Let $G$ be a group and let $B$ be a subgroup of $G$. Let $\Delta $ be the set of all right cosets where $\Delta = \{Pg: P \text{ is a subgroup of } G \text{ containing } B, g \in G \}$. We order $\Delta$ by reverse inclusion so we write $Pg \preceq P'g'$ if and only if $P'g' \subseteq Pg $.
Now given $Pg \in \Delta$ and subgroup $P' \in \Delta$ I have shown the the greatest lower bound of $Pg$ and $P'$ is $P''= \langle P', Pg \rangle$. The proof I am reading then goes on to say that this implies that any two elements of $\Delta$ has a greatest lower bound. How is this the case.
I thought maybe somehow using the greatest lower bound of $Pg$ and $P'$ and then taking the greatest lower bound of $P'g'$ and $P$ and somehow using this to give a greatest lower bound of $Pg$ and $P'g'$ but I don't think that works.
Any ideas here on what to do?
Let $Pg,P'g'\in \Delta$. We must find a greatest lower bound.
To do this, observe the following statements are equivalent:
To see this, note that (1) and (2) are equivalent by the fact that right multiplication preserves inclusions, and (3) and (4) are equivalent for the same reason. (2) and (3) are equivalent by the fact that $H$ is the greatest lower bound for for $P$ and $P'g'g^{-1}$ under reverse inclusion, as you have already proved. Thus all four statements are equivalent.
But now the fact that (1) and (4) are equivalent says precisely that $Hg$ is the greatest lower bound for $Pg$ and $P'g'$.
(Keep in mind that we know $Hg$ is itself a lower bound since if we let $P''g''=Hg$, then (4) is trivially satisfied, hence so is (1).)