You have been told an interval is a "connected stretch of the number line." The picture in your head is a ribbon of ink from a to b, maybe with solid dots at the endpoints, maybe hollow. Ribbons have length. Ribbons contain points. So when someone writes (3, 3) and says "this interval is empty," your instinct is to push back — a ribbon from 3 to 3 is still something, even if it is only one point. Right?
No. The something you are picturing is \{3\}, a single point, and that is what [3, 3] means. The notation (3, 3) is different. It means "every real number strictly greater than 3 and strictly less than 3." Read it out loud: greater than 3 and less than 3. No number does both. The set is empty. You write it as \varnothing, the empty-set symbol, and move on.
This article is about when this happens, why it is not a bug in the notation, and why the empty set is a perfectly respectable answer to a perfectly respectable question.
Parsing (a, b) as a condition, not a picture
Interval notation is shorthand for a set-builder condition. The four flavours mean:
- [a, b] = \{x \in \mathbb{R} : a \le x \le b\}
- (a, b) = \{x \in \mathbb{R} : a < x < b\}
- [a, b) = \{x \in \mathbb{R} : a \le x < b\}
- (a, b] = \{x \in \mathbb{R} : a < x \le b\}
Plug in a = b = 3 and see what falls out.
- [3, 3] = \{x : 3 \le x \le 3\} = \{3\}. One element.
- (3, 3) = \{x : 3 < x < 3\} = \varnothing. Zero elements.
- [3, 3) = \{x : 3 \le x < 3\} = \varnothing. Zero elements.
- (3, 3] = \{x : 3 < x \le 3\} = \varnothing. Zero elements.
Only the fully closed interval survives — because it is the only one that allows x to equal the endpoint. The other three demand a strict inequality on at least one side, and a number cannot be strictly less than itself.
The visual picture is: closed brackets are like doors that are open; round brackets are like doors that are closed. At a degenerate interval where both ends sit at the same number, you can only get in if at least one door is open — otherwise there is no room in the interval for the endpoint itself. The fully closed [3, 3] lets 3 squeeze through. Every other combination locks it out.
The backwards interval: (5, 2)
The second way to get an empty interval is to write one backwards. What does (5, 2) mean? Applying the same dictionary:
Read it: x greater than 5 and less than 2. There is no real number that is simultaneously bigger than 5 and smaller than 2. The set is empty. The same holds for [5, 2], [5, 2), (5, 2] — whenever a > b, every flavour of interval is empty.
This is why the textbook convention always says "for a < b, define [a, b] = \{\ldots\}." The definition still technically works when a > b, but it produces nothing. Some books insist "interval" means a \le b by definition, ruling (5, 2) out as malformed. Others allow a > b and call the result the empty interval. The underlying set is the same either way: \varnothing.
In practice, a backwards interval often appears as a symptom — a sign that you made an algebra mistake upstream. If you solved an inequality and ended up needing x \in (5, 2), somewhere a sign flip got forgotten, or a case got crossed.
The empty set is a real answer
The bigger idea is this: the empty set \varnothing is not a mistake, a null, a "no answer yet," or a placeholder. It is a perfectly valid set, and for many problems it is the correct final answer.
Consider: solve x^2 < 0 for real x.
The question is legitimate. The inequality is written in standard form. You are asked for the set of real numbers that satisfy it. You reach for your toolkit — square both sides? No, the left side is already a square. Check the sign: x^2 \ge 0 for every real x, with equality only at x = 0. So x^2 < 0 is never true. The solution set is:
Writing \varnothing here is not dodging the problem. It is the problem's answer. The inequality has no real solutions; the set of things satisfying it is empty.
The same thing happens constantly:
- |x| = -1 has solution set \varnothing. (Absolute value is non-negative.)
- x^2 + 1 = 0 has solution set \varnothing over \mathbb{R}. (Over \mathbb{C}, it has solutions \pm i — which is why the complex numbers were invented.)
- \sin x = 2 has solution set \varnothing. (Sine is bounded by \pm 1.)
- [1, 3] \cap [5, 7] = \varnothing. (The two intervals do not overlap.)
In all of these cases, the answer is not "error" or "no solution." It is "the empty set." Same noun as "the interval (0, 1)" — just a different set, with zero elements instead of infinitely many.
Why intersections produce the empty set naturally
The most common place you actually run into an empty interval is intersections. If you solve a system of inequalities, you are intersecting the two solution sets. Sometimes they do not overlap.
Suppose a problem asks for x such that x > 7 and x < 3. The first condition is (7, \infty), the second is (-\infty, 3). Their intersection:
The two rays face each other but do not overlap. The conjunction "and" demands both conditions hold — no x does. The answer is the empty set.
The point here is subtle but important: you did not write down an empty interval. You wrote down a reasonable system and computed its solution. The empty set popped out as the result. This is the empty set showing up organically, as the natural answer to a question about when two conditions are compatible.
Contrast with the degenerate-endpoint and backwards cases. Those are cases where you wrote an empty interval directly in notation — (3, 3) or (5, 2) — and the question was whether they mean anything. They do: they both mean \varnothing. They are two different notational routes to the same destination.
"But \varnothing feels like nothing"
This is the real conceptual hurdle. In everyday language, "empty" means absent, missing, not-there. When a bucket is empty, you might say "there is no bucket," which is wrong but colloquially close. The empty set is not absent — it is a set (a perfectly good mathematical object), it just happens to have no elements. It has a symbol (\varnothing or \{\}), a cardinality (0), and shows up in theorems all the time.
A useful analogy: 0 the number. Zero was weird when it was first invented — a number that counts nothing? But 0 is a number, and arithmetic without it is a mess. Similarly, \varnothing is a set, and set theory without it is a mess. You can intersect sets, union them, take subsets — and the "zero" for those operations is \varnothing. It belongs.
When a problem's answer is \varnothing, you have learned something real: the conditions you imposed are unsatisfiable together. That is information. "No real x satisfies x^2 < 0" is a theorem, not a failure.
Quick summary
- (3, 3), [3, 3), (3, 3] are all empty. Only [3, 3] is non-empty (it equals \{3\}, one point).
- (5, 2) and any interval with a > b are empty, because no real x can be both greater than 5 and less than 2.
- \varnothing is the natural answer when a problem has no solutions (e.g., x^2 < 0) or when two constraints have no overlap (e.g., x > 7 and x < 3).
- The empty set is not an error or a missing value. It is a bona fide set with zero elements, and it is the honest answer for many honest questions.
If you want to see more about why strictness and brackets matter, the round vs square bracket article drills the notation itself, and the \infty shorthand article covers a related "this looks weird but isn't" situation.