To date, no analysis has adequately accounted for the attested distribution of front umlaut in Old Scandinavian. In this study attention is paid to unexpected outcomes that defy the generally accepted rules. In particular, the complications posed by iʀ-umlaut are refined into an acid test against which existing hypotheses fail. A genuinely novel proposal is developed, based on the assumption that in prominent syllables contrast well into the umlaut period was upheld between descendants of Pre-Germanic (PreGmc) */e/ and */i/ respectively, even upon the Pre-Scandinavian raising of *e. Upon such raising the descendants of PreGmc */e/ had in all oral contexts evolved into a markedly fronted coronal vowel *ȋ, whereas in prominent syllables descendants of PreGmc */i/ had by default, with few exceptions, in a chain shift evolved into a non-umlauting dorsal vowel *ï. Given the assumption that a light second syllable within a main stressed bisyllabic foot was prominent, the two vowels *ȋ and *ï, active and inert as triggers for front umlaut respectively, could both have occurred in this position. By explaining their distribution in the lexicon, the notoriously intricate cruxes of i-umlaut may be neatly accounted for.