A student writes a formula that works for five examples. The numbers look convincing. The pattern seems obvious.
Does that prove the formula is always true?
Not necessarily.
One carefully chosen counterexample can show that a universal mathematical claim is false, even after many ordinary examples have supported it.
This is a compelling reason to study Additional Mathematics: students learn to question the conditions behind a rule instead of treating a familiar-looking pattern as a universal truth.
Counterexamples are especially valuable in algebra, logarithms, trigonometry, graphs and calculus. They can reveal why a shortcut is invalid, which inputs have been excluded and how a statement should be corrected.
The goal is not to become suspicious of every mathematical result. It is to understand the difference between a claim, supporting examples, a proof and a case that disproves a claim.
For Singapore students preparing for more advanced Mathematics, that distinction supports accurate problem-solving and clearer explanations.
What is a counterexample?
Consider the claim:
“Every real number has a positive square.”
This is a universal claim.
It uses the word “every”.
Try the number zero.
Its square is:
0² = 0.
Zero is not positive.
Therefore the claim is false.
A single counterexample is enough because the original statement claimed something about every real number.
We do not need to test infinitely many values to reject it.
A corrected statement is:
“The square of every real number is non-negative.”
That version includes positive squares and zero.
The difference between positive and non-negative matters.
Mathematics requires exact language, even when a statement appears nearly correct.
Many successful examples do not prove a universal claim
Suppose a student checks:
1² > 0.
2² > 0.
3² > 0.
4² > 0.
All are true.
The student may conclude:
“Therefore x² > 0 for every real x.”
That conclusion is invalid because x = 0 gives x² = 0.
The successful tests used only positive non-zero inputs.
They did not cover the complete real domain.
To establish a universal claim, we need an argument that applies to every permitted input, or an appropriate theorem.
To disprove a universal claim, we only need one valid input where it fails.
Understanding this asymmetry is fundamental to mathematical reasoning.
Additional Mathematics gives students opportunities to practise it.
A false square-root shortcut
Consider the proposed rule:
√(a + b) = √a + √b,
for non-negative real a and b.
It may look plausible because both sides contain square roots.
Test:
a = 1,
b = 1.
The left side is:
√(1 + 1) = √2.
The right side is:
√1 + √1 = 1 + 1 = 2.
But:
√2 ≠ 2.
Therefore the proposed rule is false.
This is a counterexample.
A student should not use the equation as a general simplification step.
The correct algebraic rules for square roots do not allow ordinary distribution across addition.
Why the square-root shortcut fails
For non-negative a and b:
(√a + √b)² = a + 2√(ab) + b.
The middle term is generally non-zero.
But:
[√(a + b)]² = a + b.
Thus the squared expressions generally differ by:
2√(ab).
For a and b both positive, that extra term is positive.
The two sides therefore cannot be equal.
They do agree in special cases such as a = 0 or b = 0.
That does not make the original universal claim true.
One true special case is not proof of a general identity.
A-Math students benefit from checking both an easy numerical counterexample and the algebraic reason behind it.
A wrong logarithm identity
Consider the claim:
ln(a + b) = ln a + ln b,
for all positive real a and b.
Choose:
a = 1,
b = 1.
The left side is:
ln(1 + 1) = ln 2.
The right side is:
ln 1 + ln 1 = 0 + 0 = 0.
Since ln 2 is not zero, the claim is false.
The correct product identity is:
ln(ab) = ln a + ln b,
for positive a and b.
The logarithm turns multiplication into addition; it does not turn ordinary addition into addition of logarithms.
The counterexample identifies a common shortcut that would corrupt many exponential-equation solutions.
Why a calculator check is useful but not the full proof
Suppose a student tests the true identity:
ln(ab) = ln a + ln b
with a = 2 and b = 3.
The numbers agree.
That confirms the identity for that pair.
But testing one pair does not prove it for all positive a and b.
A general explanation comes from exponential laws.
Let:
a = e^u,
b = e^v.
Then:
ab = e^u e^v = e^(u+v).
Taking natural logarithms gives:
ln(ab) = u + v.
Since u = ln a and v = ln b:
ln(ab) = ln a + ln b.
