Monday, June 30, 2014

Beyond the Limit, I

In my May column, FDWK+B, I said that I would love to ignore limits until we get to infinite series. One of my readers called me out on this, asking how I would motivate the definition of the derivative. Beginning this month and continuing through September, I would like to use my postings to give a brief overview of some of the problems with limit as an organizing principle for first-year calculus and to describe research that supports a better approach.

To a mathematician, the limit of f(x) as x approaches c is informally defined as that value L to which the function is forced to be arbitrarily close by taking x sufficiently close (but not equal) to c. In most calculus texts, this provides the foundation for the definition of the derivative: The derivative of f at c is the limit as x approaches c of the average rate of change of f over the interval from x to c. Most calculus texts also invoke the concept of limit in defining the definite integral, though here its application is much more sophisticated.

There are many pedagogical problems with this approach. The very first is that any definition of limit that is mathematically correct makes little sense to most students. Starting with a highly abstract definition and then moving toward instances of its application is exactly the opposite of how we know people learn. This problem is compounded by the fact that first-year calculus does not really use the limit definitions of derivative or integral. Students develop many ways of understanding derivatives and integrals, but limits, especially as correctly defined, are almost never employed as a tool with which first-year calculus students tackle the problems they need to solve in either differential or integral calculus. The chapter on limits, with its attendant and rather idiosyncratic problems, is viewed as an isolated set of procedures to be mastered.

This student perception of the material on limits as purely procedural was illustrated in a Canadian study (Hardy 2009) of students who had just been through a lesson in which they were shown how to find limits of rational functions at a value of x at which both numerator and denominator were zero. Hardy ran individual observations of 28 students as they worked through a set of problems that were superficially similar to what they had seen in class, but in fact should have been simpler. Students were asked to find \(\lim_{x\to 2} (x+3)/(x^2-9)\). This was solved correctly by all but one of the students, although most them first performed the unnecessary step of factoring x+3 out of both numerator and denominator. When faced with \( \lim_{x\to 1} (x-1)/(x^2+x) \), the fraction of students who could solve this fell to 82%. Many were confused by the fact that x–1 is not a factor of the denominator. The problem \( \lim_{x \to 5} (x^2-4)/(x^2-25) \) evoked an even stronger expectation that x–5 must be a factor of both numerator and denominator. It was correctly solved by only 43% of the students.

The Canadian study hints at what forty years of investigations of student understandings and misunderstandings of limits have confirmed: Student understanding of limit is tied up with the process of finding limits. Even when students are able to transcend the mere mastery of a set of procedures, almost all get caught in the language of “approaching” a limit, what many researchers have referred to as a dynamic interpretation of limit, and are unable to get beyond the idea of a limit as something to which you simply come closer and closer.

Many studies have explored common misconceptions that arise from this dynamic interpretation. One is that each term of a convergent sequence must be closer to the limit than the previous term. Another is that no term of the convergent sequence can equal the limit. A third, and even more problematic interpretation, is to understand the word “limit” as a reference to the entire process of moving a point along the graph of a function or listing the terms of a sequence, a misconception that, unfortunately, may be reinforced by dynamic software. This plays out in one particularly interesting error that was observed by Tall and Vinner (1981): They encountered students who would agree that the sequence 0.9, 0.99, 0.999, … converges to \(0.\overline{9} \) and that this sequence also converges to 1, but they would still hold to the belief that these two limits are not equal. In drilling into student beliefs, it was discovered that \(0.\overline{9} \) is often understood not as a number, but as a process. As such it may be approaching 1, but it never equals 1. Tied up in this is student understanding of the term “converge” as describing some sort of equivalence.

Words that we assume have clear meanings are often interpreted in surprising ways by our students. As David Tall has repeatedly shown (for example, see Tall & Vinner, 1981), a student’s concept image or understanding of what a term means will always trump the concept definition, the actual definition of that term. Thus, Oehrtman (2009) has found that when faced with a mathematically correct definition of limit—that value L to which the function is forced to be arbitrarily close by taking x sufficiently close but not equal to c—most students read the definition through the lens of their understanding that limit means that as x gets closer to c, f(x) gets closer to L. “Sufficiently close” is understood to mean “very close” and “arbitrarily close” becomes “very, very close,” and the definition is transformed in the student’s mind to the statement that the function is very, very close to L when x is very close to c.

