Exam Technique Guide

Provenance: compiled 2026-09-28, grounded in research/veridian/WMA13-verified-facts.md (13 real WMA13 series, Oct 2020–Jun 2026), the 11 audited lessons in content/veridian/curriculum/wma13/, content/veridian/question-bank/wma13.ts, and — for the M/A/B conventions and the October 2024 series specifically — independently re-checked this pass directly against the primary Pearson text in ~/Documents/SOL/Papers/IAL_Paper3_Complete/Master_Consolidated_Docs/Mathematics_Paper3_Mark_Schemes.md and Mathematics_Paper3_Examiner_Reports.md.

#WMA13 (Pure Mathematics 3) — Exam-Technique Guide

#How to use this guide

This is a standing reference, not a content lesson — it assumes you have already worked through (or are working through) the 11 spec-content lessons in content/veridian/curriculum/wma13/. Where a point below is already taught in full there, this guide names the lesson and its trap id rather than repeating the worked material; where a point is genuinely new — not yet built into any WMA13 lesson — it is marked NEW and sourced in full, so it can be built into a lesson later. wma13-paper-anatomy.ts already covers the paper's basic shape (90 minutes, 75 marks, M/A/B definitions) as a lean orientation node; this guide is its companion — the diagnostic and technique layer paper-anatomy deliberately leaves out (it carries no trap-taxonomy, self-mark checklist, or timing table of its own).

---

#1. Paper structure and a real timing budget

The fixed facts, verbatim from the spec (Pearson Edexcel IAL Mathematics specification, Issue 3, April 2019, p.21): 1 hour 30 minutes, 75 marks, every question compulsory. There is no section structure — WMA13 is one flat sequence of questions, each with its own printed mark tariff, marked in M/A/B lines rather than AO bands.

Question count is not fixed. Real papers checked directly: January 2023 had 10 questions; October 2020, October 2021, and October 2024 each had 9. Do not expect a fixed count on the day — budget against the 75-mark total and each question's own printed tariff, not a memorised number of questions.

The average is 72 seconds per mark (90 minutes ÷ 75 marks) — a planning tool, not a per-question rule, because marks are not spread evenly across questions. A real worked example, using January 2023's full, hand-checked breakdown (6+6+5+7+9+8+9+5+11+9 = 75):

That budget sums to exactly 90 minutes — the arithmetic check that shows the "72 seconds per mark" figure is a real, usable planning number, not a rough rule of thumb. On the day: look at each question's own printed mark allocation as you reach it, multiply by 1.2, and that is roughly how long it has earned.

Typical mark tariffs by topic, drawn from real, individually-verified questions (not every series breaks down identically — these are evidenced ranges, not fixed rules):

Calculator policy — the real nuance. WMA13 is calculator-allowed in general, but only an ordinary scientific calculator: Appendix 6 bans "symbolic algebra, differentiation and/or integration." Individual questions then carry their own, further, explicit instruction on top — verified verbatim, printed directly above January 2023 Q5: "Solutions relying entirely on calculator technology are not acceptable." Section 3 below builds full technique guidance around this specific trap, because it recurs across essentially every spec area, not just one topic.

---

#2. The M/A/B mark-coding system, as this paper's real mark schemes use it

#2.1 The three mark types, verbatim (general marking guidance, printed at the start of every WMA13 scheme; independently re-checked this pass against the October 2024 primary mark scheme)

• M marks: "a correct method or an attempt at a correct method."
• A marks: "can only be awarded if the previous M mark has been earned" — "M0 A1 is impossible."
• B marks: "independent accuracy marks where there is no method" — a comment or a graph, typically.

The practical consequence: a genuine attempt at the right method still earns the M mark even if the arithmetic afterwards goes wrong. But because every A mark is explicitly chained to its own M mark, skipping the method line does not just lose the method mark — it forfeits every accuracy mark built on it too. Trying and getting it wrong scores something. Not trying, however correct the final number, does not.

