Pure Mathematics 2

Forty Eight Hour Checklist

Provenance: written 2026-09-29, against research/veridian/WMA12-verified-facts.md (read directly, §1–§9 — the lesson-audit-log entries after the ones quoted below were not re-read line-by-line for this pass, since forge-exam-technique-guide.md and codex-formula-and-method-bank.md, both read in full here, already state they were built by reading the facts bank in full and already surface its load-bearing findings), content/veridian/technique-hub/wma12/forge-exam-technique-guide.md (read in full), content/veridian/technique-hub/wma12/codex-formula-and-method-bank.md (read in full), content/veridian/technique-hub/wma12/signal-frequency-and-boundaries.md (read in full), and the 12 real lesson files' filenames and headers in content/veridian/curriculum/wma12/ (checked for trap-taxonomy coverage via FORGE's own locator table, §"Part 4", rather than re-opened file by file, since re-deriving that locator independently would duplicate work FORGE already did and verified). No new primary-source PDF was opened for this document — everything below is assembled and re-framed from the four sources above, none of it re-derived from a past-paper PDF directly. That reframing — turning existing reference material into a time-boxed countdown with explicit self-test items and an explicit "don't start this now" list — is what makes this document additive; where a fact already has a home in FORGE, CODEX or SIGNAL, this document points at it by name rather than re-stating it in full.

#WMA12 — 48-Hour Pre-Exam Checklist

Pearson Edexcel International A-Level, Unit P2: Pure Mathematics 2. 90 minutes, 75 marks, no choice of question, calculator permitted throughout with no non-calculator section or paper anywhere in this qualification (spec Appendix 6; confirmed on every cover page reviewed, June 2019 through January 2026 — see the facts bank §3 and FORGE Part 1).

#What this document is, and isn't

This is not a content lesson and not a fourth copy of the exam-technique guide. It assumes you have already worked through the relevant lessons in content/veridian/curriculum/wma12/ and read (or at least skimmed) FORGE's Part 2–4 and CODEX's Part 1–2 at some point before the last 48 hours. What this document adds is a time-boxed triage: what actually has a documented payoff in the last two days, in what order, self-tested against this paper's own confirmed weak points rather than a generic revision checklist — and, just as concretely, what this paper's own real record says is a poor use of your remaining hours. Every claim below is traceable to the facts bank, FORGE, CODEX or SIGNAL by name; nothing new is asserted here that those documents don't already establish.

#Two things to accept about this paper before the countdown starts

The paper's shape is not fixed. Confirmed question count ranges from 9 to 11 across 20 archived series; per-question marks have ranged 3–14 within a single series (facts bank §3; FORGE Part 1). Proof usually opens the paper but was Question 10 — the closer — in June 2025 (FORGE Part 1). Don't walk in with a memorised template of "Q1 is proof, Q9 is the circle." Budget from the numbers printed on the paper in front of you, not from a practice paper you sat weeks ago.
A correct number with no visible method is a real, repeated zero, not a technicality. Two independent series confirm this directly for a calculator-shortcut: January 2024 Q5(i) (sum to infinity via the calculator's summation button) and January 2025 Q5(a), whose own mark scheme states outright that a correct value with no working shown scores 0 (FORGE Part 2). This governs almost everything in the self-test bank below.

#The countdown

#T-48h to T-24h — the last full review session

Run one complete self-test against the mark-scheme mechanics, not just the maths. Pick one past paper you have already attempted (or attempt one now if you have not) and re-mark your own work using the real M/A/B rules in FORGE Part 2 — specifically the dM1/ddM1 dependency distinction and the "quote the formula first" rule — rather than just checking final answers against the scheme. If you have never done this, do it now: it is the single highest-value 90-minute block left, because it tests exam mechanics you cannot pick up from reading alone.
Work through the cross-topic self-test list below once, in full, marking honestly which items you can do from memory and which you can't.
If — and only if — you found a genuine, specific gap in the self-test below, spend the remaining T-48h to T-24h block closing that one gap, using the relevant curriculum lesson file or the matching CODEX Part 2 method skeleton. Do not open a new topic you weren't already shaky on going into this session — see "What not to start" below.
Do not attempt a second full past paper today if you already ran one. One honest, fully re-marked paper against the real mark-scheme mechanics is worth more in this window than two rushed ones marked only against final answers.

