You have read the table: [a, b] is closed, (a, b) is open, and the half-open forms are [a, b) and (a, b]. You have also seen the picture: filled dots for included endpoints, hollow dots for excluded ones. What is missing is the reflex — the feeling that a square bracket is a filled dot and a round bracket is a hollow dot, and that flipping one changes the exact set of real numbers you are naming.
This page is a small workshop for that reflex. Slide the two endpoints anywhere on the line. Click a dot to toggle it between filled (closed, square bracket) and hollow (open, round bracket). The notation at the top rewrites itself as you do, and the inequality form — a < x ≤ b, say — rewrites with it. After a minute of play, the four interval types stop being four separate table rows and become one object with two switches.
The builder
What each bracket is telling you
The bracket character next to a number is a one-bit switch — it says "this number is in the set" (square) or "this number is not in the set" (round). That is the whole story. Everything else is consequence.
When you slide b towards a and both are closed, the interval shrinks continuously. At the moment a = b with two filled dots, the set has collapsed to \{a\} — the single point, which the notation [a, a] captures exactly. Slide further so b < a and the interval is empty: an interval with its left endpoint to the right of its right endpoint contains nothing. Try it in the builder and watch the notation keep updating even after the visual band disappears.
Now keep a < b but click the right dot to make it open. The segment looks almost identical — the difference is literally one point, the endpoint b itself. But mathematically the set has changed: [a, b] includes b, [a, b) does not. If you are defining a function like f(x) = \frac{1}{b - x}, the difference matters: f is defined on [a, b) but not on [a, b], because at x = b the denominator is zero. A single excluded point can be the entire content of a theorem. Play with the right-endpoint toggle a few times and watch the length of the interval not change at all while the set changes by exactly one element — that is the whole conceptual payload of interval notation in one gesture.
The language bridge: brackets versus inequality signs
There is a one-to-one dictionary between the bracket and the inequality sign.
| Endpoint | Dot | Bracket | Inequality at that end |
|---|---|---|---|
| Included | filled | [ or ] | \le or \ge |
| Excluded | hollow | ( or ) | < or > |
So the interval (a, b] — round on the left, square on the right — is exactly the inequality statement a < x \le b. Round bracket on the left pairs with strict <. Square bracket on the right pairs with non-strict \le.
The builder shows this pairing live. Toggle the left endpoint between open and closed and watch the inequality switch from a < x to a \le x without the right side changing. That is the whole grammar of interval notation — two independent switches, one per endpoint, each choosing between strict and non-strict.
Four endpoints, four shapes, one object
Before the builder, the four intervals [a,b], (a,b), [a,b), (a,b] look like four separate definitions from a textbook table. After a minute of clicking, they reveal themselves as one object with two independent binary flags — a two-bit description space.
- Both closed \rightarrow [a, b], the closed interval.
- Both open \rightarrow (a, b), the open interval.
- Left closed, right open \rightarrow [a, b), half-open.
- Left open, right closed \rightarrow (a, b], half-open.
Four combinations of two bits. That is why mathematicians almost never memorise the four names — you memorise the rule (square = include, round = exclude) and rebuild the name in the moment you need it.
Why the distinction matters: two quick scenarios
Scenario 1. The maximum of a function. Consider f(x) = x on the closed interval [0, 1]. The maximum is 1, achieved at x = 1. Now consider the same function on the open interval (0, 1). There is no maximum. You can get arbitrarily close to 1 — try 0.9, 0.99, 0.999 — but you can never reach it, because 1 is not in the set. The difference between filled and hollow at x = 1 is the difference between "maximum exists" and "maximum does not exist." That single bit is what the extreme value theorem depends on.
Scenario 2. Solving inequalities. When you solve 3x - 6 \ge 0, the answer is x \ge 2, which is the interval [2, \infty) — square bracket at 2 because the inequality is non-strict. If the problem had been 3x - 6 > 0, the answer would be x > 2, the interval (2, \infty) — round bracket because the inequality is strict. The bracket you choose is determined by whether the original inequality includes equality. Graders mark half credit off for closed versus open errors. Don't lose the marks.
A drill you can run on the builder
Try to produce each of these intervals in the widget, then check the readout:
- The closed interval [-2, 3] — both sliders in position, both dots filled.
- The open interval (0, 1) — both hollow.
- The half-open interval [0, 5) — left filled, right hollow. This is the set of numbers you use when counting from 0 up to but not including 5, like array indices
0, 1, 2, 3, 4. - The single point \{4\} — slide both sliders to 4 and close both endpoints. The interval is [4, 4], which contains only the number 4.
- The empty interval — slide b < a with any endpoint configuration. The readout shows the notation, but no real numbers satisfy the inequality.
After drill (3), a reflex should start forming: filled dot on the included end, hollow on the excluded end, and the bracket matches. You stop needing the table.
The connection to "strictly" and "at most"
Informal mathematical English has two distinct phrases that map cleanly onto the bracket choice. When a question says "x is strictly less than 5," you should hear "round bracket at 5" — strict inequality, endpoint excluded, (-\infty, 5). When a question says "x is at most 5" or "x does not exceed 5," you should hear "square bracket at 5" — non-strict inequality, endpoint included, (-\infty, 5].
The word strict always means a round bracket. The phrases at most, at least, no more than, no less than always mean a square bracket. These phrases appear in competitive exam questions in Hindi-English-Bengali blended Indian classroom English all the time — translate them into brackets before you write the answer.
On the builder, open the right endpoint to produce (a, b) and read the inequality as "a < x < b" — then toggle it closed and watch the right sign become \le. The bracket and the sign move together; neither moves alone.
The infinity side: why a bracket next to \infty is always round
The builder only handles finite endpoints, but the same reading extends to infinity. When you write (-\infty, 3], you are saying "everything from a left end that has no last point, up to and including 3." The round bracket next to -\infty is forced: -\infty is not a real number, so it cannot be an element of any subset of \mathbb{R}. You cannot include something that is not a valid candidate for membership, so the bracket next to any infinity is always round.
The one-bounded-one-unbounded intervals have only four forms: (-\infty, b), (-\infty, b], (a, \infty), [a, \infty). The bounded endpoint gets one of two brackets; the unbounded endpoint always gets a round bracket.
Back to the main article
Interval notation is just two switches — include the left endpoint or not, include the right endpoint or not — and a clean dictionary that maps the switches to bracket characters and inequality signs. Once the builder makes the switches tangible, every table row in the main article becomes obvious, and you stop second-guessing which bracket to write when you solve -2x + 7 \ge 1 or |x - 3| < 4. The interval is the target set, the bracket is the door, and square versus round is whether the door is open or closed at each end.
Related: Intervals and Inequalities Preview · Absolute Value Inequalities · Quadratic Inequalities · Supremum of Open Interval Exists, Max Does Not