On the AP Calculus AB and BC exams, differentiability and continuity are usually introduced in the same week of class, tested in the same unit, and graded with overlapping rubric language, yet they are not the same property. A function can be continuous at a point and still fail to be differentiable there; a differentiable function is always continuous, but the converse does not hold. That asymmetry is the single most reliable source of points lost on free-response questions, and it is also the source of several multiple-choice distractors that the College Board reuses year after year. This article walks through the precise relationship, the three failure modes that catch students out, the rubric language that scores differentiability versus continuity, and the preparation sequence that converts this topic from a weak spot into a routine 3-of-3 or 4-of-4 answer.
The one-way theorem: differentiability implies continuity, never the reverse
Every AB and BC exam assumes a working command of one foundational theorem: if a function is differentiable at a point, it must be continuous at that point. The proof is short and is, in my experience, worth memorising in three lines, because a free-response prompt will sometimes award a point for showing the link explicitly. You start with the limit definition of the derivative, factor the numerator as a difference of the function values plus the function value at the point, and then split the limit into two pieces. The first piece is the derivative, the second piece is the function value. As long as the derivative exists, the limit of the difference quotient is the function value at the point, which is exactly the epsilon-delta statement of continuity.
The reverse, however, is false. A function can be continuous at a point and still fail to have a derivative there, and the AP exam exploits this gap in roughly one of every three free-response questions on Unit 2 of the AB syllabus and Unit 1 of the BC syllabus. The classic example is the absolute value function at the origin. The function is continuous at zero because the limit from the left, the limit from the right, and the function value are all zero. But the derivative from the left equals negative one, the derivative from the right equals positive one, and the two one-sided limits disagree, so the derivative at zero does not exist. The exam expects you to say so, in writing, with the one-sided limits named. Saying only "the function is not differentiable because of a corner" is not enough for the rubric's justification point.
There is a second version of the same trap that appears more often in the multiple-choice section: a piecewise function that joins two smooth pieces at a vertical line. The continuity check asks for agreement of three things at the join, and the differentiability check asks for agreement of the two one-sided derivatives. Many candidates who breeze through the continuity check then skip the differentiability check entirely, or confuse the two by computing the same quantity twice. Get into the habit of treating continuity and differentiability as two separate audits at every piecewise join. The audit takes roughly ten seconds and saves a question.
A useful tactical note: the AP exam rarely asks the abstract question "is differentiability at a point equivalent to continuity?" It asks the operational question, which is "determine whether the function is differentiable at x = a, justify your answer, and find the value of the derivative where it exists." The three tasks are scored in that order, and a candidate who answers only the first and third parts without the middle justification gives up a point that is easy to keep. The next section expands on the three failure modes the rubric expects you to recognise before you can write the justification.
The three failure modes: corners, cusps, and vertical tangent lines
Every AP Calculus candidate is expected to recognise three geometric situations in which continuity holds but differentiability fails. These are corners, cusps, and vertical tangent lines. They are not interchangeable: the rubric distinguishes them because the algebraic test for each is different, and a free-response prompt will sometimes ask you to name which one is present. The corner is the absolute-value case, where the two one-sided derivatives are finite and unequal. The cusp is the cube-root case, where the two one-sided derivatives are both infinite, or both fail to exist, with the same sign. The vertical tangent line is a function like the cube root of x cubed, equal to x, which is differentiable, so this is a useful negative example: vertical tangent lines in the informal sense of "the graph shoots up" can still be differentiable if the function is smooth enough on both sides.
The algebraic test for a corner is to evaluate the two one-sided derivatives as finite numbers and show that they disagree. The algebraic test for a cusp is to evaluate the two one-sided derivatives and show that at least one of them is infinite, with the function approaching the point along both sides. The algebraic test for a vertical tangent line is to evaluate the limit of the difference quotient and show that it diverges to positive or negative infinity while the function remains continuous. A useful shorthand: at a corner the slopes are finite and different, at a cusp the slopes are infinite, at a vertical tangent the slope is infinite and the function is still continuous, but a cusp and a vertical tangent are not the same object on the rubric.
