The Difference of Two Squares
Multiply 43 \times 37 in your head — faster than reaching for a
calculator. The two numbers sit either side of 40, and a single
pattern collapses the whole thing to 40^2 - 3^2 = 1591. That shortcut
is the difference of two squares, and the very same pattern runs quietly through
simplifying fractions, solving equations and even calculus.
Expand (x + 3)(x - 3) and watch closely. Using
FOIL
— first, outer, inner, last — you get four terms:
(x + 3)(x - 3) = x^2 \;\underbrace{-\,3x + 3x}_{\text{annihilate!}}\; - 9 = x^2 - 9
The two middle terms are -3x and +3x —
equal and opposite, so they annihilate each other and vanish without trace.
What's left is strangely clean: a square minus a square, with no x-term
in the middle at all. That "missing middle" is the fingerprint of this pattern.
Here's why it matters: you can run the trick backwards. Any time you spot one square
subtracted from another square — x^2 - 9,
4x^2 - 25, even the plain number 51 \times 49
in disguise — it factorises instantly, in a single step, no trial and error:
a^2 - b^2 = (a + b)(a - b)
This is quite possibly the most-recycled identity in all of algebra. It
reappears when you simplify fractions, rationalise surds, solve equations, integrate in
calculus, and even in the mental-arithmetic tricks stage performers use. Learn to spot it on
sight and it will pay you back for years.
Watch the middle vanish — in general
One example could be a fluke, so let's do it with letters and see that the cancellation
always happens. Expand (x + a)(x - a) term by term:
\begin{aligned}
(x + a)(x - a) &= x \cdot x \;+\; x \cdot (-a) \;+\; a \cdot x \;+\; a \cdot (-a) \\
&= x^2 - ax + ax - a^2 \\
&= x^2 - a^2
\end{aligned}
The outer term -ax and the inner term +ax
are always equal and opposite — one bracket has +a, the
other -a, and that sign difference is precisely what kills the
middle. Compare this with (x + a)(x + a) = x^2 + 2ax + a^2, where
both middles have the same sign and pile up instead of cancelling. The minus sign is
doing all the work.
So the identity cuts both ways, and you should be fluent in both directions:
- Expanding: (x + 7)(x - 7)? Don't FOIL it —
just write x^2 - 49 straight down.
- Factorising: x^2 - 49? Don't hunt for two
numbers that multiply to -49 and add to 0
— just write (x + 7)(x - 7) straight down.
Why it works — an area picture
Algebra says the middle terms cancel; geometry shows you the identity with scissors and paper.
Start with a square of side a — its area is
a^2 — and snip a small square of side b
out of one corner. The L-shaped piece left over has area exactly
a^2 - b^2: that's the difference of two squares, made of
actual squares.
Now cut the L into two rectangles and slide the thin one round to the end. The pieces fit
together perfectly into a single rectangle that is (a + b) wide and
(a - b) tall. Nothing was added and nothing was thrown away — the
same two pieces, rearranged — so the two areas must be equal:
a^2 - b^2 = (a + b)(a - b). Step through the dissection below and
watch it happen.
The Greeks knew this picture over two thousand years ago — a version of it appears in Euclid's
Elements (Book II), written around 300 BC, long before anyone wrote algebra with
letters. You are learning one of the oldest tricks in mathematics.
Factorising: three levels of disguise
Factorising with this pattern is a recognition game: can you see both terms as perfect
squares? Take the square root of each term to find a and
b, then write the two brackets. The disguises get progressively
sneakier.
Level 1 — plain sight. x^2 - 49. Both terms are
obviously squares: x^2 = (x)^2 and
49 = 7^2. So:
x^2 - 49 = x^2 - 7^2 = (x + 7)(x - 7)
Level 2 — a coefficient in front. 4x^2 - 25. The
first term isn't just x^2, but it is still a perfect
square: 4x^2 = (2x)^2, because
(2x)^2 = 2x \times 2x = 4x^2. And
25 = 5^2. So here a = 2x and
b = 5:
4x^2 - 25 = (2x)^2 - 5^2 = (2x + 5)(2x - 5)
The key move at this level is recognising the whole first term as a square —
coefficient and all. 9x^2 = (3x)^2,
16x^2 = (4x)^2, 100x^2 = (10x)^2. Check
by squaring: (3x)^2 = 9x^2 ✓.
