How about drilling a column of ice from antarctica then extracting air from bubbles and measuring it? There are more abstract ways such as looking at stomatae on fossil plants, which are unlikely to be as reliable but can be used to corroborate findings.