Am 06.01.2013 00:34, schrieb Bart: > On 1/5/13, Michael Van Canneyt <michael at freepascal.org> wrote: > >> Because they don't fit in 32 bits. > > Which (to me) raises the question: why do "true ordinal" values have > to fit in 32 bits? Because operations like array accesses require that they fit into one register.