On Thursday, 3 April 2014 at 20:13:35 UTC, Bill Buckels wrote: > The sfloat24 data type is half way between a float and a double > in storage size. Wait. 24 bits would be halfway between half (16 bits) and single precision (32 b its).