This contribution aims to highlight a neglected aspect of Samuelson’s famous 1937 paper “A Note on Measurement of Utility”. Although the 1937 paper is usually regarded as the foundation of discounted utility theory, and rightly so, it is primarily concerned with utility measurement and deals only indirectly with decision over time – intertemporal issues appearing as a by-product of the realisation of a unique utility measure. But the treatment of discounted utility in turn influenced Samuelson’s understanding of cardinality. Cardinality appears here as the result of a cognitive ability that manifests when agents face a decision experiment over time in which they are compelled to cardinalize their utility functions. The result is the weak plausibility of cardinality in a more general context, such that, contrary to the usual views, we may say that Samuelson’s ordinalist approach was already in the making in 1937.