Paper 2 tests five reasoning skills that no A-Level teaches and the MAT never assessed. Here is each one explained properly, with a full worked example: necessary vs sufficient, implication vs the converse, negating quantifiers, choosing a proof method, and finding exactly where a proof breaks.
Check where you stand, freeUpdated August 2026
Why this matters: our classification of every published TMUA question found that 43.8% of Paper 2 sits in the three reasoning topics, and that share has risen from 38.8% (2016–2019) to 48.8% (2020–2023). Reasoning terms alone — necessary and sufficient conditions, implication, equivalence — is 17.5% of the paper, the single largest subtopic in either paper. It is also the one part of the TMUA with no A-Level equivalent, and the MAT never assessed it either. Most candidates arrive at Paper 2 having never formally studied any of it.
This page teaches the five skills directly, in the order UAT-UK's own content specification sets them out, each with a full worked example built in the TMUA's own style. It is not a list of definitions to memorise — every idea here is followed by a question you have to actually answer, with the reasoning behind the answer shown in full, not just the letter.
None of the worked examples below are reproduced from an official TMUA paper. They are written by us, in the TMUA's style, to teach the underlying skill — the same way a tutor would build a question on a whiteboard. Two of them are built directly on the exact fallacies UAT-UK's own specification names as examples, which is disclosed where it happens.
Paper 2 draws on Section 2 of UAT-UK's content specification, which is short enough to read in ten minutes and precise enough to be worth reading. It has three named parts:
| Part | What it covers |
|---|---|
| The Logic of Arguments | True and false; and, or, not; if–then statements, including “A only if B” and “A if and only if B”; the converse and the contrapositive of a statement; necessary and sufficient conditions; the quantifiers “for all” and “for some”; negating any of the above. |
| Mathematical Proof | Following and constructing direct deductive proof, proof by cases, proof by contradiction, and disproof by counterexample; deducing implications from given statements; making and justifying conjectures from small cases; rearranging a jumbled sequence of statements into a correct proof; longer chains of reasoning. |
| Identifying Errors in Proofs | Spotting the exact step at which a purported proof breaks down; recognising common invalid deductions. |
One detail is worth knowing before you revise a single question: the specification states plainly that candidates will not be expected to recognise or use symbolic notation for any of these terms, and will not be expected to complete formal truth tables. Paper 2 is tested entirely in plain English. If a prep resource has you memorising ∀/∃ notation or drilling truth tables, that effort is aimed at the wrong target — understanding what the ideas mean is what is actually assessed.
A condition is sufficient for a statement if satisfying the condition guarantees the statement is true — though there might be other ways to make the statement true as well. A condition is necessary for a statement if the statement cannot possibly be true without it — though satisfying the condition on its own doesn't guarantee the statement holds. A condition that is both makes the two statements equivalent: each is true exactly when the other is.
The confusion almost always runs one direction: candidates who correctly spot that P guarantees Q assume P must also be required for Q, when a completely different route to Q might exist. The fix is mechanical, not conceptual — check both directions separately, every time, and never assume the second follows from the first.
Question. For a positive integer n, let P be the statement “n is divisible by 4” and Q be the statement “n is divisible by 2”. Which of the following correctly describes the relationship between P and Q?
Answer: B. Every multiple of 4 is a multiple of 2, so whenever P is true, Q must be true — P is a sufficient condition for Q. But Q does not force P: 6 is divisible by 2 but not by 4, so Q can be true while P is false, which means P is not necessary for Q.
“If A then B” is a one-way statement: A being true forces B to be true, but nothing is claimed about what happens when A is false. Two related statements are worth knowing by name, because the TMUA tests the difference directly:
This is one of the most useful facts on the whole paper. Proving a contrapositive is often much easier than proving the original statement directly, because you get to assume the negation of the conclusion and work forward from something concrete — see the proof-ordering example in the next section.
Question. Which of the following statements has the same truth value, for every shape, as “If a shape is a square, then it has four equal sides”?
Answer: B. B is the contrapositive of the original statement, and a conditional statement always shares its truth value with its contrapositive. A is the converse, and it doesn't hold in general — a rhombus has four equal sides without being a square, so A is false while the original statement is true. C asserts the full biconditional, which fails for the same reason as A. D means “having four equal sides rules out being a square”, which is false since squares themselves are a counterexample.
“For all” claims that something holds in every case, with no exceptions. “For some” (the specification's own phrasing for “for at least one”) claims only that at least one case exists. The two behave completely differently under negation, and negating one incorrectly is the single most common way a candidate loses easy marks on this part of the paper.
| Statement | Correct negation |
|---|---|
| For all x, P(x) | There exists an x such that P(x) is false |
| There exists an x such that P(x) | For all x, P(x) is false |
| If A then B | A and not B |
| A and B | Not A, or not B |
| A or B | Not A, and not B |
Notice what stays fixed: negating “for all” only ever produces a “there exists” statement, never another “for all” statement — and if the original statement restricts x to some domain (primes, positive integers, and so on), that restriction stays exactly as it was. Only the property being claimed gets negated.
Question. Which of the following is the correct negation of the statement “For every prime number p greater than 2, p is odd”?
Answer: B. Negating “for all x in a domain, P(x)” flips the quantifier to “there exists” and negates only P(x) — the domain itself (prime numbers greater than 2) is untouched. That rules out C, which wrongly renegotiates the domain too. A keeps the universal quantifier and just swaps odd for even, which is a stronger, different claim, not a negation. D asserts the negation as another universal statement (“no prime … is odd”), which is the classic over-negation error — the true negation only needs one even prime greater than 2 to exist, not for every such prime to be even.
