Taylor and Maclaurin series sit inside Unit 10 of the AP Calculus BC course description and reappear on the exam in two places: a multiple-choice set that asks candidates to identify coefficients, intervals of convergence, or values of a function, and a free-response question that walks through a polynomial approximation in two or three connected parts. For most candidates the terminology is the first obstacle, because a Maclaurin series is simply a Taylor series built at the centre x = 0, and the AP rubric does not award separate credit for naming the centre; it awards credit for writing the correct coefficients, finding the correct interval, and using the polynomial for the purpose the prompt states. The skill the FRQ actually tests is mechanical: expand a known function, manipulate it, and answer a chained question about a derivative, integral, or approximation error. The article below walks through the four shapes the rubric scores, the substitution row that catches most candidates, and the one inequality the alternating-series-error bound sits behind in a Taylor context.
What the AP Calculus BC syllabus expects from a Taylor or Maclaurin series
Unit 10 of the AP Calculus BC course description lists four knowledge components that the FRQ writer has to choose from. The first is the construction of a Taylor polynomial of degree n for a function centred at x = a, with the explicit formula P_n(x) = f(a) + f'(a)(x-a) + f''(a)(x-a)^2/2! + … + f^(n)(a)(x-a)^n/n!. The second is the construction of a Maclaurin series, which is the same formula evaluated at a = 0, producing only powers of x. The third is the interval of convergence, found by applying the ratio test to the general term and solving for the values of x where the limit L is less than 1, then checking the endpoints separately. The fourth is the manipulation of a known series — geometric, binomial, sin, cos, or e^x — into a new function through substitution, differentiation, or integration.
For most candidates reading this, the syllabus expectation boils down to a single sentence: write the series, find where it converges, and use it. The exam never asks for a hand-derived Taylor series of an unfamiliar transcendental function; the prompt always gives you a function whose derivatives at the centre follow a clean pattern, or it gives you a closed-form geometric or binomial series to manipulate. The FRQ, in particular, is built so that part (a) tests the coefficients, part (b) tests the interval or radius, and part (c) tests an applied use — approximation, integration, or evaluation of a limit. If you can read the prompt for those three steps, the question becomes mechanical.
The exam-format impact of this unit is also worth naming. AP Calculus BC contains Unit 10 as a BC-only domain, which means the topic can appear on the BC section of the exam in either MCQ or FRQ form. The MCQ tends to test a single step — recognise that the Maclaurin series of cos x has only even powers, or read off the radius of convergence from a substitution — while the FRQ tests a chain. A typical BC FRQ in this area runs 15 minutes and is worth 9 points across three parts. Knowing this scoring density helps with pacing: a 9-point Taylor FRQ should be budgeted at roughly 5 minutes per part, and part (a) should be wrapped before the 5-minute mark so that the harder interval-of-convergence and approximation rows have time to be written cleanly.
Building a Maclaurin polynomial: the coefficient row and the substitution row
Most Taylor FRQs on the AP exam open with a prompt of the form "the Maclaurin series for sin x is given by Σ (-1)^n x^(2n+1)/(2n+1)!." The first scored step is almost always a substitution: write the series for sin(x^2), or for sin(2x), or for x·cos x, by replacing every x in the given series with the new expression. Candidates who treat this as a typing exercise rather than a substitution exercise lose the row. The substitution is not just "put the new expression where x is"; it changes the power of every term and, when the new expression is a constant multiple of x, multiplies every coefficient by that constant raised to the same power.
For sin(x^2), for example, the general term becomes (-1)^n (x^2)^(2n+1)/(2n+1)! = (-1)^n x^(4n+2)/(2n+1)!. The 4n + 2 in the exponent is the row the rubric reads, because it determines whether the function is even or odd and therefore which coefficients are zero. Candidates who write x^(2n+1) instead of x^(4n+2) lose the coefficient row entirely. For sin(2x), the general term becomes (-1)^n (2x)^(2n+1)/(2n+1)! = (-1)^n 2^(2n+1) x^(2n+1)/(2n+1)!, and the 2^(2n+1) is the row that gets scored. The 2 does not become 2^n; the exponent n travels with it. This is the single most common error I see in marking practice, and it is worth drilling before the exam.
The Maclaurin polynomial of degree n is the same series truncated after the nth nonzero term, and the rubric awards one point for writing the truncated expression and one point for the interval of convergence or radius. The interval for a sin, cos, or e^x series, and for any polynomial in x obtained from them, is always the entire real line, because the ratio test gives L = 0 for every substitution of the form (kx)^m. The interval row is therefore free credit if the function is one of the four memorised series, but the prompt may disguise this by asking for the interval after a substitution. Always re-derive; never assume the interval survives the substitution unchanged.
