The convention is blunt. \sqrt{4} = 2, not \pm 2. Both 2^2 and (-2)^2 equal 4, so why does the symbol refuse to hand you both? The sibling page Why Does √9 Equal Just 3, Not ±3? gives the rule: the radical is a function, functions return one output. This page is about the consequence — what exactly would break if we let \sqrt{4} mean the set \{+2, -2\}?
Short version: if \sqrt{\phantom{x}} returned two numbers instead of one, almost every downstream piece of mathematics would splinter into ambiguity. The convention is the smallest surgical fix that keeps graphs, calculus, computation, and the quadratic formula working.
What \sqrt{\phantom{x}} is trying to be — a FUNCTION
A function is a rule that takes an input and returns exactly one output. When you write f(4), there is a single number on the other side of the equals sign. Not two, not a set — one number.
If \sqrt{4} returned the set \{+2, -2\}, then \sqrt{\phantom{x}} would not be a function. It would be a multi-valued relation — a rule that sometimes hands you more than one answer. Multi-valued relations are legal mathematical objects, but you cannot plug them into the function-shaped slot in a formula. Every tool you use — graphs, derivatives, composition, substitution — assumes the thing inside it returns one value.
What breaks if √ is multi-valued
Run through the damage, one field at a time.
Graphs. The curve y = \sqrt{x} is the upper half of a sideways parabola — one y for each x \geq 0, passing the vertical-line test. If \sqrt{\phantom{x}} were multi-valued, the graph would be the whole sideways parabola, two y-values at every positive x. It would fail the vertical-line test and stop being a function-graph at all. You would have to pick one branch first, and you'd be right back at the principal-root convention by another name.
Calculus. The derivative \dfrac{d}{dx}\sqrt{x} = \dfrac{1}{2\sqrt{x}} depends on \sqrt{x} being a well-defined function. Feed a multi-valued relation into derivatives, limits, or integrals, and the machinery gums up. At every x you would have to track which branch you were differentiating, and the answer would flip sign based on that choice. The chain rule would need sign-bookkeeping. Every integral involving \sqrt{\phantom{x}} would have to specify a branch.
Composition. f(g(x)) plugs g's output into f, and assumes g returns a single number. If \sqrt{\phantom{x}} were multi-valued, \sqrt{\sin x} would have two values for every x with \sin x \geq 0. Stack compositions — say \sqrt{\sqrt{x^2 + 4}} — and you get 2 \times 2 = 4 possible values per input. Stack more and the tree explodes.
Computation. Every programming language implements sqrt as deterministic: one input in, one number out. Math.sqrt(4) is 2 in JavaScript, Python, C, Java — all of them. Redefine sqrt to return both +2 and -2, and a program that computes \sqrt{a} + \sqrt{b} + \sqrt{c} would have to fork into 2 \times 2 \times 2 = 8 parallel paths. Exponential cost in the number of radicals. No language implements it this way because no useful code could sit on top.
The quadratic formula — the cleanest example
Look at the formula you have used a hundred times:
The \pm sits outside the radical. The \sqrt{\phantom{x}} gives one non-negative value; the \pm explicitly creates the two solutions. The whole formula reads as "take the one non-negative square root, then choose its sign."
If \sqrt{\phantom{x}} were multi-valued, the \pm would be redundant — the radical would already produce both signs. But the ambiguity would now be hidden inside the symbol rather than visible in front of it. The current form is elegant precisely because \sqrt{\phantom{x}} is single-valued: the sign-choice lives in one visible place, not buried inside a radical.
Historically — why mathematicians picked the non-negative choice
Once you commit to single-valuedness, you still have to choose which value to return. Why the positive one?
- The square root was first introduced as the side-length of a square of given area. Lengths are non-negative; the positive root matches the geometry.
- The identity \sqrt{x} \geq 0 makes inequality reasoning cleaner: \sqrt{x^2 + 1} > 0 without sign cases.
- The function y = \sqrt{x}, defined as the non-negative root, is the right-inverse of y = x^2 on x \geq 0.