#2.2 Dependent method marks: dM1 and ddM1

A method mark can itself depend on an earlier method mark already being scored. Depth of dependency is set by how many independent method marks actually precede that step in that specific question — not fixed by topic. Two real worked structures, chosen because they match exactly the shapes the spec names as recurring for this paper (product-rule/chain-rule differentiation, and exponential-modelling turning points):

Product rule + chain rule, one dependency (Oct 2024 Q7(a)), independently re-verified against the primary mark scheme this pass: differentiate $f(x) = x^3\sqrt{4x+7}$ to show $f'(x) = \frac{kx^2(2x+3)}{\sqrt{4x+7}}$.

• M1: attempts the product rule with the chain rule on the root factor, reaching $Px^2(4x+7)^{1/2} + Qx^3(4x+7)^{-1/2}$.
• A1: the correct unsimplified two-term derivative.
• dM1: dependent on the M1 — "correctly producing a single fraction with a denominator of $\sqrt{4x+7}$."
• A1: the final simplified fraction, $\frac{7x^2(2x+3)}{\sqrt{4x+7}}$.

The examiner report is blunt about where marks actually go on this shape: most candidates "carried [the product rule] out to a high degree of accuracy" — the differentiation itself is the accessible half; "the last two marks... were more discriminating." dM1 sits exactly on the combining-into-one-fraction step, not the differentiation.

Exponential modelling, two-stage dependency (Oct 2024 Q8(c)), independently re-verified against the primary mark scheme this pass: a horse's heart rate is $H = 32 + 40e^{-0.2t} - 20e^{-0.9t}$; find $T$, the time of maximum heart rate.

• M1, A1: differentiate to $\frac{dH}{dt} = -8e^{-0.2t} + 18e^{-0.9t}$.
• dM1: sets the derivative to zero and rearranges to collapse the two different rate constants into one: $4e^{0.7t} = 9$.
• M1: solves the single-exponential equation with correct log work.
• A1: $T =$ awrt 1.158 minutes.

Here the dM1 belongs to the algebraic combining of two differently-rated exponential terms into one solvable equation — a technique with no equivalent in a single-exponential-term model, and, per the examiner report, the exact place where "many weaker candidates gave up... and others took logs at too early a stage." See section 3 for the calculator-only version of this same question, which scores 2 of these 5 marks regardless of how correct the final number is.

A third, independently confirmed case, ddM1 (doubly dependent): Oct 2023 Q9(c) (a modulus-of-a-logarithm question — solve $|2-4\ln(x+1)|=3$ then select the outside region) needs two separate method marks before the region-selection step (one for forming an equation, one for solving both cases), so that step is doubly dependent: "dependent on both previous method marks." Oct 2020 Q5(a) (the $\sin 3x$ triple-angle proof) independently confirms the same ddM1 convention in an unrelated topic. The lesson for building your own worked examples or self-marking a mock: count how many independent method marks genuinely precede a step before labelling it dM1 vs ddM1 — the label follows the question's own structure, not the topic.

#2.3 "Quote the formula first"

The single most repeated piece of technique advice in the whole WMA13 archive, confirmed independently across at least five series (Oct 2020 Q3, Oct 2021 Q9(a), Jan 2022 Q6(a), Jun 2023 Q10(a), Jan 2024 Q7(a)). The mechanism is structural, not stylistic — the general marking principles state that an unquoted method mark "may be lost if there is any mistake" in the working it's inferred from. Write the quotient rule, the product rule, the chain-rule substitution, the $\log_a$ change-of-base — whatever formula the method rests on — as its own line, before substituting values into it. A slip two lines later then costs only the accuracy mark it actually affects, not the method mark underneath it too. This is why the lesson worked chains in differentiation-rules.ts (trap rule-not-quoted-before-use) and reciprocal-and-inverse-trigonometric-functions.ts (trap identity-not-quoted-before-use) both teach this as a first habit, not an afterthought.