Finding the interval of convergence after a substitution or a manipulation
The second scored row in nearly every Taylor FRQ is the interval of convergence. The standard method is the ratio test: write the general term a_n(x), compute |a_(n+1)(x) / a_n(x)|, take the limit as n goes to infinity, set the limit less than 1, and solve for x. The ratio test gives a radius R, and the interval is one of four forms: (-R, R), [-R, R), (-R, R], or [-R, R], depending on what happens at the endpoints. The AP rubric scores the radius as one point and the endpoint behaviour as one point, and it awards no partial credit inside the ratio computation. The row the reader checks is the solved inequality, not the limit expression.
For a geometric series starting at n = 0 with general term x^n, the ratio is |x|, so R = 1 and the interval is (-1, 1). For Σ x^n/n, the ratio is |x|·(n/(n+1)), which tends to |x|, giving R = 1 again, but at x = 1 the series is the harmonic series (divergent) and at x = -1 it is the alternating harmonic series (convergent), so the interval is [-1, 1). For Σ x^n/n^2, the interval is [-1, 1] because the p-series with p = 2 converges at both endpoints. For Σ (x-3)^n, the centre has shifted to 3, the radius is still 1, and the interval is (2, 4) or a half-closed version depending on the endpoints. The general principle is: shift the centre, keep the radius, and check the endpoints with the series that remains after the substitution is undone.
Where candidates lose this row is in the algebra between the ratio and the inequality. A common error is to set the limit less than 1, divide by |x|, and forget the case |x| = 0, which is fine because x = 0 always makes the series converge. A second common error is to solve |x| < 1 and write the answer as -1 < x < 1, then forget to test the endpoints. The endpoint row is one of the rubric's three scored rows in a typical interval question, and it cannot be skipped. If a candidate finds the radius correctly and then leaves the interval blank, only one of the two points is awarded. If a candidate writes the radius as 1 and the interval as -1 < x < 1 with no endpoint check, the rubric typically awards the radius point and no interval point, because the closed endpoints are part of the interval, not separate.
Manipulating a known series: the substitution, the derivative, the integral
The third scored step in a Taylor FRQ is the manipulation of a memorised series into a new function. There are three manipulation techniques, and the AP exam rotates through them across years. The first is substitution, covered in the section above. The second is term-by-term differentiation, which is allowed inside the interval of convergence and produces a new series whose general term is the derivative of the original general term. The third is term-by-term integration, also allowed inside the interval, which produces a series whose general term is the integral of the original general term, and which introduces a constant of integration that the rubric scores as the +C row.
For example, given the geometric series Σ x^n = 1/(1-x) on (-1, 1), term-by-term differentiation gives Σ n·x^(n-1) = 1/(1-x)^2, valid on the same interval. Term-by-term integration gives Σ x^(n+1)/(n+1) = -ln(1-x), again on (-1, 1), and the rubric awards credit for the +C row when the candidate writes -ln(1-x) + C. The +C is the row that candidates most often forget, and the rubric is unforgiving: an antiderivative of a series on the FRQ is graded as a definite antiderivative would be in Unit 8, with the constant present.
A second manipulation the exam tests is the binomial series for (1 + x)^k, where k is a negative integer or a fraction, written as 1 + kx + k(k-1)x^2/2! + k(k-1)(k-2)x^3/3! + … with radius 1. The rubric scores the coefficient of the x^n row, not the general formula; candidates who write the binomial coefficient form C(k, n) without the alternating sign or without the factorial in the denominator lose the row. A typical AP prompt asks for the first four nonzero terms of (1 + x)^(-1/2) and then for the interval of convergence, which is -1 < x ≤ 1 for this function because the series converges at x = 1 by the alternating series test. The endpoint row is the discriminator between a 2-point answer and a 3-point answer, and it is the place where the binomial series differs from the geometric series even though both have radius 1.
Common pitfalls and how to avoid them in Taylor series FRQ rows
The first pitfall is treating the centre as x = 0 by default. A Maclaurin series is centred at 0, but a Taylor series can be centred at any a, and the prompt will say so. If the prompt asks for the Taylor series of f at x = 2, the general term is f^(n)(2)(x-2)^n/n!, not f^(n)(0)x^n/n!. Candidates who default to Maclaurin form lose the centre row and every subsequent row. The rubric reads the centre first and the coefficients second, so this is a high-cost error.
