I am trying to understand the following example
Let $X=\{1,2,3 \}$ and $\mathscr{S}=\{ \{1,2\},\{2,3\} \}$ so that $\cup_{S\in\mathscr{S}}S=X$. If we take the collection $\mathscr{B}$ of all sets that can be obtained by intersecting a finite number of members of $\mathscr{S}$ we have $\mathscr{B}=\mathscr{S}\cup\{X,\{2\} \}$.
Now I can see why $\mathscr{S}$ in $\mathscr{B}$ as $\{1,2\}\cap\{1,2\}=\{1,2\}$ and $\{2,3\}\cap\{2,3\}=\{2,3\}$. Similarly, $\{1,2\}\cap\{2,3\}=\{2\}$, but how do we get $X$ in $\mathscr{B}$?
EDIT: I think there is a bit of confusion. To set things correct $\mathscr{S}$ is a collection of subsets of $X$ such that $\cup S=X$. This example is given in Stromberg Introduction to Classical Analysis, p.96 (example f).
There is no way to get $ X$ in $\mathscr{B}$ because we can not describe $X$ as an intersection of members of $\mathscr{S}$
This $$\mathscr{B}=\mathscr{S}\cup\{X,\{2\} \}$$
is simply not true unless we change the usual definition of intersection to include $X$ an mentioned in Arnaud Mortier 's solution..