What determines consonant doubling in English? This question is pursued by using a large lexical database to establish systematic correlations between spelling, phonology and morphology. The main insights are: Consonant doubling is most regular at morpheme boundaries. It can be described in graphemic terms alone, i.e. without reference to phonology. In monomorphemic words, consonant doubling depends mostly on the word ending. Certain endings correlate with double consonants (e.g. <er> as in <summer>), while others correlate with single consonants (e.g. <it> as in <visit>). What is more, it is the graphemic form of the word ending that determines the presence or absence of double consonants: The word endings <-ic> and <-ick>, for example, are homophonous, but the former almost always occurs with single consonants (e.g. <panic>), the latter with double consonants (e.g. <derrick>). That makes graphemic word endings peculiar entities: Like suffixes, they are recurring and they have distributional properties—but unlike suffixes, they have no morphosyntactic or semantic function. (PsycINFO Database Record (c) 2019 APA, all rights reserved)