AP Calculus AB and BC
Integral of 1/(x ln x): Answer and Substitution
The integral of 1/(x ln x) with respect to x is ln|ln x| + C. The substitution u = ln x works because du = (1/x) dx is already present in the integrand, turning the problem into the integral of 1/u. The absolute value is required because ln x is negative for x between 0 and 1.
How to integrate 1/(x ln x) by substitution
This integrand is a substitution problem that announces itself. The derivative of is , and there is a sitting in the integrand already, so the inner function and its derivative are both present.
Write the integrand so the pairing is visible before you substitute.
Now let , so . Every piece of the integrand is accounted for, with no leftover to convert.
The problem becomes the single most important basic antiderivative on the course.
Substituting back gives a logarithm of a logarithm, which looks strange but is exactly right.
Check it by differentiating
Differentiate with the chain rule. The outer derivative is and the inner derivative is , and their product is , the integrand.
Why the absolute value is not optional
The domain of the integrand is with , since makes the denominator zero. On the value of is NEGATIVE, so writing without bars would be undefined on half the domain.
Take . Then , and does not exist, while is fine. The absolute value is what lets one formula cover both sides of .
This is the same reason carries bars, and it is the most commonly dropped notation in Unit 6.
One antiderivative, two separate intervals
Strictly, splits the domain into two pieces and the constant may differ on each. AP answers write a single , but you cannot integrate ACROSS : the integrand blows up there, so a definite integral spanning it is improper and in fact divergent.
Where the integral of 1/(x ln x) shows up on the AP exam
Integrating using substitution is Topic 6.9 in Unit 6, on both AB and BC, and this is a standard example of the pattern where the inner function is a logarithm. Unit 6 carries a weighting of 15 to 20 percent on both exams.
A definite integral over an interval that avoids is straightforward.
On BC this integrand is the engine behind a well known convergence result. The integral test compares the series to , and because grows without bound, the integral diverges and so does the series.
That result is worth remembering as a boundary case. The terms shrink faster than but still not fast enough to converge, which is why a shrinking term is never on its own evidence of convergence.
Common mistakes with the integral of 1/(x ln x)
- Writing . The rule requires the DERIVATIVE of to be present as a factor. Here has derivative , which is nowhere in the integrand.
- Dropping the inner absolute value and writing , which is undefined for .
- Choosing , which produces and changes nothing.
- Integrating across as though the integrand were continuous there. It is not, so any definite integral whose interval contains is improper and divergent.
- Reading the answer as territory. That is the antiderivative of , the reciprocal arrangement, which gives .
Two lookalike integrands, two different answers
has the logarithm downstairs, while has it upstairs. Both use ; only the power of differs, against .
Every answer on this page is machine checked
An automated test differentiates the antiderivative above and confirms it returns the integrand. A wrong sign or a missing factor fails the build, so it cannot reach you.
Frequently asked questions
What is the integral of 1/(x ln x)?
It is . Substitute , so , and the integral becomes .
Why does the answer need an absolute value?
Because is negative for , and the logarithm of a negative number is undefined. The bars let the single expression cover the whole domain instead of only .
Is the integral of 1/(x ln x) on the AP Calculus AB exam?
Yes. It is an ordinary -substitution, Topic 6.9, which is on both AB and BC. The infinite version used in the integral test is BC material.
Does the integral of 1/(x ln x) from 2 to infinity converge?
No, it diverges. The antiderivative grows without bound, so the limit is infinite. By the integral test this is why also diverges.
How is this different from the integral of ln x over x?
Both substitute , but becomes , while becomes . The answers are and .