Oh, I hadn't noticed that emphasis. I guess I'm still stuck with the logarithm trick. I also just checked Montgomery and he reports that another simple transform is y^0.25 -- ie, square root of square root. I haven't tried it though.