Dutch ordinals are constructed by adding one of two suffixes, -de or -ste, to thecorresponding cardinal. The distribution of the two suffixes over the ordinalparadigm and the motivations behind it are not evident and no serious attemptshave been made so far in the literature to formulate an analysis which covers allthe facts. Works like Barbiers (2007), Booij (2010) and Zonneveld (2007) havebriefly touched upon the subject and hinted at possible phonologicalexplanations, in terms of optimization processes and OT rankings, which do notseem to hold. Instead, I propose that for Dutch, Brabantish and West Flemish (twodialects of Dutch) and a number of other Germanic languages, the suffix selectiondivides the ordinal number lines of each language into morphosyntacticallydistinguishable subparts; and that the driving force behind this selection is anunderlying morphological complexity of some of the cardinal stem forms. Theprimary research question is can we find evidence for morphological complexityamong the cardinal numbers of the Germanic languages? In my talk I will show the data from Standard Dutch, the two dialects (Sleeman,2015), and from some (not all) of the other Germanic languages that I will belooking at. The latter include German, Afrikaans, Frisian, English, Swedish, Danish,Nynorsk, Bokmål, Icelandic and Faroese. I will present where the similarities anddifferences lie, and put the patterns in a theoretical perspective: I will demonstratesome phonological Optimality Theory attempts at predicting the forms and thenshow that these predictions are not borne out. I will compare the predictions ofthe OT framework to the predictions made by a different framework calledmorphological subcategorization (Paster, 2006, 2009; Bye, 2007), which hopefullywill do a better job at explaining the data. In this framework, suffixes determinetheir distribution by means of requirements to their potential stems. I propose thatsome of the ordinal suffix allomorphs in the Germanic languages require theirstems to be ...