java - Primitives float and double: why does f+=d not result in Type Mismatch Error?

Possible Duplicate:
Java += operator

Code example:

    double d = 1;
    float f = 2;
    f += d;  // no error?

    f = f+d; // type mismatch error, should be f = (float) (f+d);

So why does f+=d not produce an error (not even at runtime), although this would decrease the accuracy of d?