Upper Primary

How the Square Root

Algorithm works

ClassRoom
O
AMIT CHAND ne of the algorithms taught at the elementary
level is deriving the square root of a number using
the long division method. Like many algorithms
taught in high school, the focus is on the ‘how’ of
performing the steps rather than the ‘why’ these steps
produce the square root. This article focuses on developing
conceptual understanding of the Square Root Algorithm.
The difference between the factorization method and the
division algorithm is that the former gives only the exact
value of the square root of a whole number which is a perfect
square whereas the latter may be used to find the square root
of any positive number. Secondly, while factorization works
for numbers with small factors, it becomes tedious when
numbers have large prime factors.
√
We will find 576 by the division algorithm and in the
process, we will raise some questions.

Figure 1

Step 1: Make pairs of digits from the right, in this case we

get 5 and 76.
Step 2: Select the left most pair (5 here) and find the greatest
number (2 here) whose square is equal to or less than
it. Subtract this square from the selected pair. Write 2
on the left as well as above (as part of the result). This
is the first digit which will appear in the quotient.

Keywords: algorithm, square root, reasoning, visualization,

understanding

Azim Premji University At Right Angles, March 2021 29

 Step 3: Write the second pair from the left (76 here) on the right side of subtracted result (1 here), to get a
new dividend (176 here).
Step 4: Double the first number (the quotient 2) and write it down on the left of the new dividend.
Step 5: Now, by trial and error find the greatest digit __ such that (4__× __) ≤ 176. Write this second
new digit (4 here) on the right side of 2 in the quotient, now the quotient is 24.
Step 6: Subtract this product (4__× __) from the dividend. Repeat steps 3 to 6 until there are no more
pairs to be brought down.
Step 7: If the last difference is zero, the square root is the quotient (here the square root of 576 is 24). If
not, append pairs of zeroes and find the square root to the required degree of accuracy. At the
same time place a decimal point to the right of the quotient at this stage. Future entries will go to
the right of the decimal point.

Questions raised here (and often not addressed)

• Why do we make such pairs, and why do we start making pairs from the right?
• How do we make pairs if the number has a decimal part?
• Why do we double the quotient each time and write it on the left as part of the divisor?
• Why do we append a new digit to the divisor and why do we append the same digit to the
quotient and multiply the new divisor with the appended digit?

Theoretical part of Algorithm

The algorithm is based on estimating the largest square within the given number (with the help of
place-value) and then refining that estimate by adding layers to the estimated square (again with the help
of place-value) to get as close to the given number as possible.
There are three main important parts to understand the algorithm.
i. Understanding why we make pairs
ii. Understanding the idea of the formula (a + b)2 = a2 + 2ab + b2
iii. Understanding the iteration

i. Understanding why we make pairs

• Square of the smallest 2-digit number i.e. 10 is the smallest 3-digit number i.e. 100. So, any
square number of 1-digit or 2-digit must be the square of a 1-digit number
• Similarly, square of the smallest 3-digit number i.e. 100 is the smallest 5-digit number i.e. 10000.
So, any square number of 3-digit to 4-digit is the square of a 2-digit number
• Similarly, any square number of 5-digit to 6-digit is the square of a 3-digit number
• And any square number of 7-digit to 8-digit is the square of a 4-digit number and so on…
So, generally speaking, the square of the smallest n digit number (i.e. 10n−1 ) is the smallest (2n − 1)-digit
number (i.e. 102n−2 ). Therefore, a square number of (2n – 1)-digits or 2n digits must be the square of an
n-digit number. So, if the digits of a number of 2n – 1-digits or 2n digits are paired, then the number of
pairs provides the number of digits of the square root.

30 Azim Premji University At Right Angles, March 2021

Now, if the pairing is done from the left then sometimes the units digit would be left alone (for odd
number of digits e.g. for 729, 72 and 9) and otherwise it would be along with the tens digit (for even
number of digits e.g. for 1296, 12 and 96). This is not consistent. But if the pairing is done from the
right, then we get a consistent system since the units digit is always clubbed with the tens. Also, when we
pair the digits, we are essentially splitting the number in n parts as follows: 729 = 700 + 29, 1296 = 1200 +
96, 21316 = 20000 + 1300 + 16, 182329 = 180000 + 2300 + 29, etc.

ii. Application of the formula. (a + b)2 = a2 + 2 ab +b2

Figure 2

Any square number is of the form (a + b)2 = a2 + 2ab + b2 which also can be written as
a2 + (2a + b) b…(A)
(We are interested in square numbers ≥ 100)
In order to find the square root (a + b) from the square number (a + b)2 , we need to find an a and will
subtract a2 from a2 + (2a + b) b which will leave (2a + b) b. Then our work is to find such b for which
(2a + b) b holds.
Let us take an example to understand this part.
We have to find the square root of the number 4225.

Side=?
4225

Figure 3

From 4225, we have to find a square number of the form 100x2 (since 100 < 4225 < 10000)
The largest square of the form a2 which is a multiple of 100 and less than 4225 is 3600
Therefore, remaining area = 4225 − 3600 = 625

Azim Premji University At Right Angles, March 2021 31

 =
Figure 4

So. we need to find b such that

( )
2 × 60b + b2 = (120 + b)b = 625

Figure 5 Figure 6

The process to find b involves some trial and error, and with some educated guess, especially if the given
number is a perfect square.
We can check if b = 5 then (120 + 5) × 5 = 625 (the last digits give us a clue to find b).
To find the square root of numbers with 5 or more digits, iteration is required where a’ is the side of the
next layer of the square i.e. a + b and b’ is the 2nd iteration.

iii. Understanding iteration. We will understand the iteration in two cases. The first case will be with the
square of a 3-digit number and more. The later case will be with a number which is not a perfect square.

Case 1: Square of a 3-digit number. (The square will be between 10000 and 1000000.)
The 3 digit number may be written as 100x + 10 y + z (where x, y and z are single digit numbers). The first
objective is to find the first square of the form (100x)2 . So, we will find a number a = 100x, the square of
which may be 10000, 40000, 90000, 160000, 250000, 360000, 490000, 640000 or 810000.
The next part (orange in Figure 8) depends on the selection of b = 10y.
b should be chosen so that the orange part does not exceed the remaining part (as in Figure 9).
This orange and gold part together represent (2a + b) b = (2 × 100x + 10y) 10y

32 Azim Premji University At Right Angles, March 2021

 10000
10000 10000

Figure 7 Figure 8 Figure 9

′
Finally, take a = a +
( b ′= 100x + 10y and then select b’ = z to cover the remaining gray part (as in Figure
′) ′
8) which represents 2a + b b = {2 × (100x + 10y) + z}z
Now, (100x)2 + (2 × 100x + 10y) 10y + {2 × (100x + 10y) + z} z = (100x + 10y + z)2
Let us take an example (Figure 10) of finding the square root of 21316.

= = =

2 =

Figure 10

So, after choosing a suitable b = 10y = 40 in this example, we get a′ = a + b = 140

Now we are to find b′ .
′ ′ ′
Therefore, the square root is a + b = (a + b) + b = 146 after two iterations (in Figures 10-11).

Azim Premji University At Right Angles, March 2021 33

 = Remaining area
= 140b' + 140b' + (b')2
= (2 x 140 + b') b'
= (280 + b') b'

=
140

Remaining area
=140 +140 +
= (2× 140+
= (280+

Figure 11

This process can be continued for squares of 4-digit numbers, squares of 5-digit numbers and so on.

Case 2: Now we will see its use to find the square root of a number which is not a perfect square. The
iteration process is the same. Let us try to find the square root of 21320.
From the above case of finding square root of 21316, we can reach up to the step when 4 square area
remains.

14 6
21320
10000
240)11320
9600
286) 1720
1716
4

So, we will continue the iteration process by representing the number with decimal places like 21320.00
00 00…

34 Azim Premji University At Right Angles, March 2021

Here is the square root of 2

1.41421
1 2.000000000000
1
24 100
96
400
281 281
2824 11900
11296
60400
28282
56564
383600
282841
282841
2828423 10075900

ය 2 = 1.41421......

For decimal numbers, the same process of pairing digits will be followed in the same direction from the
whole number part to the decimal part of the number.
It is interesting to observe that the key theory behind this algorithm, i.e., (a + b)2 is integral to upper
primary algebra. However, even when its application, i.e., the division algorithm of finding the square root
is included in the syllabus or textbooks, the connection is rarely made. This is an attempt to fill that gap.

References
1. [Link]
2. [Link] picture only
3. [Link]
Each%20time%20the%20algorithm%20is,value%20of%20the%20square%20root.
4. [Link]
5. [Link]

AMIT CHAND is a Resource Person at Azim Premji Foundation. He has an [Link] in mathematics from Burdwan
University, Burdwan, West Bengal. He has been closely working with teachers on issues related to Mathematics
and conducts workshops focusing on conceptual as well as pedagogical strategies in teaching mathematics. He
has been doing mathematics with children for more than 6 years and is deeply interested in exploring the base of
mathematical ideas. He may be contacted at [Link]@[Link].

Azim Premji University At Right Angles, March 2021 35