The specification names four proof types candidates must be able to follow, and construct in simple cases. Picking the right one first saves most of the time a proof question costs:
| If the statement… | Try… |
|---|---|
| Follows in a short, direct chain from definitions | Direct deductive proof |
| Behaves differently across a small number of situations (e.g. odd/even, positive/negative/zero) | Proof by cases |
| Claims something is impossible, or that no such object exists | Proof by contradiction |
| Is a “for all” claim you suspect is actually false | Look for a counterexample first, before trying to prove it |
A direct proof is the simplest to follow: state what you're given, apply definitions and known results in a chain, and arrive at the conclusion. For example, to prove that the sum of any two odd integers is even: write the two integers as 2a + 1 and 2b + 1 for integers a and b; their sum is 2a + 2b + 2 = 2(a + b + 1), which is even by definition, since it is twice an integer.
The specification also names a skill most prep resources skip entirely: rearranging a jumbled sequence of statements into the correct order for a proof. It's directly testable, and it's exactly the skill the contrapositive fact from the last section unlocks.
Question. The four statements below, in some order, form a proof that “for all integers n, if n² is odd then n is odd”, by proving the contrapositive. Which order is correct?
Answer: A. A proof by contrapositive has to state the strategy before using it, so statement II comes first. III then sets up the hypothesis the contrapositive needs (n even), I carries out the algebra, and IV draws the conclusion the strategy in II promised. Proving “n² odd ⇒ n odd” directly is awkward — there's no clean way to start from an odd square and extract a factor. Proving its contrapositive, “n even ⇒ n² even”, is a two-line direct proof. Recognising that swap is most of the difficulty in this question — the reordering itself is quick once you see it.
Disproof by counterexample deserves its own worked example, because it is the one method that doesn't build an argument at all — it just needs one value that breaks the claim.
Question. Which of the following is a counterexample to the statement “for all real numbers x, x² ≥ x”?
Answer: C. At x = 0.5, x² = 0.25, which is less than x — the statement fails. Checking the others: at x = 2, 4 ≥ 2 holds; at x = −1, 1 ≥ −1 holds; at x = 1, 1 ≥ 1 holds (equality, not a failure). The statement is only false on the open interval between 0 and 1 — exactly the region a candidate has to think to test. Reaching for an integer by instinct, which is the natural first move, would never find this counterexample.
The specification's own guidance names two example errors it considers candidates should watch for: claiming that “if ab = ac, then b = c”, and assuming that “if sin A = sin B, then A = B” — stating plainly that neither is a valid deduction. Both fail for the same underlying reason: each performs a step (dividing by a quantity, or “undoing” a function) that is only reversible under a condition the argument never checks.
Question. The argument below claims to prove that “for all real numbers x and y, if xy = 5x, then y = 5.”
At which step does the argument first go wrong?
Answer: B. Step 2 divides both sides by x without first checking that x ≠ 0 — this is exactly the “ab = ac therefore b = c” error the specification names directly. If x = 0, the equation xy = 5x becomes 0 = 0, which holds for every value of y. Take x = 0 and y = 17: xy = 0 and 5x = 0, so the hypothesis holds, but y ≠ 5. That single counterexample confirms Step 2 is where the argument breaks — not Step 1, which is just the given hypothesis, and not Step 3, which would follow validly if Step 2 were sound.
The specification's other named example runs the same way. Claiming “sin A = sin B implies A = B” fails because sin 30° = sin 150° = 0.5, yet 30° ≠ 150° — sine is not one-to-one across that range, so “undoing” it doesn't recover a unique angle. Once you've seen both named examples, the pattern to check for on any proof question becomes concrete: every time a step divides by something, takes a square root, or applies a function to both sides, ask whether that operation is actually reversible for every value the argument allows — not just the ones it happened to have in mind.
Want to find out where your own gaps sit? Our free TMUA readiness check is a short adaptive set of TMUA-style questions. It gives you a readiness band and a topic-by-topic map across both papers, so you know whether reasoning specifically — rather than Paper 2 in general — is where to spend the next few weeks.
UAT-UK's own content specification splits Paper 2 into three parts: the logic of arguments (true and false, and/or/not, if–then statements, the converse and the contrapositive, necessary and sufficient conditions, and the quantifiers “for all” and “for some”), mathematical proof (direct proof, proof by cases, proof by contradiction and disproof by counterexample), and identifying errors in proofs. None of it is examined in a standard A-Level Maths course.
No. The content specification states explicitly that candidates will not be expected to recognise or use symbolic notation for these terms, and will not be expected to complete formal truth tables. Everything is tested in plain English.
A sufficient condition guarantees the statement is true, though other routes to the statement might exist. A necessary condition is required for the statement to be true, though satisfying it alone doesn't guarantee the statement. A condition that is both makes the two statements equivalent.
Swap the quantifier to “there exists” and negate only the part that follows it — any domain restriction stays exactly as it was. “For all x, P(x)” negates to “there exists an x such that P(x) is false”, not to another “for all” statement.
A claim that follows in a short chain from definitions usually wants a direct proof. A claim that behaves differently across a small number of cases usually wants proof by cases. A claim that something is impossible usually wants proof by contradiction. A “for all” claim you suspect is false is worth testing for a counterexample before you try to prove it.
UAT-UK's free Notes on Logic and Proof and full content specification, plus every past paper from 2016 to 2023 with worked answers. Then use our topic breakdown to see how reasoning is weighted, and our free readiness check to find your own gap.
This page explains what Paper 2 is actually asking. The free check tells you where you personally stand against it — and a tutor turns that into a route with an order and a deadline.
Start your free TMUA check Book a free meeting