For a worked example, take the function f of x equals the absolute value of x minus two, raised to the one-third power, and ask whether it is differentiable at x = 2. The function is continuous at 2 because both one-sided limits equal zero. From the right, the inside of the absolute value is positive, so the expression simplifies to the cube root of x minus 2, and the derivative from the right is one over three times the cube root of x minus 2 squared, which tends to positive infinity as x approaches 2 from the right. From the left, the inside of the absolute value is negative, so the expression is the cube root of negative x minus 2, and the derivative from the left is negative one over three times the cube root of 2 minus x squared, which also tends to positive infinity because the numerator is negative and the denominator is positive. The two one-sided derivatives agree in the infinite sense, so the failure mode is a cusp, not a corner, and the correct justification is "the one-sided derivatives both diverge to positive infinity, so f is not differentiable at x = 2."
Many candidates at this point are tempted to write a single sentence and move on. The AP rubric, however, expects at least two lines: one naming the one-sided derivatives, and one stating the conclusion. Some prompts go further and ask for a sketch, in which case a small graph with the one-sided slopes labelled will pick up a justification point that the algebra alone may not. In a multiple-choice setting, the same setup gives four standard distractors: differentiable everywhere, continuous but not differentiable, neither continuous nor differentiable, and differentiable except at the join. The right answer is almost always the middle one, and the wrong answers correspond to the three most common algebraic errors: forgetting the absolute value when computing the one-sided derivatives, dropping a negative sign, and confusing a vertical tangent with a cusp.
The deeper lesson is that the three failure modes are not just vocabulary. They are decision trees, and the decision the rubric is grading is whether you chose the correct one. A candidate who says "vertical tangent" when the function actually has a corner loses the same point as a candidate who does not name the failure mode at all. With that in mind, the next section moves from the geometry to the algebra: how to compute the one-sided derivatives cleanly, and how to present the computation in a way that the rubric accepts.
Computing one-sided derivatives the way the rubric accepts
The AP Calculus free-response rubric awards a justification point whenever a candidate explicitly evaluates the two one-sided limits of the difference quotient at a point of suspected non-differentiability. The format is not arbitrary: the rubric expects the limit, the algebraic work, and the conclusion, in that order, and it expects the work to be shown, not implied. Writing "the one-sided derivatives disagree, so the function is not differentiable" is not enough. Writing the limit symbol, the algebraic simplification, and the resulting finite or infinite value, is. A useful target is three to four lines of work per one-sided derivative, with the conclusion sentence sitting on its own line at the bottom.
Consider the piecewise function f of x equals x squared when x is less than 1, and 3x minus 2 when x is greater than or equal to 1. The continuity audit at x = 1 is the easier part: the left-hand limit is 1, the right-hand limit is 1, and the function value is 1, so f is continuous. The differentiability audit asks for the two one-sided derivatives. From the left, the derivative is 2x, which at x = 1 equals 2. From the right, the derivative is 3. The two values disagree, so f is not differentiable at x = 1. A complete free-response answer would show the left-hand derivative computation as a limit, the right-hand derivative computation as a limit, and the sentence "because 2 does not equal 3, f is not differentiable at x = 1." Three lines of algebra, one sentence of conclusion, one point for the audit, and one point for the conclusion. Total work: under two minutes.
For BC candidates, the same audit shows up in disguised form on parametric and vector-valued questions. A common prompt defines a particle's position parametrically and asks for the velocity vector at a specific time. The velocity is the derivative of position with respect to time, and the differentiability audit is the same one-sided check, except that the components of the position vector are the functions being audited. The continuity check asks whether the position vector is continuous; the differentiability check asks whether the velocity vector exists. A particle whose x-component is continuous but whose y-component has a corner at t = 1 has a continuous position and a discontinuous velocity, so the speed is undefined at t = 1 even though the particle is somewhere well-defined. This is the same one-way theorem, applied in a different coordinate system.
The cleanest preparation tactic is to drill a small set of piecewise functions until the audit becomes a reflex. For most candidates, ten piecewise examples covering a corner, a cusp, a join of polynomials, a join of a polynomial and a trigonometric function, and a join of a polynomial and a square root will exhaust the question bank. The next section extends the audit from a single point to an interval, which is where the AP exam's multiple-choice section does most of its damage.
