This is faster than the (higher scoring, as I type) alternative using .max whenever there's a significant probability of a false result.
(...and more succinct than my idea using .each, too!)