#2.4 ft is an A/B-only tag — never M

Verbatim: "A and B marks may be f.t. – follow through – marks." M marks are absent from that sentence. The reason is structural: an M mark rewards the method, independent of what value it is applied to — there is nothing for a follow-through flag to signal on a mark that doesn't test correctness of the underlying value in the first place. A method correctly re-applied to an earlier wrong answer earns a plain M1, with no tag needed.

This is worth stating explicitly because this course's own lesson-audit history got it wrong three separate times before catching it — differentiation-rules.ts, numerical-methods.ts, and reciprocal-and-inverse-trigonometric-functions.ts each originally mislabelled a follow-through method mark "M1 ft" before the pattern was found and corrected (see each lesson's own header audit log). If you are self-marking a mock against a scheme you've written yourself, or building your own, treat any "M ft" you write as a signal to re-check — it is very likely a plain M1 with the reasoning simply carried over from an earlier value.

#2.5 Misreads, multiple attempts, and the abbreviation list

Verified verbatim, general marking guidance (independently re-checked this pass against the October 2024 primary mark scheme): bod — benefit of doubt; ft — follow through; cao — correct answer only; cso — correct solution only, no errors permitted anywhere in that part; isw — ignore subsequent working; awrt — answers which round to; SC — special case; oe — or equivalent; dep/indep — dependent/independent; dp/sf — decimal places/significant figures; an underlined star for "the answer is printed on the paper"; an arrow for "the second mark is dependent on gaining the first mark."

Two further real rules worth knowing precisely: for a misread that doesn't change the character of the question, examiners "deduct two from any A or B marks gained, in that part" — so a misread is not automatically catastrophic. And on multiple attempts: if all but one is crossed out, only the surviving attempt is marked; if everything is crossed out, or nothing is, every attempt is marked and the highest-scoring one counts. Practical implication: crossing out a wrong attempt cleanly, leaving exactly one visible, is a real, costless way to make sure the right attempt is the one scored — don't leave two live attempts side by side hoping the better one is picked, and don't leave a stray, illegible smudge instead of a clean cross-out.

#2.6 Exact answers vs decimals

A general convention, corroborated by examiner-report commentary in nearly every series reviewed: where an exact answer is asked for, or surds are clearly in play, "marks will normally be lost if the candidate resorts to using rounded decimals." Two specific, well-evidenced instances worth internalising rather than re-deriving under pressure: harmonic-form construction marks (Oct 2020 Q7(a)) require the exact surd for $R$ and $\alpha$ in the unit the question names — the mark scheme explicitly refuses a numerically-correct value in the wrong form ("the degree equivalent α = awrt 75.96° is A0"); and a log-linear-graph answer such as $T = \frac{1}{36}$ (Jun 2023 Q2(a)(ii)) explicitly does not accept the decimal or the unexpanded power form as a substitute.

---

#3. The calculator-only-method penalty: a paper-wide discipline

This is not a topic-specific quirk — it is the single most consistently enforced, paper-wide requirement in the whole archive, and October 2024's own examiner report opens by naming it as the one thing worth knowing above everything else that series: "candidates should make their methods clear." The two bold instruction sentences that appear above specific questions — "show all stages of your working" and, separately, "solutions relying entirely on calculator technology are not acceptable" — are a warning, not decoration.

The exact cost, in marks, not just in general terms. The real Oct 2024 Q8(c) mark scheme states the rule outright: "answers without full working may not gain credit." A candidate who writes $-8e^{-0.2t}+18e^{-0.9t}=0 \Rightarrow t=1.158$ with no algebra shown "scores 1,1,0,0,0" — the M1 and A1 for the derivative itself are still earned (differentiating is visible, checkable work), but the dM1, M1 and A1 for actually solving the equation are all zero, regardless of how correct 1.158 is. That is 2 of 5 marks for a numerically perfect final answer. This exact number — not "loses some marks," but the specific 1,1,0,0,0 breakdown — is worth knowing precisely, because it shows the penalty is not a vague deduction; it is every mark attached to a step the working never demonstrated.