The positivity conditions permit those real logarithms.
This illustrates the difference between numerical verification and general mathematical justification.
The square of a number can lose its sign
Consider:
x² = 9.
The real solutions are:
x = 3,
and:
x = -3.
Both square to nine.
Now examine the claim:
“If x² = y², then x = y.”
Choose:
x = 3,
y = -3.
Their squares are equal:
3² = (-3)² = 9.
But x and y are not equal.
Therefore the claim is false.
The correct implication is:
x² = y²
implies:
x = y or x = -y.
Equivalently:
|x| = |y|.
Squaring can remove sign information, so reversing the operation requires care.
Why √(x²) is not always x
A related false claim is:
√(x²) = x for every real x.
Take:
x = -4.
Then:
x² = 16.
The principal square root is:
√16 = 4.
But x = -4.
Therefore:
√(x²) ≠ x
for this counterexample.
The correct identity over the real numbers is:
√(x²) = |x|.
If x ≥ 0, this equals x.
If x < 0, it equals -x.
The domain and sign condition matter.
Students who automatically cancel a square and square root without considering the principal root can lose valid cases or introduce incorrect signs.
Squaring can create an extra solution
Solve:
√(x + 5) = x – 1
over the real numbers.
The principal square root is non-negative, so the right side must also be non-negative:
x – 1 ≥ 0.
Therefore x ≥ 1.
Square both sides:
x + 5 = (x – 1)².
Expand:
x + 5 = x² – 2x + 1.
Rearrange:
x² – 3x – 4 = 0.
Factorise:
(x – 4)(x + 1) = 0.
The algebraic candidates are x = 4 and x = -1.
But x = -1 violates x ≥ 1 and does not satisfy the original equation.
The only real solution is x = 4.
The squaring step produced a candidate that was not valid in the original problem.
A counterexample can expose a lost condition
Imagine the student claims:
“Every solution obtained after squaring an equation is a solution of the original equation.”
The previous problem supplies a counterexample.
After squaring, x = -1 satisfies the derived quadratic.
But substituting x = -1 into the original equation gives:
√4 = -2.
The left side is 2.
The right side is -2.
They are not equal.
Therefore x = -1 is not a solution.
The correct habit is to check all candidates in the original equation after an operation that may not preserve equivalence in both directions.
A-Math requires students to distinguish an implication from an equivalent transformation.
What does equivalence mean?
Two equations are equivalent over a specified domain when they have the same solution set within that domain.
For example:
x – 3 = 0
and:
2x – 6 = 0
are equivalent over the real numbers because multiplying by the non-zero constant two preserves the solution set.
But squaring both sides of an equation can add candidates.
Dividing by an expression that might equal zero can remove candidates.
An equation obtained by a manipulation is not automatically equivalent to the original.
The student should ask whether every step works both forwards and backwards under the stated conditions.
This is a useful form of mathematical discipline.
It prevents a neat-looking final equation from replacing the actual question.
Dividing by a variable may lose a solution
Consider:
x(x – 1) = x.
A student might divide both sides by x and obtain:
x – 1 = 1.
Therefore:
x = 2.
But this division assumes x ≠ 0.
The original equation also holds at x = 0:
0(0 – 1) = 0.
So x = 0 is another solution.
A safer method is to bring everything to one side:
x(x – 1) – x = 0.
Factor:
x(x – 2) = 0.
Thus:
x = 0 or x = 2.
The correct solution set has two values.
The division shortcut lost one because it silently excluded x = 0.
When division by an expression is legitimate
Suppose we have:
ab = ac.
Can we cancel a and conclude b = c?
Only when a is non-zero.
If a = 0, the equation becomes:
0 = 0,
which holds for every b and c.
It does not imply b = c.
For a counterexample, choose:
a = 0,
b = 2,
c = 7.
Then:
ab = 0,
ac = 0.
The products are equal.
But:
b ≠ c.
This disproves the unrestricted cancellation claim.
The corrected statement requires a ≠ 0.
A-Math students should learn to preserve such non-zero conditions when simplifying algebraic fractions and equations.
A denominator restriction survives cancellation
Consider:
f(x) = (x² – 4)/(x – 2).
