The pen-never-lifts picture
Here is the friendliest test for continuity: imagine drawing the graph of a function by hand. If you can trace the whole curve without ever lifting your pen off the paper, the function is continuous. The temperature outside over a day, the height of a growing plant, the position of a thrown ball — these change smoothly, with no teleporting from one value to another, so their graphs are unbroken.
Now think about where the pen *would* have to leave the paper. Maybe the curve suddenly jumps up to a new level. Maybe there is a single missing dot — a tiny hole — where the function is undefined or sits at the wrong spot. Maybe the curve rockets off toward infinity near some value. Each of those is a break, and each is a place the function fails to be continuous. The whole formal story below is just a careful way of saying: at this point, the curve does not break.
Saying it precisely with a limit
On the previous rungs you met the limit: lim x->a f(x) is the single value the curve is *heading toward* as x closes in on a — regardless of what happens exactly at a. Continuity links that heading-toward value to the value the function actually takes. A function f is continuous at a when the place the curve is heading is the place it actually lands.
f is continuous at a <=> lim x->a f(x) = f(a)
- f(a) exists — the function is actually defined at a; there is a dot on the graph there, not a hole.
- lim x->a f(x) exists — the curve heads toward one definite value, the same from both sides (the left and right one-sided limits agree).
- They match — the value it heads toward equals the value it lands on: the limit equals f(a).
Three ways a curve breaks
When the equation fails, the failure has a shape, and it helps to name the common ones. A jump discontinuity is a step: the curve approaches one height from the left and a different height from the right, so the one-sided limits disagree and the limit does not exist. Think of a parking fee that is $2 for up to one hour and $5 the instant you pass an hour — the cost jumps.
A removable discontinuity is a single hole: the curve heads cleanly toward a value (the limit exists), but f(a) is missing or set to the wrong spot. It is called *removable* because you could patch the one point and make the function continuous. An infinite discontinuity is a vertical asymptote: near a the curve shoots off toward +infinity or -infinity, like 1/x near x = 0, so no finite limit exists at all.
g(x) = (x^2 - 1) / (x - 1) # equals x + 1 everywhere EXCEPT x = 1, where it is 0/0 (undefined). # lim x->1 g(x) = 2, but g(1) does not exist -> removable hole at (1, 2).
Why continuity earns its keep
Continuity is not just tidiness — it buys you guarantees. The headline one is the Intermediate Value Theorem: if f is continuous on the interval from a to b, then it takes *every* value between f(a) and f(b) somewhere along the way. An unbroken curve cannot get from one height to another without passing through all the heights in between — there is no door to skip through, because there are no gaps.
This sounds obvious, but it is a workhorse. Suppose f is continuous, f(a) is negative, and f(b) is positive. Since 0 sits between a negative and a positive number, the theorem guarantees there is *some* point c between a and b with f(c) = 0 — a root. That is exactly how you can pin down a solution to an equation you cannot solve by hand: bracket it between a minus and a plus, then squeeze.
- Find points a and b where f(a) < 0 and f(b) > 0 (a sign change). Continuity guarantees a root sits between them.
- Check the midpoint m = (a + b) / 2. Whichever half still has opposite signs at its ends must contain the root.
- Keep halving the interval. Each step doubles your accuracy, and the bracket closes in on the root as tightly as you like.