# How does a cellular automaton "know" when to halt?

I came across the (proven) claim that the cellular atomaton "rule $110$" is universal, meaning that it can do every calculation in principle. But to calculate a result, the automaton must "halt" somehow.

How does the automaton "know" when it has to halt, in other words, how is "halting" simulated in this automaton ?

A turing machine has the "halt"-state but "rule $110$" apparently only can manipulate the symbols.

The 110 never halts. The machine will, for instance, continue forever toward the left filling in $1$'s. So the only stable configuration with finitely many $1$'s (or infinitely many, for that matter) is the configuration with no $1$'s at all.