The smallest representable "float" value (FLT_MIN) is something like 1.1x10-38, but the representation only guarantees about six or seven places of decimals.
So, for instance, you can represent 1.0 in one "float" variable and FLT_MIN in another, but if you add them together, the result is still 1.0.
You can, however, add FLT_MIN to FLT_MIN and get 2 * FLT_MIN, just as you can add 1.0 to 1.0 and get 2.0