The second pitfall is the alternating sign in a Maclaurin series for sin, cos, or the alternating p-series. The sign pattern in sin x is (-1)^n, in cos x it is (-1)^n, in the alternating harmonic series it is (-1)^n, and in 1/(1+x) it is (-1)^n. The sign is part of the general term and the rubric scores it as a separate row from the coefficient. Candidates who write a positive sign in front of an alternating term lose the sign row even if the coefficient is correct. If you are unsure, write the first four terms explicitly and check that they match the function for small x; this takes 30 seconds and catches the sign error before it is committed.
The third pitfall is the radius-of-convergence shift after a substitution. If the original series has radius 1 and you substitute x for 2x, the new series has radius 1/2, because the ratio test now gives |2x| < 1, or |x| < 1/2. The radius does not survive a substitution unchanged, and the rubric scores the new radius as a separate row. Candidates who write the interval as -1 < x < 1 after a 2x substitution lose the interval row, even if every other row is correct.
The fourth pitfall is the +C row in a term-by-term integration. As noted above, integrating a series produces an antiderivative that the rubric grades with the constant present. A candidate who writes -ln(1-x) without the +C loses one point, and on a 3-part FRQ that is the difference between a 5 and a 6. The constant is free; the row is graded separately from the integral; write it.
How the rubric scores a polynomial approximation: the error bound row
The fifth scored step in a Taylor FRQ is the approximation error, which on the AP exam is usually framed through the Alternating Series Estimation Theorem or the Lagrange error bound formula. The ASET is the more common test on the BC exam, because it requires no derivatives beyond what is already in the series. The theorem states that if a series is alternating, its terms decrease in absolute value monotonically to zero, and S is its sum, then the error of truncating after the nth term is bounded by the absolute value of the (n+1)th term. The rubric scores the error bound as a single row, and the bound itself is the row the reader checks.
For example, given the Maclaurin series for cos x truncated after the x^4 term, the next term is -x^6/6!, and the error of approximating cos x by 1 - x^2/2 + x^4/24 is at most |x|^6/720. The rubric awards one point for identifying the next nonzero term, one point for writing the absolute value, and one point for evaluating or simplifying the bound. Candidates who write the bound without the absolute value lose the second point, and candidates who pick the wrong term — say, the x^5 term from sin x by mistake — lose all three.
The Lagrange error bound, by contrast, is R_n(x) ≤ M|x-a|^(n+1)/(n+1)!, where M is an upper bound on the (n+1)th derivative of f on an interval containing a and x. The AP exam tests this bound in a question that gives the bound in the prompt or asks the candidate to find a value of n that makes the bound less than a stated tolerance. The row the rubric scores is the solved inequality, not the formula. A candidate who writes the formula correctly and then stops loses the row; the row is the answer to "how many terms do you need?" in terms of n. This is the single longest step in a Taylor approximation problem, and it should be budgeted 4–5 minutes in a 15-minute FRQ.
Worked example: a three-part Taylor FRQ with the rubric rows marked
Consider a prompt of the form: "The Maclaurin series for ln(1 + x) is Σ (-1)^(n+1) x^n/n for -1 < x ≤ 1. (a) Write the first four nonzero terms of the Maclaurin series for ln(1 + x^2). (b) Find the interval of convergence of the series in (a). (c) Use the first two nonzero terms to approximate ln(1.04), and bound the error of your approximation." This question is the canonical BC Taylor FRQ, and each part has a clean row to score.
Part (a) is a substitution. Replacing x by x^2 gives general term (-1)^(n+1) (x^2)^n/n = (-1)^(n+1) x^(2n)/n, so the first four nonzero terms are x^2 - x^4/2 + x^6/3 - x^8/4. The rubric scores the substitution (one row), the powers of x (one row), the sign pattern (one row), and the denominators (one row). A 4-point question, one point per term. The error to watch for is the power: a candidate who writes x^4 instead of x^8 for the fourth term loses the row, because the general formula puts 2n in the exponent.
Part (b) is the interval. The ratio test gives |x^2| < 1, or |x| < 1, so the radius is 1. At x = 1 the series is Σ (-1)^(n+1)/n, which converges by the alternating series test. At x = -1 the series is Σ (-1)^(n+1)/n, which is the same series, so it also converges. The interval is [-1, 1]. The rubric scores the radius (one point) and the endpoint behaviour (one point). Candidates who write -1 < x ≤ 1 lose the endpoint point at x = -1.
Part (c) is the approximation. Setting x^2 = 0.04 gives x = 0.2. The first two nonzero terms are 0.2^2 - 0.2^4/2 = 0.04 - 0.0008 = 0.0392. The next term is 0.2^6/3 = 0.000064/3 ≈ 0.0000213, so the error is at most 0.0000213 by ASET. The rubric scores the substitution (one point), the evaluation of the polynomial (one point), the identification of the next nonzero term (one point), and the bound (one point). Four points, four rows, and a clean 12-point question across the three parts.
