The & is the address-of operator. It returns the address of the variable that stores a certain value. Normally, one would do float *f_ptr = &f_var; to get a float pointer to a value of type float. If you instead use long *l_ptr = (long *)&f_var; you get a pointer of type long* pointing to the same variable (which is still a float). This is dangerous and will rarely give the expected result, for two reasons: a) sizeof(long) might be larger than sizeof(float), causing an access to l_ptr to result in undefined behavior. b) Reading the value with long l = *l_ptr will be mostly useless, since a float is not a long, and no value conversion took place. The value of l will not be the integer part of the float, as you might expect.

This double *d_ptr = (double *)l_ptr; has similar problems: While now, both sides are pointers, interpreting a pointer to a long as a pointer to double is only useful in very rare cases (e.g. when working with unions). And double is now for certain longer than float, so doing double d = *d_ptr will result in undefined behavior.

Bottom line: Neither of these instructions are useful, at least certainly not in beginners code.