Consider pythagoras' theorem and a vector sum interpretation of the individual component gradients.
Thank you for the reply, but I'm still confused. So with pythagoras' theorem I would have Y=(x1,x2,x3,x4)? I know that for Y(s)=(sqrt(1-s^2),s) the gradient is (-s/sqrt(1-s^2),s) and the magnitude of the gradient is 1/sqrt(1-s^2), and I'm supposed to get an expression similar to this, but I'm not sure how I can rewrite the original equation. If I put r=(x2,x3,x4) then it would be Y(r)=(sqrt(1-r.r),r), and this is where I'm confused, how do I find the gradient of this?