You have been squaring both sides of equations since Class 8. If \sqrt{x + 1} = 3, you square both sides to get x + 1 = 9, so x = 8. It feels like a universal move — a hammer that works on anything with a square root. So when you meet an inequality like \sqrt{x + 1} < 3 and think "square both sides," the reflex is natural.
And it is almost always wrong.
The trap is not that squaring is "illegal" on inequalities. It is that squaring is not always a direction-preserving move. Equations are symmetric (a = b is the same as b = a), so any function you apply to both sides keeps the equality. Inequalities carry a direction — an arrow pointing from smaller to larger. Squaring can keep that arrow, reverse it, or destroy it completely, depending on the signs of the numbers you are squaring.
The clean example that exposes the trap
Start with an inequality everyone agrees is true:
Now square both sides without thinking:
That is false. The inequality flipped. You didn't make an arithmetic mistake; you did exactly what you do for equations, and the inequality lied to you.
The reason is that the squaring function f(x) = x^2 is not monotonic on \mathbb{R}. It decreases on (-\infty, 0] and increases on [0, \infty). When two numbers sit on opposite sides of zero — one negative, one positive — squaring can pull them in different directions. -5 is far to the left of zero; 3 is modestly to the right. Squaring folds the number line at zero and measures only the distance from the origin. After the fold, -5 (distance 5) overshoots 3 (distance 3), so the arrow reverses.
The rule, stated cleanly
A function preserves an inequality when it is monotonically increasing on the interval containing both sides. Squaring is increasing on [0, \infty) and decreasing on (-\infty, 0]. So:
- If both sides are non-negative (a \ge 0 and b \ge 0), then a < b \iff a^2 < b^2. Squaring is safe; the arrow is preserved.
- If both sides are non-positive (a \le 0 and b \le 0), then a < b \iff a^2 > b^2. Squaring flips the arrow — same rule as multiplying by a negative.
- If the sides have mixed signs, there is no universal rule. You must split into cases.
Contrast that with equations. The equation a = b is preserved by any function whatsoever — squaring, cubing, exponentiating, taking logs — because applying the same function to two equal inputs gives two equal outputs, no matter what the function does elsewhere. Inequalities are more fragile because they encode order, and most functions don't preserve order everywhere.
The squaring trap creates ghost solutions
There is a second reason to be careful: even when squaring preserves the direction, it can introduce solutions that don't satisfy the original inequality. Consider:
Both sides non-negative? The left side \sqrt{x} requires x \ge 0, and the right side x - 2 is non-negative only when x \ge 2. So squaring is safe only when x \ge 2. Square both sides under that restriction:
The quadratic is positive on (-\infty, 1) \cup (4, \infty). Intersect that with the restriction x \ge 2: you get (4, \infty).
If you had squared without checking that x - 2 \ge 0, you would have accepted x \in [0, 1) as well — and at x = 0, the original inequality says 0 < -2, which is false. Those are extraneous solutions: they solve the squared inequality but not the original. They sneak in because the squared inequality can't distinguish a < b from a > -b — both give a^2 < b^2 when the right side is positive.
Exam graders love this trap. You square, write a tidy answer, and lose marks because you forgot to intersect with the domain where squaring was legitimate.
A quick field guide: when can you square?
When you meet an inequality involving a square root or an absolute value, ask three questions before squaring:
- Is the side I want to square always non-negative? Square roots and absolute values are, by definition. But expressions like x - 2 are not — they are negative for x < 2.
- Is the other side also guaranteed non-negative? If it can be negative, squaring is not safe. Often you need to split into cases: one where the right side is negative (handle directly), and one where it is non-negative (then square).
- After squaring and solving, do my answers actually satisfy the original inequality? The final check is cheap and catches extraneous solutions from hidden sign issues.
For |f(x)| < g(x), there is a cleaner alternative that avoids squaring altogether: rewrite it as the double inequality -g(x) < f(x) < g(x) — which is what the intervals and inequalities preview article uses throughout. This avoids the trap entirely. Similarly, |f(x)| > g(x) splits into two cases: f(x) > g(x) or f(x) < -g(x). Direct, no squaring, no extraneous solutions.
Why equations are different
The deeper reason equations let you get away with squaring is that the equality a = b means "the same number." Squaring the same number twice gives the same result, so a^2 = b^2 follows automatically. The equation a = b is preserved by literally any function you like. Even when squaring introduces extra roots in equations — \sqrt{x} = x - 2 squared gives x = x^2 - 4x + 4 with roots x = 1 and x = 4, but only x = 4 satisfies the original — the logic is still "implication in one direction," and the check at the end catches the strays.
Inequalities carry strictly more information than equations. a < b says "they are different, and here is which is bigger." To preserve that extra structure, your function must respect order — be monotonic — on the relevant range. Squaring does, but only on one side of zero. That one-word caveat is what trips everyone up the first time they meet it. Once you know the rule, the fix is automatic: check signs first, then square.
The final takeaway: think of squaring as a move that is almost always safe for equations and almost always risky for inequalities. For inequalities, prefer the direct techniques — double inequalities for "< r" cases, case splits for "> r" cases — and reach for squaring only when both sides are known to be non-negative, and even then, verify your answer against the original at the end.