#T-24h to T-3h — the day before

Re-read CODEX Part 1's memorise-vs-given table once, out loud or by writing it from memory, then check yourself against the actual table. This is the fastest-decaying knowledge in the whole paper — a list, not a skill — and the eight areas with zero booklet support (proof, factor/remainder, circle geometry, differentiation/integration technique, plus the two trig identities and three of the five log laws) are exactly the ones a tired brain reaches for the booklet on and finds nothing.
Run the "confirmed repeat offenders" list below (traps documented on three or more independent series) as a dedicated block — these are the highest-confidence, best-attested patterns in the whole archive, not a guess at what's "likely."
Confirm your calculator is on the permitted list: no symbolic algebra, no differentiation/integration facility, no stored retrievable formulae (spec Appendix 6; verbatim on every cover page). If you're not sure, check it tonight, not tomorrow morning.
Do not start a brand-new past paper this late in the day. If you haven't already run one today, a partial, rushed attempt the evening before produces more anxiety than signal. Use the time on the self-test bank instead.

#T-3h to T-0 — exam morning

Ten-minute skim only, not new work: the "confirmed repeat offenders" list, plus your own T-48h self-test items you marked as shaky.
Physically check: Yellow booklet (Mathematical Formulae and Statistical Tables, Issue 2 or current), permitted calculator, nothing else that could look like a stored-formula device.
Remind yourself of the three cross-topic rules that cost real marks regardless of topic: quote the formula before you use it; keep exact forms until the question says otherwise; if a part says "using your answer to (a)" or "hence," use that answer, don't recompute it from scratch (FORGE Part 2–3; confirmed real cost, twice over, at June 2025 Q5(b): zero marks and lost time for re-doing work the question told you to reuse).
Do not use this window to learn anything new. Everything in "What not to start" below applies with extra force this close to the exam.

#The last few minutes before you're told to begin

Read every question's own instruction line before starting it — a specific minority of questions each series carry a bold "show all stages of your working" or "solutions relying entirely on calculator technology are not acceptable" instruction (confirmed October 2023 onward; FORGE Part 1 and Part 5). These are not boilerplate.
Write the mark allocation next to each question as you reach it and budget from that number, not from memory of a practice paper's shape (FORGE Part 1, Part 5).

#Cross-topic habits — self-test these first, they apply to every question

These are not topic knowledge; they are five real, mark-scheme-mechanical habits, each confirmed costing marks on multiple independent series regardless of which spec area the question is testing. Full sourcing for all five is FORGE Part 2–3; this list exists so you can drill them as a five-item routine rather than hunting for them scattered across a longer document.

Quote the formula before you substitute into it. Self-test: pick any formula-based question type and write the general formula line first, on its own, before any numbers. If your habit is to go straight to numbers, that's the gap — confirmed real zero-mark outcomes exist for a correct final value with an invisible method (Jan 2024 Q5(i); Jan 2025 Q5(a)).
Keep exact forms until told otherwise. Self-test: solve a fractional-root quadratic (e.g. from $3x^2-2x-8=0$ ) and check whether your final line is $x=-\tfrac43,2$ or a rounded decimal. Real cost of getting this wrong: a correctly-factorised stationary point loses its accuracy mark if the $x$ -coordinate is rounded (Jan 2021 Q2), and the identical rule costs marks on circle radii and combined-log final forms too.
If a part says "using your answer to (a)" or "hence," use it — don't recompute. Self-test: can you name, right now, the difference between a "hence" instruction and a "you may use any valid method" instruction? Real cost of missing this: a fully correct trapezium-rule re-calculation in part (b) scored zero because the question demanded the part (a) answer be reused (June 2025 Q5(b)), and the examiner report explicitly adds that this also cost time.
Isolate your final answer from any list, expansion, or multi-line working. Self-test: after any question producing more than one number on the page (a Pascal's-triangle row, a set of candidate roots, a list of case values), do you circle, box, or otherwise mark the one the question actually asked for? Real cost: the January 2024 Q2 binomial question's final accuracy mark was "commonly not awarded to candidates who failed to isolate the required term from a list" (CODEX Part 2 §5).
Check degrees vs. radians before you touch the calculator's mode. Self-test: does the interval in the question use $\pi$ or a degree symbol? Confirmed as a real, repeated error independently on two series (October 2022 Q5; June 2025 Q8(i)) — working an entire radian-interval question in degrees, not merely rounding a radian answer to a decimal (a related but distinct trig error, see the per-topic list below).