Factorise:
f(x) = (x – 2)(x + 2)/(x – 2).
For x ≠ 2, cancel the common factor:
f(x) = x + 2.
But the original expression is undefined at x = 2.
The simplified polynomial x + 2 is defined there.
Therefore the two formulas describe the same values only on the restricted original domain x ≠ 2.
A student who claims the original function includes the point (2, 4) has introduced an invalid value.
The counterexample x = 2 shows why domain conditions cannot be discarded with cancelled algebraic factors.
A quadratic need not have real roots
Consider the claim:
“Every quadratic equation has two real roots.”
Choose:
x² + 1 = 0.
For every real x:
x² ≥ 0.
Therefore:
x² + 1 > 0.
It cannot equal zero.
So the equation has no real solutions.
The claim is false.
A different quadratic:
(x – 2)² = 0
has one distinct real root, x = 2, with multiplicity two.
And:
x² – 1 = 0
has two distinct real roots, x = -1 and x = 1.
The correct classification depends on the discriminant.
Students should not assume that the degree alone determines the number of distinct real solutions.
A repeated root is not two distinct points
For:
(x – 2)² = 0,
the factor x – 2 occurs twice.
We say x = 2 has multiplicity two.
But there is still only one distinct real x-value satisfying the equation.
A graph of y = (x – 2)² touches the x-axis at (2, 0).
It does not meet the axis at two separate points merely because the root is repeated.
This distinction matters when an examination asks for the number of roots or intersections.
A-Math teaches students to read the exact wording.
“Two roots counted with multiplicity” and “two distinct real roots” are not identical claims.
A counterexample such as this one can correct a common overgeneralisation.
A zero derivative does not guarantee a maximum
Consider the claim:
“If f'(a) = 0, then f has a local maximum at x = a.”
Choose:
f(x) = x³.
Differentiate:
f'(x) = 3x².
At x = 0:
f'(0) = 0.
But the function x³ increases through the origin.
It does not turn downwards there.
Therefore x = 0 is not a local maximum.
It is a stationary point of inflexion.
The counterexample disproves the original universal claim.
The correct approach is to examine derivative signs or use a suitable classification test under its conditions.
Finding f'(a) = 0 identifies a stationary candidate; it does not fully classify it.
A second-derivative test also has conditions
Suppose:
f'(a) = 0.
If f”(a) > 0, the usual second-derivative test gives a local minimum under appropriate differentiability conditions.
If f”(a) < 0, it gives a local maximum.
But what if:
f”(a) = 0?
The test is inconclusive.
Consider:
f(x) = x⁴.
At zero:
f'(0) = 0,
f”(0) = 0.
Yet x = 0 is a local minimum.
Now consider:
g(x) = x³.
At zero:
g'(0) = 0,
g”(0) = 0.
But it is a stationary point of inflexion.
The same first- and second-derivative values can correspond to different behaviour.
Students should not turn an inconclusive test into a conclusion.
A larger derivative does not mean a larger function value
Suppose someone claims:
“If f'(x) is larger than g'(x) at a point, then f(x) must also be larger than g(x) there.”
Consider:
f(x) = 100 + x,
and:
g(x) = 2x.
At x = 0:
f(0) = 100.
g(0) = 0.
But the derivatives are:
f'(0) = 1.
g'(0) = 2.
Thus:
g'(0) > f'(0),
while:
g(0) < f(0).
The function with the greater rate has the smaller current value.
The claim is false.
A derivative describes local change, not the original function’s absolute height.
This distinction is important in motion, growth and optimisation.
Two functions can have the same derivative
Consider:
f(x) = x² + 10,
and:
g(x) = x² – 5.
Differentiate:
f'(x) = 2x.
g'(x) = 2x.
The derivatives are equal for every real x.
But:
f(x) – g(x) = 15.
The original functions differ by a constant.
Therefore knowing the derivative does not determine one unique original function without an additional condition.
This is why indefinite integration introduces + C.
The counterexample corrects the claim:
“Equal derivatives always mean equal functions.”
The accurate statement on a connected interval is that functions with equal derivatives differ by a constant, under the usual differentiability assumptions.
