I know that Cov(i,j)=Beta(i)*Beta(j)*Variance(M). I cannot memorize:

Cov(i,j)=Beta(i,1)*Beta(j,1)*Var(F1)+Beta(i,2)*Beta(j,2)*Var(F2)+(Beta(i,1)*Beta(j,2)+(Beta(i,2)*Beta(j,1))*Cov(F1,F2)

Will we have to use this longer one, or is it safe to say just shorter one-factor version?