Why this recurs across nearly every spec area, not just exponential modelling — cross-referenced, not repeated, from six separately audited lessons: differentiation-rules.ts (calculator-relied-on-with-no-method-shown), exponential-growth-and-decay.ts (the Oct 2024 Q8(c) marked solution's own commonWrongPath), exponentials-and-logarithmic-graphs.ts (calculator-equation-solver-used-with-no-algebra-shown), modulus-and-graph-transformations.ts (calculator-used-with-no-algebraic-method-shown), numerical-methods.ts (calculator-solver-used-despite-the-no-calculator-instruction, and separately ans-key-substitution-not-shown — embedding an iterate via the calculator's ANS key with no substitution line visible), and rational-expressions-and-algebraic-division.ts (calculator-technology-cited-as-the-method). Six spec areas, six independent lessons, the same mechanism every time: a correct final value with no method line earns whatever B/independent marks exist and nothing chained to an M.

The technique, stated once, for reuse everywhere: before touching a calculator to solve anything (as opposed to evaluate a already-derived expression), write the equation you are about to solve as its own line. On an exponential-modelling maximum/minimum question specifically, that means: derivative found and quoted → set to zero and rearranged so one exponential term stands alone on each side → the two rate constants combined into one by dividing → then, and only then, logs taken. Per the Oct 2024 examiner report, "others took logs at too early a stage" is named as a real, separate failure mode from calculator-only working — taking logs before the two exponential terms are combined is mathematically invalid, not just under-shown. Both failures are avoided by the same discipline: write the rearrangement down, in order, before reaching for a solver of any kind.

---

#4. Self-mark checklist

Run this after finishing a question, before moving on — each line is tied to a real, repeated mark-scheme or examiner-report finding, not a generic reminder.

Method visibility

• Did I quote the rule/formula I'm using as its own line before substituting into it? (§2.3)
• If this question carries the "show all stages... calculator technology" rubric, is every step of my working visible, not just the final value? (§3)
• For an iteration, did I show the value being substituted into the formula each time, not just the calculator's running answer? (numerical-methods.ts, ans-key-substitution-not-shown)

Show-that and proof completeness

• If the answer is printed on the paper, does every intermediate line actually appear — not just a correct dx/dy or f'(x) jumping straight to the given result? (confirmed across 4+ series — Oct 2020, Jan 2022, Jun 2023, Jan 2024)
• Did I state the identity I substituted (e.g. which cos2A form) rather than letting it appear only implicitly?

Exact values and units

• Does the question ask for an exact value, and if so, have I left surds/exact fractions rather than converting to a decimal? (§2.6)
• If my answer is a rate, area, or modelled quantity, have I written the unit and, where relevant, a direction word ("increasing"/"decreasing"/"loss")? (confirmed across 4+ series: Oct 2021, Jan 2022, Oct 2022, Jan 2025)
• Have I stated an angle in the unit the question actually asked for (degrees vs radians), including on a construction mark like $R$/$\alpha$, not just the final solved angle?

Domain, range, and solution-set discipline

• If I found an inverse function, did I also state its domain — and is it written in $y$/$f(x)$ notation, not $x$-notation? (§5.2, NEW)
• Have I checked that I haven't attached the correct algebra to the wrong function's letter, on a question that defines more than one function?
• On a "solve in a given range" trig equation, have I found every solution in range, and rejected every solution outside it?
• If the question asks for a specific solution ("the smallest," "the first"), have I rejected every other genuine solution I found, not just reported all of them? (§5.2, NEW)
• On a modulus equation or inequality, have I solved both branches, and selected the combined region the question actually needs, not just one branch's critical value?