Open intervals, closed intervals, and the interval audit
The AP exam asks two different questions about intervals, and conflating them is one of the most common reasons candidates lose points on Unit 2 of AB and Unit 1 of BC. The first question is whether a function is differentiable on an open interval, which means differentiable at every point inside the interval. The second question is whether a function is differentiable on a closed interval, which means differentiable on the open interval and also having matching one-sided derivatives at the endpoints. The first is a global audit, the second is a global audit plus a one-sided audit at the boundary. The two are not equivalent, and the rubric treats them differently.
For a worked example, consider f of x equals the absolute value of x on the closed interval from negative 2 to 2. The function is differentiable on the open interval from negative 2 to 0 and on the open interval from 0 to 2. It is not differentiable at 0. So the function is differentiable on the open interval from negative 2 to 2 minus the point 0, but it is not differentiable on the closed interval from negative 2 to 2. A candidate who writes "f is differentiable on the interval from negative 2 to 2" loses the point, and a candidate who writes "f is differentiable on the open interval from negative 2 to 2 except at x = 0" keeps the point. The wording is the difference between a 3 and a 4 on the rubric.
For BC candidates, the closed-interval audit is the setup for the Mean Value Theorem and the Extreme Value Theorem, both of which appear in the second half of the syllabus. The Extreme Value Theorem requires continuity on a closed interval, and the Mean Value Theorem requires continuity on a closed interval and differentiability on the corresponding open interval. A function that is continuous on the closed interval but fails to be differentiable at an interior point still satisfies the hypothesis of the Extreme Value Theorem but not of the Mean Value Theorem. The AP exam has been known to test this distinction by giving a piecewise function, asking for the absolute maximum, and then asking whether the Mean Value Theorem applies. The answer to the first is a routine audit, and the answer to the second is no, because the function has a corner.
A common tactical mistake is to confuse differentiability on an open interval with continuity on the corresponding closed interval. The two are not the same condition. A function can be differentiable on an open interval and discontinuous at the endpoints, and a function can be continuous on a closed interval and fail to be differentiable at an interior point. Reading the interval carefully is the only safeguard, and the next section catalogues the rubric language that the AP graders actually look for.
Rubric language: the words that earn points and the words that cost them
AP Calculus free-response rubrics are unusually literal. The graders are trained to award a point for a specific sentence structure, and a candidate who uses a synonym or paraphrases the conclusion sometimes loses the point even when the algebra is correct. The most important phrases to internalise are "f is continuous at x = a because the limit of f of x as x approaches a exists and equals f of a," "f is differentiable at x = a because the limit of the difference quotient exists," and "f is not differentiable at x = a because the one-sided limits of the difference quotient disagree." Each phrase is worth one justification point on a typical prompt, and the wording is not interchangeable.
A second cluster of phrases governs the interval statements. "f is continuous on the open interval from a to b," "f is continuous on the closed interval from a to b," "f is differentiable on the open interval from a to b," and "f is differentiable on the closed interval from a to b" are four distinct claims, and the AP exam has been known to award or withhold a point based on the choice of the word "open" or "closed." The safest habit is to name the interval explicitly every time. A candidate who writes "f is continuous on the interval from a to b" is ambiguous, and ambiguity is the friend of no one on a free-response prompt.
A third cluster governs the geometric vocabulary. The phrases "corner," "cusp," and "vertical tangent" are graded for precision. A corner is a point at which the one-sided derivatives are finite and unequal. A cusp is a point at which at least one one-sided derivative is infinite and the function is continuous. A vertical tangent is a tangent line that is itself vertical, which is the case when the derivative is zero and the function behaves like the cube root of x at the origin, no, that is differentiable, so the cleaner example is a function like x to the one-third at the origin, which is differentiable with a vertical tangent. The vocabulary is precise, and the rubric expects precision. With that in mind, the next section turns to the preparation sequence that converts this material into a 5.
