In Ahn and Sauerland’s recent studies (Sauerland 2014; Ahn and Sauerland 2015), constructions of relative measures such as percentage nouns and fractions can give rise to two interpretations: one is consistent with the Conservativity Universal formulated by Keenan and Stavi (1986), but the other is not. In this paper, I propose a novel way to cash out the two interpretations. Specifically, on the basis of Mandarin data, I argue that the “conservative” interpretation involves measurement of individual objects, whereas the “non-conservative” interpretation involves measurement of event-related objects, i.e., 〈event, object〉 pairs (Doetjes and Honcoop 1997; cf. Barker 1999).