#Confirmed repeat offenders — traps documented on three or more independent series

Everything in the archive is a real, cited finding — but these five specific patterns are the best-attested in the whole record: each has been independently confirmed recurring on at least three separate series, not found once and generalised. If your remaining self-test time is genuinely short, these five are where the evidence is thickest.

Trapezium-rule strip-width and bracketing errors — confirmed across at least four series (October 2020, January 2023, June 2024, October 2023; facts bank §6.9, CODEX Part 2 §9). Self-test: given $n+1$ ordinates, can you state $h$ without hesitating? Do you close the summation bracket only after every term — first, last, and every doubled interior ordinate — is inside it?
The log-free equation as its own standalone accuracy mark — confirmed on four independent series (January 2021 Q3(ii), June 2023 Q5, June 2024 Q3(i), June 2025 Q6(i); CODEX Part 2 §6). Self-test: after combining log terms on both sides of an equation, do you write the resulting equation with no logs in it as its own explicit line before solving — and do you explicitly reject any root that would make an original log's argument $\leqslant 0$ , rather than silently dropping it?
"Quote the formula first" / no marks for a calculator-only value — confirmed directly on two series with an explicit mark-scheme statement to that effect, and stated as a general rule in every series' own marking guidance (January 2024 Q5(i); January 2025 Q5(a); FORGE Part 2). See cross-topic habit 1 above.
Premature rounding costing the final accuracy mark on a trig or geometry question — confirmed on at least three series across two different sub-topics (October 2022 Q5, June 2025 Q8(ii) for trig; January 2023 circle radius for geometry; facts bank §6.7, §6.3). Self-test: habit 2 above, applied specifically to any question where an intermediate value is a surd or an ugly decimal.
Arithmetic/geometric series formula confusion mid-question — confirmed across at least two series directly, and named as a recurring category rather than a one-off (June 2019, October 2020; facts bank §6.4). Self-test: given a series, can you state in one sentence which of the four given formulae (arithmetic $n$ th term, arithmetic sum, geometric $n$ th term, geometric sum) it needs, before you write any of them down?

#Per-topic self-test — one pass through every spec area

Short, specific self-checks per topic, each tied to a real citation already established in FORGE, CODEX or the facts bank. This is not a re-teaching of the topic — if any item below is a genuine "no," go to the named lesson file, not this document, to fix it.