Odd degree does not mean exactly one real root
A student sees a cubic:
x³ + x = 0.
It has one real root, x = 0.
The student then claims:
“Every odd-degree polynomial has exactly one real root.”
Consider:
x³ – x = 0.
Factorise:
x(x² – 1) = 0.
Therefore:
x(x – 1)(x + 1) = 0.
The distinct real roots are:
x = -1, 0, 1.
There are three.
The claim of exactly one real root is false.
A corrected general statement is that every real polynomial of odd degree has at least one real root.
That follows from continuity and opposite end behaviours.
It does not say there is only one.
A single graph window can hide a counterexample
Suppose someone claims that a polynomial is positive everywhere because the calculator shows only positive outputs on a narrow range.
Consider:
f(x) = x⁴ – 100x² + 1.
At x = 0:
f(0) = 1 > 0.
Near enough to zero, some values may remain positive.
But at x = 1:
f(1) = 1 – 100 + 1 = -98.
The claim is false.
The graph’s appearance depends on the displayed window.
A-Math students should not mistake a finite picture for proof of a statement about every real input.
An algebraic counterexample can be found even when the first few plotted values look supportive.
Two functions agreeing at two points need not be identical
Consider:
f(x) = x,
and:
g(x) = x².
At x = 0:
f(0) = g(0) = 0.
At x = 1:
f(1) = g(1) = 1.
The two functions agree at both tested inputs.
But at x = 2:
f(2) = 2.
g(2) = 4.
They differ.
Therefore matching two observations does not prove that two general functions are identical.
The same warning applies to mathematical models fitted to a small data set.
Agreement with a limited number of points may be useful evidence, but it is not automatically a proof of equality for all inputs.
The domain and functional structure still matter.
A trigonometric counterexample
Consider the proposed identity:
sin(A + B) = sin A + sin B.
Choose:
A = 30°,
B = 30°.
The left side is:
sin 60° = √3/2.
The right side is:
sin 30° + sin 30° = 1/2 + 1/2 = 1.
These values differ.
Therefore the proposed identity is false.
The correct sum formula is:
sin(A + B) = sin A cos B + cos A sin B.
The counterexample can reveal the wrong rule.
A general proof of the correct formula requires appropriate mathematical reasoning, such as unit-circle rotation geometry.
This contrast between disproving and proving is especially useful in trigonometry.
A product of increasing functions need not increase
Suppose someone claims:
“If two real functions are increasing, their product is always increasing.”
Consider the functions:
f(x) = x,
g(x) = x.
Both are strictly increasing on the entire real line.
Their product is:
h(x) = x².
But x² decreases as x increases from -2 to -1.
For example:
h(-2) = 4,
and:
h(-1) = 1.
The input increased from -2 to -1, but the product decreased.
The unrestricted claim is false.
The issue is the sign of the factors.
Positive increasing functions behave differently from functions that can take negative values.
Conditions about positivity may be essential when establishing a product-monotonicity statement.
Correct the product claim with a useful condition
Suppose f and g are non-negative and non-decreasing on the same interval.
For any a < b in that interval:
0 ≤ f(a) ≤ f(b),
and:
0 ≤ g(a) ≤ g(b).
Because the factors are non-negative:
f(a)g(a) ≤ f(b)g(a).
And because f(b) is non-negative:
f(b)g(a) ≤ f(b)g(b).
Combine:
f(a)g(a) ≤ f(b)g(b).
Therefore the product is non-decreasing.
The positivity assumptions make the proof work.
The earlier counterexample showed that removing them can invalidate the conclusion.
This is how a false universal statement can be repaired by identifying the missing conditions.
An implication is not always reversible
Consider the true statement:
“If x = 2, then x² = 4.”
Does the converse hold?
“If x² = 4, then x = 2.”
No.
The counterexample x = -2 satisfies x² = 4 but not x = 2.
The original implication is true.
Its converse is false.
This is a basic logical distinction.
A-Math students encounter it when squaring equations, using derivatives and solving conditions involving roots.
A method may establish that one property implies another without proving they are equivalent.
Before reversing an argument, ask whether the converse has been justified.
Necessary and sufficient conditions
A condition is sufficient when it guarantees a conclusion.