Level 3 — a square hiding inside a square. x^4 - 16.
At first glance neither term looks like the familiar pattern — but
x^4 = (x^2)^2 (a power with an even exponent is always a
perfect square) and 16 = 4^2. Apply the identity with
a = x^2, b = 4:
x^4 - 16 = (x^2)^2 - 4^2 = (x^2 + 4)(x^2 - 4)
Done? Not yet! Look at that second bracket: x^2 - 4
is itself a difference of two squares. Factorise again:
x^4 - 16 = (x^2 + 4)(x^2 - 4) = (x^2 + 4)(x + 2)(x - 2)
Two layers, peeled one at a time. The first bracket x^2 + 4 is a
sum of squares, so it stays put — three factors, and now it's fully factorised.
Whenever you finish a factorising pass, glance back at each bracket and ask: can any of
these split again?
These three mistakes catch students out year after year:
-
It's the difference of two squares — the minus sign is not optional.
A sum of two squares like x^2 + 9 does not
factorise (over the real numbers). Try it: (x + 3)(x + 3) = x^2 + 6x + 9
and (x - 3)(x - 3) = x^2 - 6x + 9 — every attempt produces a middle
term you can't get rid of. Only opposite signs make the middle annihilate.
-
Both terms must actually be perfect squares.
x^2 - 10 looks tempting, but 10 is not a
perfect square, so there's no whole-number factorisation. (Later you'll meet
\left(x + \sqrt{10}\right)\left(x - \sqrt{10}\right) using surds —
but at GCSE, "factorise" means nice whole numbers, and x^2 - 10
doesn't have them.) Perfect squares to know on sight:
1, 4, 9, 16, 25, 36, 49, 64, 81, 100, 121, 144.
-
One pass may not be the end. After factorising, check each bracket:
x^4 - 16 = (x^2 + 4)(x^2 - 4) is only half factorised —
the second bracket splits again into (x + 2)(x - 2). Exam mark
schemes routinely reserve the final mark for spotting the second layer.
The arithmetic party trick
The identity isn't just for algebra homework — it turns certain multiplications into mental
arithmetic. Try 51 \times 49. It looks like long-multiplication
territory, until you notice the two numbers sit symmetrically around
50:
51 \times 49 = (50 + 1)(50 - 1) = 50^2 - 1^2 = 2500 - 1 = 2499
In your head, in two seconds. It works whenever two numbers straddle an easy middle:
73 \times 67 = (70 + 3)(70 - 3) = 4900 - 9 = 4891
It also runs the other way — subtracting big squares without squaring anything. Asked
for 102^2 - 98^2, do not compute 102^2 = 10404
and 98^2 = 9604 and subtract. Factorise instead:
102^2 - 98^2 = (102 + 98)(102 - 98) = 200 \times 4 = 800
A sum, a difference, one easy multiplication. Examiners love setting these precisely because
the "long way" is slow and error-prone while the factorised way is instant — a question like
7.5^2 - 2.5^2 = 10 \times 5 = 50 takes longer to read than to solve.
With this very identity! Before cheap calculators (well into the 20th century), engineers and
accounting clerks used books of quarter-square tables. The trick: rearrange the
difference of two squares into a multiplication machine,
a \times b = \frac{(a + b)^2 - (a - b)^2}{4}
(expand the two squares and check it — everything cancels except 4ab,
divided by 4). So to multiply any two numbers, you looked up the quarter-square
\tfrac{n^2}{4} of their sum and of their difference in the table, and
subtracted. One subtraction instead of a whole long multiplication — and lookup tables
of squares are easy to print. Stage "lightning calculators" used the same idea live: asked for
996 \times 1004, they'd silently compute
1000^2 - 4^2 = 1{,}000{,}000 - 16 = 999{,}984 and answer before the
audience finished writing the question down.
See it explained
Sal Khan walks through the same pattern from the expanding side — multiply
(a + b)(a - b), watch the middle cancel, then use the result to
factorise examples. A good second voice if the idea hasn't clicked yet: