Integration by substitution
Some integrals refuse to match any standard form on sight, yet hide a simple structure underneath: an "inner function" tangled with its own derivative. Integration by substitution is the technique that unmasks this structure by swapping the variable, collapsing a messy integral in into a plain one in a new variable . It is the first of the three integration techniques in the CBSE Class 12 Integrals chapter (the others being partial fractions and integration by parts), and it is nothing more than the chain rule for differentiation, read backwards. Spotting the right substitution is a skill built through pattern recognition — look for a bracket, root, exponent or angle whose derivative is quietly sitting elsewhere in the integrand. Once found, the rest is bookkeeping: replace, integrate, and substitute back.
If an integrand can be written as , the substitution turns it into the simpler integral , which is then converted back to .
The core idea
Substitution is a change of variable, not a new integration rule — it is a way of packaging the chain rule so an unfamiliar integral becomes a familiar one. The formal statement:
where: is the "inner function," is the new variable, is its derivative (which must be present, up to a constant multiple, elsewhere in the integrand), and is any antiderivative of .
Recognizing when to substitute. Scan the integrand for an inner expression that looks awkward standing alone — a bracket raised to a power, a square root, an exponent, an angle multiplied by a constant — and check whether its derivative (up to a constant factor) is sitting nearby. If it is, that expression is almost always the right choice for . Four patterns recur constantly:
- : put .
- : put ; this always gives .
- : put ; gives .
- (including ): put .
A second family is trigonometric substitution, where itself is replaced (e.g. ) to tame expressions like , turning them into . This is used heavily for the standard square-root forms later in the chapter, but the mechanics — replace, integrate, substitute back — stay identical.
Two habits matter throughout: always substitute back to the original variable before writing a final answer, and always check for a missing constant factor between and the differential term actually present in the integrand.
The goal is to justify, from nothing but the chain rule for differentiation and the definition of an antiderivative, why replacing inside and integrating with respect to instead of produces the correct antiderivative once is converted back to — this is what licenses every substitution problem in the chapter.
Step 1. Start from an antiderivative in . Suppose is any function with , so that by the very definition of an indefinite integral.
Step 2. Build the composite function and differentiate it with respect to using the chain rule. . Since , this becomes .
Step 3. Read the result as an integration statement. The last line says the derivative of is exactly the integrand . By the definition of an antiderivative, that means — precisely the substitution formula, with no extra assumption beyond the chain rule.
Step 4. Connect this to the mechanical procedure. Writing gives , so treating as a formal replacement and computing produces the same as Step 1; substituting back recovers exactly from Step 3. The "replace and substitute back" recipe is therefore not a trick — it is shorthand for the chain rule run in reverse.
Worked example
Evaluate .
Step 1. Identify the inner function. Here , and its derivative is exactly the other factor present in the integrand — a strong signal to substitute with .
Step 2. Substitute. Let , so , giving . The integral becomes purely in :
Step 3. Integrate in and substitute back. , and replacing gives
Step 4. Verify by differentiating: , which matches the original integrand exactly.
Final answer: .
Competitive stretch
Evaluate .
Step 1. Rewrite and notice the numerator is (up to sign) the derivative of the denominator, since .
Step 2. Substitute , so , i.e. . This turns the integral into .
Step 3. Substitute back and simplify using .
Final answer: , one of the standard integrals worth memorizing directly, though deriving it once shows substitution recognizing the pattern in disguise.
Common traps
- Substituting but forgetting to also replace correctly — every part of the original integral, including the differential, must be rewritten in terms of .
- Leaving a stray behind after substitution — a valid substitution must leave the integral entirely in terms of (or a constant that can be pulled outside the integral sign).
- Missing a constant factor: treating (missing the coefficient ) as if it matched the worked example directly. Since , here , so a compensating factor of is required.
- Forgetting to substitute back to at the end, leaving the final answer stranded in terms of — a definite/indefinite integral in must be answered in .
Ask the tutor: not sure which substitution to try for a tricky integrand — describe it to the AI tutor and ask for a hint on spotting the derivative pairing.
How do I know which part of the integrand to call t?
Look for an inner expression that would be awkward to integrate on its own — a bracket raised to a power, a square root, an exponent, or an angle multiplied by a constant — and check whether its derivative, possibly with a missing constant multiple, appears elsewhere in the integrand. If both pieces are present, that inner expression is almost always the right choice for . If no such pairing exists no matter what you try, substitution alone will not work and the problem likely needs a trigonometric identity, partial fractions, or integration by parts instead.
Why do I need to substitute back to x at the end?
Because the original question asked for the integral of a function of , and is only an intermediate result written in a variable you invented for convenience. Leaving the answer as is mathematically incomplete and will lose marks, since has no meaning outside your working. Always replace with in the final line so the answer is expressed in the same variable the question was posed in.
What if the derivative of my chosen inner function is off by a constant factor?
That is completely fine and very common — it does not mean the substitution is wrong. If gives for some constant , just solve for the piece you actually need, such as , and carry that constant through the integral. Only worry when the leftover expression after substitution still contains that cannot be written in terms of at all — that signals a genuinely wrong choice of , not just a missing constant.