Estimating square roots

Not every whole number is a perfect square — 2\sqrt{2}, 70\sqrt{70}, 200\sqrt{200} have no whole-number square root at all. Yet you can still say, with confidence and without a calculator, roughly what such a square root equals. Estimating square roots is the skill of trapping a non-perfect square's root between two consecutive whole numbers and then refining that guess to one decimal place, using nothing more than the perfect squares you already know by heart. This closes the ICSE Class 8 chapter Squares and Square Roots. After finding exact roots of perfect squares by prime factorisation and long division, this article asks: what do you do when the number is not a perfect square? The answer builds real number sense — the kind that lets you catch a wrong calculator answer at a glance — and lays the groundwork for meeting irrational numbers like 2\sqrt{2} formally in Class 9.

The core insight

A non-perfect square's root always lies strictly between the roots of the two nearest perfect squares, and comparing how close the number sits to each of those squares turns that bracket into a usable one-decimal estimate.

The core idea

Every whole number xx that is not a perfect square sits somewhere between two consecutive perfect squares, a2a^2 and b2=(a+1)2b^2 = (a+1)^2. Because squaring is an increasing operation for positive numbers, taking square roots preserves the order:

a2<x<b2  ⟹  a<x<ba^2 < x < b^2 \quad \implies \quad a < \sqrt{x} < b

where: aa and b=a+1b=a+1 are consecutive whole numbers, xx is the number whose root you want, and a2a^2, b2b^2 are the perfect squares bracketing it. This single line already answers the first, coarsest kind of question — "between which two whole numbers does 70\sqrt{70} lie" — without any arithmetic beyond recalling that 82=648^2=64 and 92=819^2=81. The bracket by itself is often too rough for classwork, which usually asks for one decimal place. The fix is to notice that x\sqrt{x} does not sit exactly halfway between aa and bb just because xx sits between a2a^2 and b2b^2 — it sits closer to whichever end xx is closer to. Comparing the two gaps, x−a2x-a^2 and b2−xb^2-x, tells you which way to lean. Pushing that idea into a formula gives a simple linear estimate:

x ≈ a+x−a2b2−a2\sqrt{x}\ \approx\ a + \frac{x-a^2}{b^2-a^2}

where: x−a2x-a^2 is how far xx has climbed above the lower perfect square, and b2−a2b^2-a^2 is the full gap between the two perfect squares, so the fraction is "how far across the gap" xx sits, added on to aa. This is not exact — the true relationship between xx and x\sqrt{x} is a curve, not a straight line — but for a one-decimal classroom estimate it is close enough, and it is always checked afterwards by squaring the estimate back. Notice, too, why estimation matters as a habit and not just a technique. Suppose a long-division calculation for 70\sqrt{70} came out to 8787 instead of 8.378.37 because of a misplaced decimal. The bracket 8<70<98<\sqrt{70}<9, found in seconds from perfect squares alone, would flag that answer as absurd before you even checked the working line by line. Estimation is therefore a fast first pass that catches large errors early, and it is also the intuition you carry forward when, in later classes, you place irrational numbers such as 2\sqrt2 or 3\sqrt3 on a number line. The figure below shows why the linear guess works reasonably well over a short stretch: between two consecutive perfect squares, the curve y=xy=\sqrt{x} is nearly straight, so sliding along the chord from (a2,a)(a^2,a) to (b2,b)(b^2,b) lands close to the true curve.

Figure
The curve y equals square root of x between two perfect squares with the straight line estimate lying close beside it
The linear estimate hugs the curve between two perfect squares x square root of x a squared, a b squared, b linear estimate true curve, slightly higher
Derivation AEstimating a square root

The goal is to turn a bare, non-perfect-square number like x=70x=70 into a one-decimal estimate of x\sqrt{x} using only perfect squares you can recall or compute, without a calculator or the long-division algorithm, and to trust that estimate because you can check it in seconds by squaring.

Step 1. Bracket xx between two consecutive perfect squares a2<x<b2a^2<x<b^2, with b=a+1b=a+1. This immediately gives a<x<ba<\sqrt{x}<b, the whole-number part of the answer.

Step 2. Compare the two gaps x−a2x-a^2 and b2−xb^2-x. Whichever is smaller tells you which perfect square xx sits closer to, and hence whether x\sqrt{x} leans toward aa or toward bb.

Step 3. Apply the linear estimate x≈a+x−a2b2−a2\sqrt{x}\approx a+\dfrac{x-a^2}{b^2-a^2} to convert that lean into a one-decimal number.