| Property | Definition in AP Calculus rubric terms | Algebraic test | Geometric signature |
|---|---|---|---|
| Continuous at a point | Limit equals function value | Three-way agreement of left, right, and value | No break in the graph |
| Differentiable at a point | Limit of difference quotient exists | Two one-sided derivatives agree as finite numbers | Single well-defined tangent line |
| Continuous but not differentiable | Continuous, one-sided derivatives disagree | One-sided derivatives finite and unequal, or at least one infinite | Corner or cusp |
| Differentiable implies continuous | Theorem, proof by limit definition | Factor the difference quotient and apply the derivative | One-way implication |
Common pitfalls and how to avoid them
The first pitfall is treating "continuous" and "differentiable" as synonyms. Roughly one in five candidates in any given cohort does this, and the cost is a free-response point that the rubric does not give back. The fix is mechanical: at every piecewise join, run two audits, one for continuity and one for differentiability, and write the conclusion of each on its own line. The audits take ten seconds together and they remove a class of errors that no amount of algebraic skill will catch.
The second pitfall is skipping the one-sided derivative computation because the function is continuous. Continuity is not a substitute for differentiability. The two are related by a one-way theorem, and the rubric awards the justification point only when the one-sided derivatives are computed explicitly. A candidate who writes "the function is continuous at x = 1, so it is differentiable there" is making exactly the error the rubric is designed to catch.
The third pitfall is misidentifying a cusp as a corner. The two are distinguished by whether the one-sided derivatives are finite or infinite. A function like the absolute value of x to the one-third has a cusp at the origin, not a corner, because the one-sided derivatives diverge. Naming the wrong failure mode costs the same as naming no failure mode at all, and the fix is to evaluate the one-sided derivatives and check whether they are finite or infinite before assigning a label.
The fourth pitfall is using the word "open" when the rubric expects "closed," or vice versa. The Mean Value Theorem and the Extreme Value Theorem have different hypotheses, and a wrong interval costs a point. The fix is to underline the interval in the prompt and to copy the wording, including "open" or "closed," into the answer. The habit takes a few seconds and it pays off across multiple units of the syllabus.
The fifth pitfall is leaving the conclusion sentence implicit. A free-response answer that ends at the algebra and does not state "f is not differentiable at x = 1" forfeits a point that the rubric awards for the conclusion, not the work. The fix is to end every audit with a one-sentence statement of the result, written in plain language, with the point named. The graders read the last line of the audit, and a clear last line is the difference between a 3 and a 4 on a typical prompt.
Preparation sequence: from definition to 5 in four weeks
Most candidates I work with need roughly four weeks of focused preparation to convert differentiability and continuity from a weak spot into a routine 3-of-3 answer. The first week is definition and theorem: write the limit definition of the derivative from memory, prove that differentiability implies continuity, and write the three failure modes with one example each. The second week is the interval audit: drill ten piecewise functions, paying attention to the open-versus-closed distinction, and write the rubric sentences for each. The third week is the free-response format: time yourself on past prompts, practise the two-audit habit at every piecewise join, and self-grade using the published rubrics. The fourth week is the multiple-choice format: drill College Board released items until the corner-versus-cusp distinction becomes reflexive.
The single most efficient exercise is what I call the two-audit drill. Take a piecewise function, run the continuity audit in one column and the differentiability audit in the other, and write the rubric sentence at the bottom of each column. Repeat for ten functions. The exercise takes about forty minutes and it covers roughly eighty percent of the free-response points available on this topic. For BC candidates, the same drill works on parametric and vector-valued functions, with the position vector playing the role of the function and the velocity vector playing the role of the derivative. The audit structure is identical, and the four-week sequence transfers cleanly.
Conclusion and next steps
Differentiability and continuity are not a single topic but a small cluster of audits: a point audit, an interval audit, a one-sided audit at the boundary, and a justification audit for each. The AP Calculus AB and BC exams reward candidates who run all of them explicitly, in the rubric's preferred order, and who name the conclusion in plain language. For candidates targeting a 5, the next concrete step is to drill the two-audit exercise on ten piecewise functions and to self-grade the conclusion sentences against the published rubrics. AP Courses' AP Calculus AB and BC programmes run that drill against each student's free-response work, isolate the specific audit that is costing points, and turn the corner-versus-cusp distinction into a ten-second reflex on exam day.