Numerical methods specifically

• Does my root-location justification name continuity explicitly, not just "sign change"? (the single most repeated specific error in the entire archive — Oct 2022, Jan 2022, Jan 2024, improving only by Jan 2025)
• Have I given each iterate to the number of decimal places asked, without rounding prematurely partway through the chain?

Mark-code discipline (when self-marking against a scheme you've written)

• Have I only tagged A or B marks "ft," never an M mark? (§2.4)
• Have I checked how many independent M marks genuinely precede a step before calling it dM1 vs ddM1, rather than copying a label from a different question? (§2.2)

---

#5. Common-error digest

#5.1 Already documented — cross-referenced by lesson, not reproduced

Organised by spec area. Full worked treatment, MCQ drilling, and verbatim mark-scheme sourcing live in the named lesson file; only the mechanism and a fresh citation are given here.

#5.2 Genuinely new patterns — not yet built into any WMA13 lesson

(a) Reject the extra solution when the question asks for a specific one, not the full set — NEW. Verified verbatim, Jan 2022 Q2(b) mark scheme: the accuracy mark for "the smallest positive solution" is "ONLY" — awarded solely for that one value. The examiner report confirms candidates who correctly found both solutions to $\sin 2\theta = \frac{3}{4}$ but failed to discard the larger one "lost the accuracy mark which was even more of a shame," despite doing strictly more correct mathematics than the question needed. This is the mirror image of the already-documented "missing a second solution" trap: here the error is not discarding a genuine extra solution, not failing to find one. Technique: when a question names a specific solution ("the smallest," "the first," "the only"), finding every solution is necessary but not sufficient — the final line must actively reject every one the question didn't ask for.

(b) Forgetting to halve when recovering θ from a solved 2θ — NEW. Verified verbatim, the same Jan 2022 Q2(b) examiner report: "candidates did not divide by 2, leaving their answer as 48.59…°" instead of the correct 24.3°. This is a distinct arithmetic-recovery slip from the θ−α recovery already taught in harmonic-form.ts (trap stops-after-the-shifted-angle-never-recovers-theta, which is about adding back a phase shift) — this one is about dividing back a compound-angle multiplier once the equation is solved for $2\theta$ rather than $\theta$. Technique: the last line of any "solve for $2\theta$/$\frac{1}{2}\theta$" equation is not the value of $\theta$ itself — write the intermediate compound-angle value, then explicitly convert it, as its own visible step.

(c) Range of a function must be stated in $y$/$f(x)$ notation, not $x$-notation — NEW, and explicitly flagged as an open gap in the underlying research file. Verified verbatim, Jan 2023 Q1(a) mark scheme: the range of $f$ is accepted as "$y \leqslant 9$, $f \leqslant 9$, $y \in (-\infty, 9]$" but "explicitly not 'x ⩽ 9'" — a numerically correct range loses the mark if written in the wrong variable's notation. This is a distinct failure mode from the already-documented "domain of inverse omitted" trap (functions-composition-and-inverses.ts): here the domain/range is stated, with the right number, and still scores zero because of which letter it's attached to. Technique: a range is a set of outputs — write it using $y$ or $f(x)$, never $x$, even when the number itself is correct; a domain, symmetrically, is a set of inputs and takes $x$.