Proof (proof.ts): Can you state, without hesitating, the difference between what "prove by exhaustion" and "disprove by counter-example" each require you to write? Confirmed real cost of confusing them: unrequested algebraic proofs attempted where exhaustion was asked for score worse and cost more time, confirmed on at least two series (October 2022, October 2023; facts bank §6.1). Do you know that a counter-example question needs exactly one case, with its reason stated, not several?
Factor & remainder theorems (factor-and-remainder-theorems.ts): Given $f(x)$ and a value of the divisor's root, can you set up the correct equation (the right sign, the right side) in one line? The single most common confirmed error here is a sign slip on which root pairs with which factor form (January 2024 Q1; October 2023 Q4). If a question says "hence factorise," do you know that reaching the right answer by a different route scores zero for that part (October 2023 Q4(c) (ii); CODEX Part 2 §2)?
The circle (the-circle.ts): Do you sketch before you calculate? Real, direct examiner advice: a sketch is "advisable" and candidates who skip one lose marks finding the centre even after succeeding elsewhere (January 2021 Q9). Can you state the radius-⊥-tangent gradient rule without taking the reciprocal a second time by mistake (a confirmed real slip, January 2023 Q6)? Do you know to distinguish a diameter's length from a radius before squaring it (June 2025 Q2)?
Sequences & arithmetic series (sequences-and-arithmetic-series.ts): Can you write the standard proof of $S_n=\tfrac12n\{2a+(n-1)d\}$ from memory, in the real three-line B1/M1/A1* structure (write $S_n$ with the last term shown, reverse and add, reach the $2S_n$ line and halve)? Confirmed real weakness: examiners call a fully correct version of exactly this proof "surprisingly unusual to see" (January 2020 Q8(i); facts bank §6.4).
Geometric series (geometric-series.ts): If an arithmetic series feeds into a geometric one starting from its final term, do you know not to restart the geometric series from the arithmetic series' own first term, and not to double-count the shared term (January 2022; facts bank §6.4)? Can you unpack a $\sum_{r=1}^{\infty}$ expression into explicit terms and check a second ratio before asserting $r$ (January 2024 Q5(i))?
Binomial expansion (binomial-expansion.ts): Do you default to the $(a+bx)^n$ combinatorial form rather than the general $(1+x)^n$ series — even though both are printed on the same booklet page? Confirmed real cost of the wrong choice: examiners directly name it as the harder, make-little-progress route on a two-unknown question (June 2025 01R, Q7). When you raise a bracketed term to a power, do you raise the whole thing, including any coefficient — not just the variable (a confirmed slip, January 2024 Q2)?
Logarithms & exponentials (logarithms-and-exponentials.ts): Do you know log subtraction is not division of the log expressions themselves, and that a log is never distributed across a sum inside its own argument (January 2023 Q4(i), two named, confirmed misconceptions)? If a question specifies a particular combined single-log final form, do you know that stopping at an un-combined exact expression only earns an earlier mark, not the final one (January 2021 Q3(i))?
Trigonometric equations (trigonometric-equations.ts): When an equation has a term you could factor out (a common $\sin\theta$ , $\cos\theta$ , or $\tan\theta$ ), do you factorise rather than divide it out — dividing silently discards a whole solution branch (October 2023 Q3)? If the equation's argument carries a coefficient or is a compound angle, do you search the scaled interval before converting back (June 2025 Q8(i))? For a squared compound-angle equation, do you take both square roots before running the range-doubling method on each sign separately (October 2021 Q10(i))?
Differentiation — stationary points (stationary-points-and-curve-sketching.ts): After finding the second derivative, do you actually substitute each $x$ -value into it? Confirmed real weakness: roughly half of candidates who correctly reached the second derivative never took this step (January 2021 Q2). Do you compute $y$ -coordinates only when the question actually asks for them?
Integration — definite integrals & areas (definite-integrals-and-areas.ts): When a region's boundary is an axis rather than a solved intersection, do you state that limit directly with no method attached, rather than confusing a function's value there with the limit itself (January 2020 Q4)? Do you check for a sign change across the interval that would force you to split the integral?
Trapezium rule (the-trapezium-rule.ts): Covered above under "confirmed repeat offenders" — if you can state $h$ and the bracket structure without hesitating, this topic is in good shape. One further, distinct check: can you state the over/underestimate reason in one sentence that names the mechanism (how the chords sit relative to the curve), not just the direction — a combined mark, not two separate ones (October 2023 Q6(c))?

#What NOT to start learning fresh in the last 48 hours

Each item below is something this course's own research has explicitly flagged as either unconfirmed, low-value, or actively counterproductive to invest revision time in this late — not a guess at what's unimportant, but a direct read of the gaps and low-value patterns FORGE and CODEX already named.

