What a Complete Limit Argument Has to Establish

Put $x = 2$ into that expression and the denominator is $2 - 2 = 0$, so the expression has no value there. Then 4 cannot describe what happens at 2; it describes inputs near 2, and that shift is the whole problem.

Here is what makes 4 correct. The numerator factors, since $(x-2)(x+2) = x^2 - 4$, so for every input other than 2 you may divide top and bottom by $x - 2$:

That short condition carries the argument. The original expression and $x + 2$ are the same function except where the original is undefined. A limit at 2 depends only on inputs near 2, never on 2 itself, so the two have the same limit. And $x + 2$ is a polynomial, whose limit at 2 is its value:

Three separate things had to be true; the bare 4 asserts none.

Two researchers described the gap between the picture of a limit in a student's head and the written definition, and applied it to limits and continuity as taught in school and at university^[1]. That names the problem without measuring how often it happens^[1]. The picture feels like knowledge; it never was the definition.

A study of university mathematics students who had finished a calculus course sorted their ways of picturing a limit into groups^[2]. One group treated the limit as the function's value at the point; another as a procedure you carry out^[2]. The study describes these ways of thinking; it does not count them^[2]. If your limit simply is the substitution, answer and argument are one object, and nothing was left out.

A researcher looked at how 120 students in introductory calculus described their thinking, in writing and out loud, and found a handful of everyday pictures: something collapsing away to nothing, or a scale so small nothing exists below it^[3]. Which picture a student used showed up in the claims they made and the reasons they gave^[3]. The private picture surfaces on the page.

In one study, ten calculus students at a university were interviewed, then shown other ways of thinking about limits and given problems their own way could not handle^[4]. Their moving, sliding-towards picture barely shifted^[4]. One reason the researcher gives: they read the awkward problems as small exceptions rather than a sign the picture was wrong^[4]. So the rest of this post is a writing habit, not a lecture.

In two teaching sessions where small groups of students built a definition of a limit themselves, the researchers came away with an account of how: they had to stop starting from the input variable, and stop needing the endless approaching to be finished^[5]. That describes a few students, untested across classrooms^[5].

Both halves appear in "as $x$ gets closer to 2, the value gets closer to 4": it starts from $x$ and leans on a process nobody can finish.

Start from the output instead. Say you want the value within $0.01$ of 4. For $x \neq 2$ the expression equals $x + 2$, and

so the value is within $0.01$ of 4 exactly when $x$ is within $0.01$ of 2 and $x \neq 2$. A tighter tolerance is answered by the same line. Nothing must be finished; you produce a distance on demand.

Consider $\lim_{x \to 0} \frac{\sin x}{x}$. Writing "by L'Hôpital's Rule" here is circular: the derivative of $\sin$ at 0 is defined as $\lim_{h \to 0} \frac{\sin h - \sin 0}{h} = \lim_{h \to 0} \frac{\sin h}{h}$, the very limit you were asked for, so the rule assumes its conclusion. The commoner failure is plainer — the rule is quoted where the form was never $\frac{0}{0}$ or $\frac{\infty}{\infty}$, and the line checking it is the one nobody wrote.

A squeeze does not assume its conclusion. For $0 < x < \frac{\pi}{2}$,

and both $\frac{\sin x}{x}$ and $\cos x$ are unchanged if $x$ becomes $-x$, so the same bounds hold just below 0. Both bounds tend to 1 — $\cos x \to 1$, the upper bound is constantly 1 — so by the squeeze theorem the middle is trapped and the limit is 1.

A review that pooled many studies found that prompting students to explain to themselves why something works gives a small to moderate improvement on math tests taken right afterwards, bigger when they are first shown what a good explanation looks like^[6]. Evidence that it helps in a real classroom, or that the gains last, was much thinner^[6]. Its authors also recommend asking why a common wrong idea is wrong — a recommendation from the wider literature, not from this review^[6].

In two classroom experiments, algebra students on a computer tutor were put at random into versions of the same unit^[7]. Those who studied finished solutions and explained why each was right or wrong came out understanding the concepts better than those given practice alone^[7]. Wrong solutions looked especially useful, though the researchers put that more cautiously^[7]. That was first-year algebra in a tutoring system, not calculus and not limits: borrowed evidence.

Not every limit needs the extra sentence. A polynomial is defined at its target and its limit there is its value, so $\lim_{x \to 3} (2x + 1) = 2(3) + 1 = 7$ is the whole argument: nothing was rewritten, so nothing needs a condition. The habit earns its keep where substitution fails, and nothing where only the answer is asked for.

Put a bare $= 4$ beside the three-line version, write the sentence it leaves unsaid, then check:

• Does a line say which inputs the rewriting is valid for?
• Does a line say why the final substitution is allowed?
• If a theorem is named, is its condition checked on paper?

That is my own judgement, not a finding: a clause per line, usually the step you trusted least.

The studies behind this post looked at how students think and talk about limits, not at how anyone marks them^[4][3]. None examined markers or marking schemes, and nothing here tells you what a course will accept. The only local evidence is the worked solutions you were handed: read one and count its lines of argument per limit.

The studies describing students' pictures of limits report which pictures turned up, not how many students hold each^[2][3]. The framework naming the gap between picture and definition describes it rather than counting it^[1]. The resistance finding rests on interviews with ten university calculus students^[4].

The writing advice sits on the thinnest ground. The self-explanation review measured tests taken right after the work, with much less support for classrooms and for later recall^[6], and explaining something to yourself is not writing an argument someone else reads. The example experiments ran in first-year algebra with a computer tutor^[7]. No study here shows that writing "for $x \neq 2$" changes anything.