(d) On $\frac{dy}{dx}=\frac{1}{dx/dy}$, a turning point comes from setting $\frac{dx}{dy}=0$ directly — setting the reciprocal "equal to zero" and solving via its denominator is invalid algebra, even when it lands on the right numbers — NEW, independently verified this pass against the primary October 2024 mark scheme and examiner report. Question: $x = 2y^2+5y-6$; find the turning point. Correct route: differentiate to $\frac{dx}{dy}=4y+5$, invert to $\frac{dy}{dx}=\frac{1}{4y+5}$ (M1, A1), then set $4y+5=0$ directly (M1, dM1, A1) — the real mark scheme's own working. The report names the real error precisely: many candidates instead set the reciprocated expression, $\frac{1}{4y+5}$, "equal to zero," which the mark scheme states plainly "scored 0 marks" — "an incorrect assumption." A fraction is never made equal to zero by zeroing its denominator — that operation finds where the fraction is undefined, not where it equals zero — so the reasoning is invalid even though, for this particular curve, it happens to land on the same $y$-value as the legitimate method (because the turning point of $x=f(y)$ is exactly where $\frac{dx}{dy}=0$, i.e. exactly where $\frac{dy}{dx}$ is undefined). Technique: on a $\frac{dy}{dx}=\frac{1}{dx/dy}$ turning-point question, work from $\frac{dx}{dy}$ (before inverting) to find where it is zero — do not invert first and then try to "solve" the reciprocal equalling zero.

(e) Differentiating an exponential model with the power rule instead of $a^x \to a^x\ln a$ — NEW, independently verified this pass against the primary October 2024 examiner report. Verified verbatim, Oct 2024 Q4(c): candidates modelling $N = 100 \times 2.24^t$ produced $\frac{dN}{dt} = 100t \times 2.24^{t-1}$ — treating the variable exponent as though it were a fixed power and $2.24$ as though it were the variable base, exactly backwards from the model's actual structure, and scoring nothing. This is a distinct failure mode from the already-documented "substituted instead of differentiating" trap (exponential-growth-and-decay.ts) and from the "negative-exponent sign dropped" trap — both of those assume the correct rule was at least attempted. This is the wrong rule entirely: applying "bring down the power, reduce the power by one" (valid for $x^n$) to a term where $x$ is the exponent, not the base. Technique: before differentiating any $a^{f(t)}$ term, name which variable is in the exponent and which is the fixed base — if the variable is in the exponent, the rule is $a^x \to a^x \ln a$ (or the memorised $e^{kx}\to ke^{kx}$ form after rewriting), never the power rule.

(f) Terminating algebraic division early, leaving a linear expression where a genuine constant remainder is required — NEW, independently verified this pass against the primary October 2024 examiner report. Verified verbatim, Oct 2024 Q9(c) (algebraic long division, spec 1.1) — the examiner report names candidates who "terminate the division early," left with $(x-2)$ standing in for the remainder. A polynomial division by a quadratic divisor is not finished while the remainder still contains an $x$ term — for a quadratic divisor, division stops only once the remainder is a genuine constant (degree strictly less than the divisor). This is distinct from the already-documented cancel-and-simplify-step-abandoned trap (which is about not finishing a cancellation after the division is otherwise complete) — this is stopping the division itself one step too soon. Technique: before writing a final remainder, check its degree is strictly less than the divisor's — a remaining $x$ term after dividing by a quadratic is a signal the division isn't finished, not a valid final answer.

---

#6. Exam-day protocol

Short, and only what is directly grounded in the material above:

1. Before starting each question, scan for the two-sentence rubric — "show all stages... calculator technology are not acceptable." If it's there, no step in that question can be calculator-only, however fast the calculator route is (§3).
2. Quote the formula/rule as its own line before substituting into it — every time, not just when it feels necessary (§2.3).
3. Circle or underline the word that fixes the answer's required form before finishing a question — "exact," "to 3 d.p.," "in the form $\ln p$," "the smallest," "not involving logs" — because the mark scheme enforces that word literally, not the nearest correct-looking number (§2.6, §5.2a).
4. On a numerical-methods root-location question, say the word "continuous" out loud as you write your reason — it is the single most repeated specific loss in the entire archive, and it is a free, mechanical habit to fix (§4).
5. On any exponential-modelling turning-point question, write the rearrangement before touching logs: isolate one exponential term per side, divide to collapse two different rate constants into one, then take logs (§3).
6. On a "show that"/proof question, write every intermediate line, even ones that feel obvious — the mark scheme is reading the path, not just checking the destination matches the printed answer (§4).
7. Finish every modelling answer with a unit and, if it's a rate, a direction word — "£ per year, increasing" is a habit that costs seconds and is one of the most reliably dropped marks on the paper (§4).

