@SteamKing But does it not matter that x is a function of t, and (through the relation with r) a function of y? And the only way the book's answer makes sense is if you substituted rcos(t) = ycot(t) into x^2 then differentiate that, treating 2y^2 as a constant, which knocks it out completely.