Step 4. Verify by squaring one or two trial values close to the estimate and checking that xx falls between their squares — this catches arithmetic slips and confirms the estimate is in the right range.

Worked example

Estimate 70\sqrt{70} to one decimal place, then use the same method on 200\sqrt{200}.

Step 1. Bracket 7070: we know 82=648^2=64 and 92=819^2=81, and 64<70<8164<70<81, so 8<70<98<\sqrt{70}<9.

Step 2. Compare the gaps: 70−64=670-64=6 and 81−70=1181-70=11. Since 6<116<11, 7070 is closer to 6464, so 70\sqrt{70} leans toward 88, not 99 — a first guess around 8.38.3–8.48.4.

Step 3. Linear estimate: 70≈8+70−6481−64=8+617≈8+0.35=8.35\sqrt{70}\approx 8+\dfrac{70-64}{81-64}=8+\dfrac{6}{17}\approx 8+0.35=8.35.

Step 4. Verify by trial squaring: 8.32=68.898.3^2=68.89 (too low), 8.42=70.568.4^2=70.56 (too high). Since 7070 sits between these two squares, the estimate of about 8.48.4 is confirmed as reasonable, and the check also shows the true value must lie between 8.38.3 and 8.48.4. 70≈8.4\sqrt{70}\approx 8.4 (the true value to two decimal places is 8.378.37). This four-step routine is worth practising until it is automatic: bracket, compare gaps, apply the fraction, then square-check. The check in Step 4 is not optional busywork — it is what turns a guess into a defensible estimate, because it independently confirms the bracket without relying on the linear formula being exact.

Competitive stretch

The same four steps on x=200x=200: 142=19614^2=196 and 152=22515^2=225 bracket it, so 14<200<1514<\sqrt{200}<15. The gaps are 200−196=4200-196=4 and 225−200=25225-200=25, so 200200 sits very close to 196196, and the linear estimate gives 14+429≈14.1414+\dfrac{4}{29}\approx 14.14. Checking, 14.12=198.8114.1^2=198.81 and 14.22=201.6414.2^2=201.64 bracket 200200, confirming it. 200≈14.1\sqrt{200}\approx 14.1, between the whole numbers 1414 and 1515.

Common traps

  • Rounding to the nearer perfect square instead of actually comparing x−a2x-a^2 with b2−xb^2-x — a number can look "close to" the bigger square by eye and still be numerically closer to the smaller one.
  • Treating the linear-estimate answer as exact — it is an approximation good to about one decimal place; only long division (within the syllabus's digit cap) gives the precise value.
  • Skipping the trial-squaring check — this one step catches most arithmetic slips immediately and costs almost no extra time.
  • Confusing "estimate this square root" with "prove this number is irrational" — Class 8 asks only for a numerical estimate; classifying 2\sqrt2, 3\sqrt3, etc. as irrational is Class 9 content and should not be mixed in here.

Ask the tutor: Not sure which two perfect squares to bracket your number between — ask the AI tutor to walk through estimating any square root step by step.

Why does the linear estimate formula work even though it is not the real relationship between x and its square root?

Over a short stretch, like from one perfect square to the very next one, the curve y=xy=\sqrt{x} bends only a little, so a straight line joining its two endpoints stays close to it. That is exactly what the linear formula does: it draws a straight line from (a2,a)(a^2,a) to (b2,b)(b^2,b) and reads off a value along it. It is not exact — the true curve dips slightly above the line — but the error is small enough for a one-decimal classroom estimate.

How is estimating a square root different from finding it by long division?

Estimating gives you a quick, one-decimal-ish approximation using only perfect squares you already know, with no written algorithm required — useful for sanity checks and mental maths. Long division gives an exact decimal value, digit by digit, following a fixed procedure, and is what you use when a precise answer (up to the syllabus's decimal-place cap) is actually required. Estimation is the fast first pass; long division is the accurate final answer.

If I estimate square root of 70 as 8.4, is that the same as saying square root of 70 is irrational?

No, and it is important to keep the two ideas apart. Estimating a square root just means giving its approximate decimal value using the bracket-and-refine method above — nothing about the number's type is claimed. Proving that a number like 70\sqrt{70} is irrational, meaning it can never be written exactly as a terminating or repeating decimal, is a separate idea that Class 9 covers formally. In Class 8, stick to producing and verifying a numerical estimate.

Before this

Where this leads