That raises an interesting and inadequately explored question: Is this so bad? When we use the terminology of limits to define derivatives and definite integrals, is it sufficient if students understand the derivative as that value to which the average rates are getting closer or the definite integral as that value to which Riemann sums get progressively closer? There can be some rough edges that may need to be dealt with individually such as the belief that the limit definition of the derivative does not apply to linear functions and Riemann sums cannot be used to define the integral of a constant function (since they give the exact value, not something that is getting closer), but it may well be that students with this understanding of limits do okay and get what they need from the course.

There has been one very thorough study that directly addresses this question, published by Michael Oehrtman in 2009. This involved 120 students in first-year calculus at “a major southwestern university,” over half of whom had also completed a course of calculus in high school. Oehrtman chose eleven questions, described below, that would force a student to draw on her or his understanding of limit. Through pre-course and post-course surveys, quizzes, and other writing assignments as well as clinical interviews with twenty of the students chosen because they had given interesting answers, he probed the metaphors they were using to think through and explain fundamental aspects of calculus.

The following are abbreviated statements of the problems he posed, all of which ask for explanations of ideas that I think most mathematicians would agree are central to understanding calculus:
  1. Explain the meaning of \( \displaystyle \lim_{x\to 1} \frac{x^3-1}{x-1} = 3. \) 
  2. Let \( f(x) = x^2 + 1.\) Explain the meaning of \( \displaystyle \lim_{h\to 0} \frac{f(3+h)-f(3)}{h}.\) 
  3. Explain why \( 0.\overline{9} = 1.\) 
  4. Explain why the derivative \( \displaystyle f’(x) = \lim_{h\to 0} \frac{f(x+h)-f(x)}{h}\) gives the instantaneous rate of change of f at x
  5. Explain why L’Hôpital’s rule works. 
  6. Explain how the solid obtained by revolving the graph of y = 1/x around the x-axis can have finite volume but infinite surface area. 
  7. Explain why the limit comparison test works. 
  8. Explain in what sense \( \displaystyle \sin x = x - \frac{1}{3!} x^3 + \frac{1}{5!} x^5 - \frac{1}{7!}x^7 + \cdots . \) 
  9. Explain how the length of each jagged line shown below can be \( \sqrt{2} \) while the limit has length 1. 
  10. Explain what it means for a function of two variables to be continuous.
  11. Explain why the derivative of the formula for the volume of a sphere, \( V = (4/3)\pi r^3 \), is the surface area of the sphere, \( dV/dr = 4\pi r^2 = A. \) 

In next month’s column, I will summarize Oehrtman’s findings. I then will show how they have led to a fresh approach to the teaching of calculus that avoids many of the pitfalls surrounding limits.

Hardy, N. (2009). Students' Perceptions of Institutional Practices: The Case of Limits of Functions in College Level Calculus Courses. Educational Studies In Mathematics, 72(3), 341–358.

Oehrtman M. (2009). Collapsing Dimensions, Physical Limitation, and Other Student Metaphors for Limit Concepts. Journal For Research In Mathematics Education, 40(4), 396–426.

Tall, D., & Vinner, S. (1981). Concept image and concept definition in mathematics with particular reference to limits and continuity. Educational Studies in Mathematics, 12(2), 151–169.


  1. Thanks for posting this! The Oehrtman questions at the end are especially awesome.

    I am a non-AP calculus teacher and after years of paring down limits more and more (for the reasons you cite), last year I finally decided to take the plunge and get rid of limits for the sake of limits, and focus even more on their conceptual meaning.