Pacing and preparation strategy for Taylor series on the exam
The BC exam contains a Taylor or Maclaurin FRQ roughly every other year, and the rest of the time the topic is tested in a 2–3 question MCQ block. Preparation should reflect that distribution. Spend roughly 60 percent of your study time on the FRQ shape: writing the first four nonzero terms, finding the interval, and using the polynomial for an approximation. Spend the remaining 40 percent on the MCQ-style recognition: knowing the sin, cos, e^x, geometric, and binomial series from memory, and being able to read the radius of convergence from a given series without re-deriving it from scratch.
For most candidates reading this, the highest-leverage move is to memorise the five series above and the radius of convergence for each. sin x and cos x have radius infinity, e^x has radius infinity, the geometric series has radius 1, and the binomial series has radius 1. From these five facts, you can derive the radius of any substituted or manipulated series by chaining the ratio test on the general term. The exam will not test a series outside this set, because the rubric requires a closed-form answer and a clean interval. If a function's series does not reduce to one of these five, it is not on the exam.
The final preparation move is timed practice on a 9-point Taylor FRQ. A 15-minute budget, broken into 5 minutes per part, is the standard pacing. If part (a) takes longer than 5 minutes, the candidate is doing the question out of order; part (a) is always the coefficient row, and it should be the fastest part. If part (b) takes longer than 4 minutes, the candidate is over-checking endpoints; the endpoints are a 1-minute decision once the radius is found. If part (c) takes longer than 6 minutes, the candidate is computing the approximation by hand rather than by polynomial evaluation; the polynomial is the point of the question, not a side calculation. Practice the timing once, and the exam-day pacing will follow.
How Taylor series FRQs differ from the MCQ block on the same topic
The MCQ block tests recognition and the FRQ tests construction. This split is the reason BC students should treat the two formats as separate skills. The MCQ might ask which of four series is the Maclaurin series for f(x) = sin(x^2)/x, and the answer is reached by pattern-matching: the candidate knows sin x = Σ (-1)^n x^(2n+1)/(2n+1)!, divides by x, and reads off the new general term. No algebra, no interval, no approximation — just the recognition. The FRQ, by contrast, takes that same recognition and asks the candidate to write the series, find the interval, and use the polynomial. The MCQ is 90 seconds per question, and the FRQ is 5 minutes per part.
| Feature | MCQ Taylor question | FRQ Taylor question |
|---|---|---|
| Time budget | ~90 seconds per question | ~5 minutes per part, 15 minutes total |
| Scored rows | One row: the answer choice | Three to four rows: coefficients, interval, approximation, error |
| Typical prompt shape | "Which series represents f(x)?" | "Write the first four terms. Find the interval. Approximate the value." |
| Skill tested | Pattern recognition | Construction and chain reasoning |
| Common error | Misreading the power after a substitution | Forgetting the +C row in a term-by-term integration |
The table captures the operational difference. A candidate who prepares only the MCQ format will recognise the right series on the FRQ but lose the construction rows; a candidate who prepares only the FRQ format will be slow on the MCQ and lose recognition questions to time pressure. The preparation plan should be split: three to four FRQ practice runs under timed conditions, and one to two MCQ sets of 8–10 questions each, with the goal of recognising the five memorised series within 30 seconds of seeing the function. AP Courses' one-to-one AP Calculus BC programme runs both formats in parallel, with rubric-level feedback on the FRQ construction rows and pacing drills on the MCQ recognition rows, so the two skills do not interfere with each other in the final weeks of preparation.
Conclusion and next steps for Taylor series preparation
Taylor and Maclaurin series on the AP Calculus BC exam reduce to a small number of mechanical rows: write the coefficients, find the interval, use the polynomial. The five series to memorise are sin x, cos x, e^x, the geometric series, and the binomial series for (1+x)^k. The four manipulations to drill are substitution, term-by-term differentiation, term-by-term integration, and the alternating series error bound. The three pitfall rows to watch for are the centre of the Taylor polynomial, the alternating sign in the general term, and the +C in an integrated series. With these in hand, the BC Taylor FRQ becomes a 9-point question that scores predictably for any candidate who can read the prompt in under a minute and pace the three parts in 5 minutes each. AP Courses' AP Calculus BC Taylor-series module walks each student through a personalised error log of these three pitfall rows, then runs timed FRQ practice with rubric-level marking on the construction steps, so that the exam-day Taylor or Maclaurin question is scored for what the student actually knows rather than for what the rubric penalises by default.