Since its beginning in the 19th century, when dealing with inherited lexicon, Romance etymology and etymography always promoted Latin etyma. In a quite revolutionary paper, J.-P. Chambon (2007) recommended recreating Romance etymology on the basis of comparative grammar. This paper applies this novel method of practicing Romance etymology by reconstructing the common ancestor of lexemes like Romanian (a) cădea "to fall", Italian cadere, French choir and Spanish caer. The methodological background is provided by linguistic reconstruction (Fox 1995), which surprisingly has never before been applied to Romance etymology. This research yields two results: on the one hand, it appears that the Proto-Romance etymon of all Romance cognates is one lexeme, namely */'kad-e-/ (as opposed to Latin cadĕre/cadēre given by previous authors). On the other hand two distinct morphological bases can be observed: */'kad-e-re/ and */ka'd-e-re/. One can then correlate these with two independent sets of data. The first is extra-linguistic, since it pertains to geographical areas conquered by the Romans at different periods of Antiquity, while the second concerns comparison with written Latin of the Antiquity. The */'kad-e-/ example will provide the opportunity for presenting the DÉRom (Dictionnaire Étymologique Roman), a new online dictionary of Romance etymology whose editors would like to establish contacts with other historical lexicographical projects.