A condition is necessary when the conclusion cannot hold without it.
For example, consider a real quadratic equation:
ax² + bx + c = 0,
with a ≠ 0.
The condition:
b² – 4ac = 0
is necessary and sufficient for the quadratic to have a repeated real root.
But for a differentiable function, the condition f'(a) = 0 is not by itself sufficient for a local maximum.
It can occur at a minimum or stationary inflexion too.
Learning to identify the exact scope of a condition helps students avoid unjustified reversals.
It also clarifies what an examination question is asking the student to prove.
A counterexample must satisfy the original assumptions
Suppose the claim is:
“For all positive real numbers x, x² > 0.”
A student suggests x = 0 as a counterexample.
But zero is not positive.
It lies outside the stated domain.
Therefore it does not disprove the claim.
In fact, if x > 0, then x² > 0.
A valid counterexample must satisfy every assumption of the claimed statement while making the conclusion false.
This is an essential rule.
Choosing an invalid input may create an apparent contradiction without challenging the actual theorem.
A-Math trains students to read assumptions as carefully as the formula itself.
How to build a valid counterexample
Use a systematic process.
First, identify the universal claim.
Second, write its domain and all stated conditions.
Third, find a candidate within those conditions.
Fourth, evaluate the claim exactly at that candidate.
Fifth, show clearly which equality, inequality or conclusion fails.
For example, to disprove:
“Every real x satisfies √(x²) = x,”
choose x = -1.
It is real, so it belongs to the domain.
The left side is √1 = 1.
The right side is -1.
They differ.
That is a complete counterexample.
The explanation is short because the counterexample directly addresses the universal statement.
Why proof needs a different strategy
A counterexample disproves a universal claim.
It does not prove a corrected claim automatically.
Suppose we propose:
x² ≥ 0 for every real x.
Testing x = -2, 0 and 5 supports it.
But the universal justification is that the square of any real number is non-negative.
More elaborate statements may require factorisation, completing the square, calculus or a recognised theorem.
For example, consider:
x² – 4x + 5 > 0
for every real x.
Complete the square:
x² – 4x + 5 = (x – 2)² + 1.
The square is non-negative, so the expression is at least one and therefore strictly positive.
That proves the claim for all real x.
No finite list of test points is needed.
Counterexamples and mathematical models
Suppose a simple model claims:
“Every quantity that grows during the first two observations continues increasing forever.”
We can construct mathematical counterexamples.
For instance:
f(t) = 4t – t²,
for t ≥ 0.
The function increases on 0 < t < 2.
Then it decreases for t > 2.
Its derivative is:
f'(t) = 4 – 2t.
The derivative becomes negative after t = 2.
So early growth does not guarantee indefinite growth.
This is an algebraic counterexample to the broad claim.
Whether the function represents a realistic physical process is a separate empirical question.
The value of the example is showing that the logical inference is not valid from the limited information supplied.
Mathematical disproof and scientific evidence are different
In pure Mathematics, a universal statement can be disproved by one valid counterexample.
In empirical science, observations may have measurement uncertainty, sampling limitations and multiple possible explanations.
One surprising observation may justify investigating or revising a model.
But deciding whether an empirical claim has been falsified can require more context than simply substituting a number into an equation.
Additional Mathematics gives students precise reasoning tools.
It should not encourage them to treat every noisy measurement as a logically exact counterexample to a scientific law.
The relevant assumptions, definitions and quality of evidence still matter.
This distinction supports responsible interpretation of data and models.
A five-session learning sequence
Session one: identify universal claims
Write several statements using “every”, “always” and “for all”.
Decide which claims have clearly stated domains.
Test simple integer and zero cases where appropriate.
Session two: repair false algebraic rules
Use √(a + b) versus √a + √b.
Use ln(a + b) versus ln a + ln b.
Develop a numerical counterexample for each and derive the correct law.
Session three: investigate reversible operations
Study x² = y², √(x²) and an equation that gains a candidate after squaring.
Practise checking solutions in the original equation.
Session four: test calculus claims
Compare x², x³ and x⁴ at stationary points.
Explain why a zero first derivative does not determine a maximum.
