How to compute square roots by hand using long division or newton's method for numbers like 123?

Checked on January 10, 2026
Disclaimer: Factually can make mistakes. Please verify important information or breaking news. Learn more.

This fact-check may be outdated. Consider refreshing it to get the most current information.

Executive summary

Two reliable hand methods exist to compute square roots: Newton’s method (the iterative averaging formula often called Heron’s method) quickly refines an initial guess and is what calculators use in practice [1] [2], while the classical digit‑by‑digit "long division" square‑root algorithm produces digits one at a time but is not described in the provided reporting; therefore its stepwise description is outside the dataset used here.

1. Newton’s method: the idea and the formula that drives convergence

Newton’s method finds roots of f(x)=0 by iterating x{n+1} = xn − f(x_n)/f′(x_n); for square roots set f(x)=x^2 − N which yields the familiar iteration x{n+1} = 0.5*(xn + N/x_n) — an averaging of the guess and N divided by the guess — a specialization historically attributed to Heron and widely documented as the standard algorithm used in calculators [3] [1] [2].

2. Why Newton’s method converges fast (and what can go wrong)

When f′ at the root is nonzero and f is smooth, Newton’s method converges quadratically: roughly doubling the number of correct digits each iteration once close enough to the true root; this is why only a handful of iterations usually give high accuracy [3]. Failure modes noted in standard texts include encountering an iterate where the derivative is zero or poor global behavior leading to a different root, but for positive N and the square‑root specialization these problems are rare with a reasonable initial guess [4] [3].

3. Hands‑on: computing √123 by Newton iterations (worked example)

Start by setting N = 123 and choosing an initial guess; a sensible integer guess is 11 because 11^2 = 121 is close to 123, then apply x{n+1} = 0.5*(xn + 123/x_n) repeatedly (the formula and approach are the standard Newton/Heron specialization [1] [2]). With x0 = 11, x1 = 0.5(11 + 123/11) ≈ 0.5(11 + 11.181818) ≈ 11.090909; x2 = 0.5(11.090909 + 123/11.090909) ≈ 11.0907225, and a third iteration yields x3 ≈ 11.0905365, which is already accurate to many decimals — this rapid improvement is the quadratic convergence behavior textbooks describe [3] [5].

**4. Practical stopping rules and numeric hygiene**

Texts and programming examples recommend stopping when successive iterates change by less than a chosen tolerance or when |guess^2 − N| is small relative to the magnitude of the guess (a relative tolerance) to avoid needless iterations; SICP and other instructional sources implement a relative‑tolerance test |guess^2 − N| < guess tolerance to balance small and large N [2] [5]. Implementations also guard against divides by zero and choose initial guesses (N or a nearby perfect square) that avoid pathological trajectories [6] [5].

5. The long‑division (digit‑by‑digit) square‑root algorithm: reporting limits and where to look

The classical digit‑by‑digit “long division” method yields one decimal (or integer) digit per pass and is valuable for extracting exact decimal digits by hand; however, the provided reporting and sources do not include a worked description or authoritative steps for that long‑division algorithm, so a sourced, stepwise manual procedure cannot be reproduced from this dataset — interested readers should consult a numerical methods or elementary arithmetic text for the annotated digit‑by‑digit algorithm (reporting limitation).

6. How to choose between methods in practice

For hand calculation aiming for a few decimal places, Newton’s method is simple, fast and requires only arithmetic and division; for formal digit‑by‑digit generation of every decimal digit without intermediate rounding, the long‑division style algorithm is appropriate but was not detailed in the provided sources [1] [2]. For programming or calculator work, Newton’s method is the standard because of its simplicity and quadratic convergence which is emphasized in numerical‑analysis and educational sources [1] [3].

Want to dive deeper?
How does the digit‑by‑digit long division square‑root algorithm work step by step?
What are common stopping tolerances and error bounds used when implementing Newton's method for square roots?
How did Heron, Babylonian methods, and Newton relate historically in the development of root‑finding algorithms?