The general $(1+x)^n$ real-index binomial series. Never required at P2 — the spec restricts P2 to positive integer $n$ (spec 4.5), and it sits on the same booklet page as the form you actually need only because the booklet is organised by unit, not by difficulty. Examiners directly confirm it as the harder, unnecessary route on a real question (June 2025 01R Q7). If you don't already use the combinatorial $(a+bx)^n$ form by reflex, drill that, not this — do not spend any of the last 48 hours on the general form at all.
Proof by contradiction as a from-scratch technique, if it isn't already familiar. It's a real, credited alternative route for general algebraic proofs (CODEX Part 1 §1) — but it is not the primary method this course's own lesson content builds fluency in, and building a new proof technique from zero in the last two days is a poor trade against reinforcing the exhaustion and counter-example methods you've already practiced. If it comes up and you don't know it, fall back on direct algebraic proof instead.
The 3-D real-world optimisation question family (a stated volume or physical constraint feeding a surface-area or similar formula, then differentiation). This is confirmed on a single series so far (June 2025 Q9, 11 marks) and FORGE's own words for it are that it is "not yet built into any lesson's trap-taxonomy" and "a lead for a future content pass, not claimed as fixed" (FORGE Part 4 #5). Building brand-new familiarity with an unfamiliar full question format from zero is a weak use of 48 hours next to reinforcing the confirmed cross-series patterns above. If one appears on the day, fall back on the general method: use the constraint to eliminate a variable, differentiate, solve, and justify with the second derivative (CODEX Part 2 §8) — the same skeleton every other optimisation question in this paper follows, not a special case you need to have memorised in advance.
The circle's degenerate/non-real-locus condition ( $g^2+f^2-c>0$ failing). CODEX states plainly that "no WMA12 question in the material reviewed for this document or the facts bank was confirmed to test the boundary/degenerate case directly" (CODEX Part 1 §3, flagged gaps). Real time is better spent on the circle traps in the per-topic list above.
Hunting for a specific "trap" in sequences 4.1 (recurrence relations) or 4.3 (increasing/decreasing/periodic sequences). CODEX names this directly as a gap in the underlying research, not something to be found with more searching: "no confirmed, primary-source-verified real examiner-report error pattern anywhere in this course's research to date" (CODEX Part 1 §4, flagged gaps). A 30-second recap of the plain definitions — increasing if $u_{n+1}>u_n$ , decreasing if $u_{n+1}<u_n$ , periodic with period $k$ if $u_{n+k}=u_n$ — is enough; don't over-invest here at the expense of the confirmed-trap topics above.
Sketching or interpreting the graph of $y=a^x$ in isolation. CODEX flags this directly: no confirmed real WMA12 question tests the graph itself, separately from solving $a^x=b$ equations (CODEX Part 1 §5, flagged gaps; §5 flagged gaps). The tested skill is the equation-solving method in the per-topic list above — spend your time there instead.
A brand-new full past paper less than roughly a day out, if you have already run at least one this week. This isn't a generic wellness recommendation — it follows directly from this document's own T-24h guidance above: a rushed, partially-completed paper the evening before produces less signal than a focused pass through the self-test bank, because you no longer have time to properly re-mark it against the real M/A/B mechanics that matter more than the raw score.

#Flagged gaps in this checklist

No topic-by-topic time allocation is given (e.g. "spend 40 minutes on logs, 20 on proof") because no pass of this course, including FORGE and SIGNAL, has topic-tagged all 75 marks of enough series to support a real weighting — both documents flag this explicitly as future work, not a guess this document should fill in.
The "confirmed repeat offenders" list is bounded by what has been independently checked so far. A trap appearing on only one or two series in the current archive may well recur just as often — the per-topic self-test above still covers those — this section only names the five with the deepest current cross-series confirmation, not a claim that everything else is rare.
This document does not cover exam-day logistics beyond equipment and instruction-reading (venue, timing of arrival, general test anxiety) — FORGE's own Part 5 states directly that it found no WMA12-specific grounding for general exam-wellness advice, and this document inherits that same limit rather than inventing guidance beyond what the sources support.