The formula for finding resulting confidence interval includes upper and lower bound as 1/F(n1-1,n2-1) and F(n2-1,n1-1) But, as F(a,b)= 1/F(b,a) So F(n2-1,n1-1)=1/F(n1-1,n2-1) Therefore the upper and lower boumdary of sigma1