Session five: prove a corrected statement
Take an initially false claim and identify the condition needed to repair it.
For example, explore when a product of increasing functions is also non-decreasing.
Finish with a general proof rather than a list of numerical tests.
This routine develops both the ability to disprove and the ability to justify.
Four common counterexample mistakes
Mistake one: using an input outside the domain
The student offers x = 0 against a claim restricted to positive x.
The repair is to check all assumptions first.
Mistake two: treating several successful tests as a proof
The learner tests five inputs and declares a universal identity established.
The repair is to use a general algebraic argument.
Mistake three: finding a suspicious example but not checking it
The student chooses values without demonstrating which part of the claim fails.
The repair is to substitute and show the contradiction explicitly.
Mistake four: confusing the converse with the original implication
The learner assumes that because P implies Q, Q must imply P.
The repair is to test the converse separately and identify whether additional conditions are needed.
Each mistake involves logical structure as much as calculation.
What parents can ask during revision
Ask the student what the word “always” means mathematically.
Ask whether one valid counterexample can disprove such a claim.
Ask whether their chosen input belongs to the original domain.
Ask how they know a calculator test proves or disproves something.
Ask whether a transformation has preserved every original solution.
Ask whether the converse of a statement has been checked.
These prompts encourage precise thinking.
A learner who can explain why a shortcut fails is better equipped to avoid repeating it than one who merely memorises that the shortcut is forbidden.
Singapore’s Additional Mathematics context
The official 2027 Singapore-Cambridge Secondary Education Certificate G3 Additional Mathematics syllabus K341 includes mathematical reasoning, algebraic identities, domain restrictions, quadratic-root conditions, graphs, trigonometric identities and calculus.
The logical language of counterexamples and necessary versus sufficient conditions supports those topics, although advanced formal logic is not a separately listed G3 Additional Mathematics chapter.
For the precise assessed scope, consult the SEAB 2027 G3 syllabus directory.
These examples are designed to strengthen reasoning, not predict particular examination questions.
How a three-student tutorial can help
One learner may apply square-root rules incorrectly.
Another may divide by a variable and lose a valid root.
A third may differentiate correctly but assume every stationary point is a maximum.
All three need different explanations.
In a carefully managed three-student Additional Mathematics tutorial, the tutor can ask the learner to test a claim, identify the missing condition and then solve a changed problem independently.
The immutable Secondary 1 Mathematics small-group tutorial in Clementi illustrates why careful algebra and mathematical explanation matter before advanced reasoning tasks.
Students already progressing independently may not need tuition. Additional help should focus on an identified misconception.
Frequently asked questions
What is a mathematical counterexample?
It is a valid input or case that satisfies the assumptions of a universal claim but makes its conclusion false.
How many counterexamples are needed to disprove a universal statement?
One valid counterexample is sufficient.
Can testing many examples prove an identity?
Not generally. A universal identity needs an argument that covers every permitted input.
Why must the domain be checked first?
An input outside the domain does not disprove a statement that was never claimed to apply there.
Does a zero derivative always mean a turning point?
No. The function x³ has derivative zero at the origin but continues increasing through it.
Why can squaring an equation produce false answers?
Because squared equality can hold for values with opposite signs, even when the original equation required equal signed expressions.
How does this help in an A-Math examination?
It supports checking conditions, avoiding invalid transformations, distinguishing proof from examples and explaining why an answer follows.
The lasting reason to test a claim
Additional Mathematics develops the ability to distinguish a true pattern from an unjustified assumption.
A familiar-looking identity may fail for a simple input.
A valid operation may need a non-zero or positivity condition.
A stationary point may be a minimum, maximum or neither.
Two functions may share a derivative without sharing the same values.
A counterexample makes these distinctions visible.
It teaches students to ask not only, “Does this work?” but also, “Under which conditions must it work?”
That is a durable mathematical habit.
For connected reading, explore Why Study Additional Mathematics | Explain Why a Mathematical Claim Is True, Why Study Additional Mathematics | Know When an Algebraic Answer Is Invalid, Why Study Additional Mathematics | Recognise Stationary Points That Are Not Turning Points and the eduKateSG Additional Mathematics project library.