This type avoids loss of precision, and is likely what was intended by having 14.4 fixed point. Conversions to/from float64 are provided.