Slightly OT but regarding floats - I was wondering if we could get by just with an exponent and ditch the mantissa. Each number would be represented as an integ

Slightly OT but regarding floats - I was wondering if we could get by just with an exponent and ditch the mantissa. Each number would be represented as an integer exponent of 2^-n for n corresponding to the precision.