You read the sentence "the rationals are dense in \mathbb{R}" and a voice at the back of your head says: wait — if they are everywhere, aren't they everything? Isn't \mathbb{Q} = \mathbb{R} then? The voice is wrong, but it is wrong for an interesting reason, and until you have explicitly killed it, the word "dense" will keep leaking into claims it was never meant to support.
So let us kill it now.
The claim that sounds reasonable but isn't
Here is the misconception, stated cleanly: if a set S is dense in \mathbb{R}, then S = \mathbb{R}. Even sharper: if S is open and dense in \mathbb{R}, then S = \mathbb{R}.
Both versions feel plausible because of the everyday meaning of "dense." In a dense crowd, you cannot walk between two people without bumping into a third; in a dense forest, there are no empty clearings. Transplant that picture to the number line and you get: if rationals are dense, there is no room left for anything else.
The mathematical meaning is different, and the difference is the whole point of the word. Mathematical density is a statement about arbitrary closeness, not about filling. It says: pick any real number, demand any positive error tolerance, and I can produce an element of S within that tolerance. It does not say: every real number is itself in S.
The one example that settles it
Take S = \mathbb{Q}, the rationals. They are dense in \mathbb{R} — between any two real numbers there is a rational one (see the animated proof). If the misconception were true, that would force \mathbb{Q} = \mathbb{R}.
Now consider \sqrt{2}. You know from Number Systems that \sqrt{2} is not rational — the classical proof assumes \sqrt{2} = p/q in lowest terms, derives that both p and q must be even, and collides with "lowest terms." So \sqrt{2} \in \mathbb{R} but \sqrt{2} \notin \mathbb{Q}.
That single number is enough. \mathbb{Q} is dense in \mathbb{R} and \mathbb{Q} \ne \mathbb{R}. Density did not force equality. The misconception is dead.
What density actually says
Here is the definition, stripped down.
A set S \subseteq \mathbb{R} is dense in \mathbb{R} if, for every open interval (a, b) \subseteq \mathbb{R} with a < b, there is an element of S inside (a, b).
Read that again carefully. The claim is about intervals. It says every open interval contains a point of S. It says nothing about individual points of \mathbb{R} — it does not say every real number is itself an element of S.
Equivalently: S is dense in \mathbb{R} if \mathbb{R} \setminus S has no "neighbourhoods of absence." There is no open interval — no matter how tiny — that S completely misses. But individual points x \in \mathbb{R} \setminus S are fine. You can miss a point; you just cannot miss a whole interval around that point.
That is why the rationals can be dense and still omit \sqrt{2}. They get into every interval around \sqrt{2}, no matter how thin, but the single point \sqrt{2} itself is not among them.
Why the mismatch feels confusing: in everyday language, "close" usually implies "eventually touching." Mathematically, "arbitrarily close" is carefully built to mean closer than any positive number you name — without ever having to be zero. The rationals are within 10^{-100} of \sqrt{2}, within 10^{-10^{100}}, within any \varepsilon > 0 — but they are never at distance exactly 0 from \sqrt{2}. That last gap is the whole content of "dense but not equal."
A second example: irrationals are also dense
If the misconception were right, two dense subsets of \mathbb{R} would both have to equal \mathbb{R}, and hence equal each other. So every dense subset would be identical. That is already suspicious — and in fact it is wrong in the sharpest possible way.
The irrationals \mathbb{I} = \mathbb{R} \setminus \mathbb{Q} are also dense in \mathbb{R}. Between any two real numbers there is an irrational. (Real Numbers — Properties gives the shift trick: find a rational between a - \sqrt{2} and b - \sqrt{2}, then add \sqrt{2} back.)
But \mathbb{I} is obviously not \mathbb{R}. The number 0 is real. The number 0 is rational (0 = 0/1). So 0 \in \mathbb{R} but 0 \notin \mathbb{I}. Same with 1, \tfrac{1}{2}, -3, \tfrac{22}{7}. The rationals are exactly the reals missing from \mathbb{I}.
So you have two sets, \mathbb{Q} and \mathbb{I}, that are both dense in \mathbb{R}, neither equal to \mathbb{R}, and — here is the kicker — they are disjoint. Their intersection is empty. They are both "everywhere" in \mathbb{R}, yet they share not a single point. That is flatly impossible if "dense" means "equal." It is perfectly consistent if "dense" means "gets arbitrarily close."
Even open dense sets miss points
The sharper version of the misconception asks about open dense subsets. Maybe adding "open" fixes things — open sets feel beefier, more like solid intervals.
It does not fix things. Here is the example: S = \mathbb{R} \setminus \mathbb{Z}, the real line with the integers removed.
- S is open: it is the union of the open intervals \ldots, (-2, -1), (-1, 0), (0, 1), (1, 2), \ldots, and a union of open sets is open.
- S is dense in \mathbb{R}: every open interval (a, b) with a < b contains some non-integer. (If b - a < 1, the interval can contain at most one integer; pick any point of (a, b) that avoids it — there are uncountably many.)
- S \ne \mathbb{R}: 0 \in \mathbb{R} but 0 \notin S. Same for every integer.
So S = \mathbb{R} \setminus \mathbb{Z} is an open dense subset of \mathbb{R} and it still misses points — countably many of them, namely every integer. Adding "open" to "dense" does not close the gap between density and equality.
Why the integers can be "missed" without creating an empty neighbourhood: each integer n is surrounded by non-integer points on both sides. However close to n you look, you will find plenty of non-integers — in fact, all but one point in any tiny interval around n is non-integer. So \mathbb{R} \setminus \mathbb{Z} crowds right up against each integer without containing it. Density is satisfied; equality is not.
The pattern: "dense in" versus "equal to"
Put the three examples side by side:
| Set S | Dense in \mathbb{R}? | Equal to \mathbb{R}? | Missing from S |
|---|---|---|---|
| \mathbb{Q} | Yes | No | every irrational |
| \mathbb{I} = \mathbb{R} \setminus \mathbb{Q} | Yes | No | every rational |
| \mathbb{R} \setminus \mathbb{Z} (open) | Yes | No | every integer |
| \mathbb{R} | Yes (trivially) | Yes | nothing |
Three of the four are dense subsets of \mathbb{R} that are not \mathbb{R}. The first is countable and not open; the second is uncountable and not open; the third is uncountable and open. None of these properties — countability, uncountability, openness — forces density to become equality.
The only statement that forces S = \mathbb{R} is S = \mathbb{R} itself. There is no side door.
Why this confusion is worth naming
The reason to burn out this misconception is that it shows up in disguise all over real analysis. Whenever you are told "sequence of rationals \{a_n\} converges to L," your instinct should say L is real, but not necessarily rational. Whenever you see "\mathbb{Q} is dense in \mathbb{R}," your instinct should say close, yes; identical, no. Whenever you see "the computable numbers are dense in \mathbb{R}" or "the algebraic numbers are dense in \mathbb{R}," your instinct should not jump to "so every real is computable" or "so every real is algebraic." Neither conclusion follows.
The slogan to file: density is a closeness statement, not an equality statement. Two sets can both be dense, disjoint from each other, and neither equal to \mathbb{R}. The number line has enough room for several dense populations to live interleaved without overlapping — and several dense open populations to live interleaved while still leaving points uncovered. That is the geometry of \mathbb{R}, and misreading "dense" as "equal" flattens it into a sentence that is simply false.
Related: Real Numbers — Properties · Dense But Full of Holes · Rationals Are Dense in ℝ — Animated Proof · Number Systems