请给个理由，这短简单代码的执行结果为什么调用I2F1（），而不是I1F1（）？

[flying_fox0]

class I1
{
public:
virtual void I1F1() = 0;
virtual void I1F2() = 0;
};

class I2
{
public:
virtual void I2F1() = 0;
virtual void I2F2() = 0;
};

class CA: public I1,public I2
{
public:
virtual void I1F1(){};
virtual void I1F2(){};
virtual void I2F1(){};
virtual void I2F2(){};

};
int main()
{
CA * a = new CA;
void *pv;

pv = (I2 *) a ;
((I1 *)(pv))->I1F1() ;

return 0;
}