    I blogged about the intro activity that gets them thinking about the big ideas:

    and then how I structured the course so it worked:

    Sam Shah

  2. As Paul Lockhart wrote, in calculus we "explore the mathematics of motion, and the best ways to bury it under a mountain of unnecessary formalism. Despite being an introduction to both the differential and integral calculus, the simple and profound ideas of Newton and Leibniz will be discarded in favor of the more sophisticated function-based approach developed as a response to various analytic crises which do not really apply in this setting, and which will of course not be mentioned."

    Heaven forbid we show calculus to students the way Barrow and Newton saw it: an intuitive (albeit algebraically challenging) description of nature.

    Mathematicians were working out methods for finding tangent lines around 1660. The "epsilon-delta" formalism is from when -- 1820? We expect calculus students to make a 160-year jump in a week or so.

    Starting a calculus course with all of the formalism of limits is like teaching arithmetic beginning with Peano's Axioms.

    1. Does any one really over use the formal limit definition anymore? Even in my day (1970's) limits didn't break the course, even though just about everything stated in David's post was true. The crisis I would see is not teaching limits at all. You know, the albeit algebraically challenging stuff.:)

      This is a great discussion, but sadly, the real crisis right now in AP calculus is not limits, it is students that can't add fractions.

  3. Dear David,

    This was some of the most interesting I have read in years! At the Univ of Oslo I teach a class on the mathematically difficult parts of the K-12 curriculum, which in Norway includes a fair bit of calculus. I will add some of the Oehrtman questions to my list for next year. Thanks!

  4. Limits are not at all required in calculus. Both the derivative and integral are well defined in the New Calculus. There is no use of any of the ill-formed concepts: infinity, infinitesimal or limits. The New Calculus is the first and only rigorous formulation of calculus in human history.

    If you must teach the limit idea, then you'd be better off learning about Gabriel's New Limit theorem for mainstream calculus:!Gabriels-New-Limit-Theorem-replaces-all-previous-limit-theorems-in-mainstream-calculus/c9q/D150FC37-5EFE-412D-B2A1-604F7F7E842E

  5. Thanks David.

    Your comments make sense as well as bringing some educational studies and theory into the discussion. Your conclusion that the current long standing conviction that treating limits extensively at the beginning of a first calculus course is mistaken has been a thesis I have supported for over 30 years. Most current STEM students are still developing an understanding of numbers and functions when they start a course in calculus. This growth can be nurtured further, but cannot be forced along by premature introduction of concepts that took hundreds of years to formulate at current standards of rigor. More experience with estimation and some of the language that supports the understanding of limits is what is needed at this early stage, not more rigor and formalities.

    In an editorial published during the heyday of calculus reform I remarked on three central themes for a beginning calculus course- differential equations, estimation, and modeling. [See Flashman, Martin. "A Sensible Calculus," The UMAP Journal, Vol. 11, No. 2, Summer, 1990, pp. 93-96. ]

    My own materials on calculus (still developing after over 30 years) have followed an approach that focuses attention on these three themes while reducing and/or eliminating more formal and unhelpful treatments that spend time and student energy on a calculus of limits. [See The Sensible Calculus Program - ]

    I have not been alone in recognizing and avoiding the obsession with limits. Several of the "reform" calculus texts have also removed those lengthy treatments of limits from their contents while refocusing the treatment on more thematic approaches.

    It is a fiction that first course in calculus for the general STEM student should be as rigorous as a first course in analysis for a mathematically intense student and that this approach has benefits that make it essential.

    What is unfortunate is the continued dominance of textbooks that maintain this fiction. This dominance will likely continue until the faculty who select and control these texts and syllabi reorient their choices to available and developing resources for learning calculus that better suit their students.

    I look forward to your next column and hope that you will include some comments on existing materials that already avoid "many of the pitfalls surrounding limits."

  6. It has been known for some time that the "formal definition" of limits is not comprehended by Calculus I students, if only for the fact that this is their first (and only) exposure to formal mathematics. What I fear though is that teachers are confusing the situation with the formal definition of limits with all of the very necessary algebraic exercises students go through to find limits. I can't produce a better argument for those informal algebraic limit exercises than Oehrtman's 11 questions.