Either choice (always-positive or always-negative) would be consistent. The positive choice is the one with the fewest extra minus signs lying around.
For even roots the ambiguity is real; for odd roots it doesn't exist
The whole sign-ambiguity story is an even-root phenomenon. An even root of a positive number has two real roots — +r and -r — and the convention picks +r. But an odd root has exactly one real value. \sqrt[3]{8} is just 2; there is no other real number whose cube is 8. Similarly \sqrt[3]{-8} = -2.
So "principal root" is meaningful specifically for even roots. For odd roots, there is only one real candidate, and calling it "principal" is a formality. This is why √(x²) = |x| needs the absolute value, while \sqrt[3]{x^3} = x sits there unadorned.
What happens in complex numbers
Zoom out. In the complex numbers, every non-zero number has n different n-th roots. The number 1 has three complex cube roots — 1, and two others rotated by 120^\circ on the unit circle. Which does \sqrt[3]{1} denote? Convention picks the principal root: the one with the smallest non-negative argument. For 1, that is 1 itself. Same idea, same trade-off: the smallest, cleanest choice that turns a multi-valued relation into a function.
Could we define √ differently?
Yes. You could define \sqrt{x} for x < 0 as i\sqrt{|x|}, extending the principal branch into the complex plane — which is what calculators and Python do. You could also pick -i\sqrt{|x|}; both are consistent. The mathematics demands only that you pick some branch and stick with it.
The ± is the safety valve
The convention never loses information. Whenever both signs matter — solving x^2 = 4, applying the quadratic formula — you write \pm in front of the radical. The ambiguity is captured, explicit, and visible.
The \sqrt{4} gave you 2; the \pm gave you both signs. The convention and the safety valve together reproduce every piece of sign-information that a multi-valued radical would carry — with the advantage that the ambiguity is in plain sight.
Worked examples of how the convention works in practice
- Equation x^2 = 16. Solve: x = \pm\sqrt{16} = \pm 4. The \pm is explicit; it names both solutions.
- Expression \sqrt{16}. Evaluate: \sqrt{16} = 4. Function application; one output.
- Physical length. A square has area 16 m². Its side length is \sqrt{16} = 4 m — not -4 m, because lengths are non-negative by physical meaning. The convention aligns with the geometry.
- Simplify \sqrt{x^2}. The answer is |x|, not x. The absolute value is how the convention preserves non-negativity when the variable's sign is unknown.
Each case shows the same pattern: the radical hands back one non-negative number; any sign-ambiguity is written out with a \pm or absorbed into an |\cdot|.
The convention costs you nothing if you remember the ±
You are not losing anything by adopting the principal-root convention. Every time both signs matter, you write the \pm and both signs appear. Every time they do not, the convention keeps \sqrt{\phantom{x}} a well-behaved function. You trade the small, visible risk of forgetting a \pm for the huge, invisible win of being able to differentiate, graph, compose, and compute without branching.
Common confusions
-
"Can't I just redefine \sqrt{\phantom{x}} as \pm in my own problem?" You could privately, but you would confuse every reader. The convention is universal — textbooks, calculators, programming languages, exam keys. Use it and write \pm outside where needed.
-
"The principal root loses information." It does not. The \pm is a separate modifier that lives in the equation-solving step, not the symbol. The full sign-information is preserved whenever you need it.
-
"Why is 'principal' non-negative, not non-positive?" Because non-negative aligns with geometric intuition (side lengths), simplifies inequalities, and inverts y = x^2 on its natural domain. The non-positive choice would be consistent, just less pleasant.
Closing
The principal-root convention is a tiny piece of bureaucracy that makes the rest of mathematics mechanical. Without it, every radical is a coin-flip; with it, radicals are ordinary function-applications and the \pm is the safety valve for places where the sign genuinely matters. Learn the convention, write the \pm where it belongs, and you will never worry that \sqrt{4} is short-changing you. It is giving you one of the two roots by design — and the other is available the moment you ask for it in the open.