subinterval, with its endpoints written in the opposite order.
The descending oriented width is \(y_i-y_{i-1}\), while the corresponding ascending width is \(y_{i-1}-y_i=-(y_i-y_{i-1})\).
Using the same tags, each reversed Riemann sum is the negative of its matched forward sum. As the partition mesh shrinks to zero, the sums converge to their respective integrals, so the limiting integrals also differ by a factor of \(-1\).
More explicitly, for a matched pair of subintervals and a shared tag \(\xi\), the terms are \(f(\xi)(y_i-y_{i-1})\) and \(f(\xi)(y_{i-1}-y_i)\). Therefore
Adding this relation over all subintervals gives opposite Riemann sums. Since \(f\) is Riemann integrable on the interval, the corresponding limits exist; taking those limits proves the endpoint-reversal identity.
Worked Example: A Constant Function
Worked Example: A Constant Function
Evaluate \(\int_1^5 3\,dx\) and \(\int_5^1 3\,dx\).
On \([1,5]\), the graph of \(f(x)=3\) is a horizontal line. The interval has width \(5-1=4\), so its signed area in the increasing direction is the rectangle’s area:
Now reverse the endpoints. By the endpoint-reversal identity,
Check: The oriented width from \(5\) to \(1\) is \(1-5=-4\). Multiplying by the constant height gives \(3(-4)=-12\), agreeing with the identity.
Worked Example: Reversing an Integral of \(x\)
Worked Example: Reversing an Integral of \(x\)
Evaluate \(\int_3^1 x\,dx\).
First evaluate in increasing order. On \([1,3]\), the graph of \(y=x\) forms a trapezoid with parallel vertical sides of lengths \(1\) and \(3\), separated by a horizontal distance of \(3-1=2\). Its area is
The requested integral has the bounds in reverse order, so
Check: The same trapezoid has geometric area \(4\), but the integration direction is reversed. Its signed integral is therefore \(-4\), as computed.
Worked Example: An Expression with Mixed Endpoint Orders
Worked Example: An Expression with Mixed Endpoint Orders
Evaluate \(\int_2^6 1\,dx+\int_5^1 x\,dx\).
The first integral is forward. Its integrand is constantly \(1\), and the interval has width \(6-2=4\):
The second integral is reversed. Evaluate its increasing-order counterpart first. The graph of \(y=x\) on \([1,5]\) gives a trapezoid with vertical sides \(1\) and \(5\), and width \(5-1=4\):
Reversing those bounds changes the sign:
Now combine the two evaluated numbers:
Check: The forward integral contributes \(4\); the reversed integral contributes \(-12\). Their sum is \(4-12=-8\), so the result respects the orientation of both terms.
A Common Mistake: Using the Width Without Its Sign
A frequent error is to treat every interval width as positive, even when the bounds are reversed. For example, an incorrect calculation might claim
This calculation uses the positive distance between the endpoints, \(4-1=3\), but ignores the direction from \(4\) to \(1\). The oriented width is the endpoint reached minus the endpoint started: \(1-4=-3\). Thus the correct calculation is
Equivalently, evaluate in increasing order and then reverse:
The positive number \(6\) is the geometric area of the rectangle. It is not the value of the integral with reversed bounds.
A Reliable Way to Handle Reversed Bounds
When evaluating a definite integral whose endpoints are in decreasing order, it is useful to separate the orientation decision from the evaluation itself.
- Identify which endpoint is smaller and which is larger.
- Write the integral with the smaller endpoint first.
- Place a minus sign in front of that increasing-order integral.
- Evaluate the increasing-order integral, then retain the minus sign.
For example, applying these steps to \(\int_6^2 f(x)\,dx\), provided \(f\) is Riemann integrable on \([2,6]\), gives
This procedure is especially useful in longer expressions. Each integral must be read with its own endpoint order: one term may be forward and another reversed. Apply the sign change to the reversed term before combining numerical values.
Why the Identity Matters
Endpoint reversal is more than a notation convention. It preserves the meaning of accumulated change when the direction of travel is reversed. If an integral represents a net change accumulated while moving from one endpoint to another, reversing the trip reverses the sign of that net change.
It also prevents a common confusion between signed accumulation and geometric area. The graph and the region between the endpoints are unchanged when the direction is reversed. What changes is the orientation of the accumulation, and therefore the sign of the definite integral.
Check Your Understanding
Use the endpoint-reversal identity and signed area where appropriate.
- If \(\int_1^4 f(x)\,dx=7\), what is \(\int_4^1 f(x)\,dx\)?
- Evaluate \(\int_3^0 2\,dx\).
- The graph of \(y=x\) on \([2,6]\) forms a trapezoid. Find \(\int_6^2 x\,dx\).
- Evaluate \(\int_1^3 1\,dx+\int_5^2 1\,dx\).
- Explain why \(\int_b^a f(x)\,dx\) is not generally the positive geometric area between the graph and the horizontal axis when \(a<b\).
