Hey TheCountador.
Consider finding the random Z where Z is the distribution when conditioned and then get Var[Z] = E[Z^2] - E[Z]^2 using standard formulae.
I know the unconditional variance formula of the form:
((n)(r)(N-r)(N-n)) / ((N)(N)(N-1))
So, what is the variance formula conditioned on 1 or 2 items - particularly when there's 3 different items to choose from ?