---

#7. Genuine gaps flagged

• Per-question topic-to-number mapping is only confirmed for a handful of individual questions (e.g. Jan 2023 Q1 = functions, Oct 2024 Q1–Q9 as listed in §1), not for a complete series end-to-end beyond January 2023's mark total breakdown. Do not assume a fixed question order (e.g. "numerical methods is always Q1") — it is a strong tendency, not a rule; treat the ordering claims above as "commonly," not "always."
• The formula booklet's exact printed line for $\int \csc x \, dx$ was not independently re-verified this pass (inherited from the underlying research file's own flagged gap) — its presence on the sheet is corroborated by an examiner-report reference (Jun 2023), but the precise printed form should be checked directly against the booklet PDF before being taught as exact wording.
• The general marking guidance's boilerplate wording (§2.1, §2.4, §2.5) was independently re-checked this pass against one further series (October 2024), on top of the three series the underlying research file already checked (January 2023, October 2023, June 2022) — consistent with the assumption that this generic Edexcel-wide boilerplate does not change within one spec version, but this remains a representative sample, not an exhaustive check of every series.
• Section 5.2's six new patterns are each sourced to one series only (five to October 2024, one to January 2022). None has yet been cross-checked against a second series the way the course's existing, multi-series-confirmed traps have been — they are real and verbatim-sourced, but should be treated as single-instance findings pending a second occurrence, not yet as confirmed recurring patterns on the level of, say, the continuity-statement omission (independently confirmed across 5 series).
• This guide's own timing table (§1) is a planning tool built from real but non-uniform data — mark tariffs for the same topic genuinely vary between series (e.g. numerical methods ranges from 5 to 8 marks across the series checked), so the figures given are evidenced typical ranges, not guarantees for any specific future paper.

---

#Sources

• research/veridian/WMA13-verified-facts.md — the primary compiled source for this guide; 8 series with full mark schemes and examiner reports read (Oct 2020, Oct 2021, Jan 2022, Jun 2022, Oct 2022, Jun 2023, Jan 2024, Jan 2025), 3 further series for mark-scheme-notation cross-checking (Jan 2023, Oct 2023, Jun 2024), plus Oct 2024 (question-bank gap-mining pass) and Summer/Jun 2026 (citation-currency audits) — 13 series total.
• content/veridian/question-bank/wma13.ts — the two Oct 2024 marked-solution items (wma13-oct24-q7a-product-chain-rule, wma13-oct24-q8c-two-exponential-maximum) that anchor §2.2's worked dM1 examples.
• content/veridian/curriculum/wma13/*.ts — all 11 audited lessons, cross-referenced by trap id in §5.1 and §3 (not reproduced): paper-anatomy.ts, differentiation-rules.ts, exponential-growth-and-decay.ts, exponentials-and-logarithmic-graphs.ts, functions-composition-and-inverses.ts, modulus-and-graph-transformations.ts, reciprocal-and-inverse-trigonometric-functions.ts, compound-and-double-angle-formulae.ts, harmonic-form.ts, integration-by-recognition.ts, numerical-methods.ts, rational-expressions-and-algebraic-division.ts.
• ~/Documents/SOL/Papers/IAL_Paper3_Complete/Master_Consolidated_Docs/Mathematics_Paper3_Mark_Schemes.md and Mathematics_Paper3_Examiner_Reports.md — independently opened and read this pass (not trusted from the research file's own summary) to re-verify the general marking guidance verbatim, and to source the full October 2024 series (Q1–Q9) directly, including three of §5.2's six new findings (d, e, f) that go beyond what the research file itself had mined from that series.