[tromp]

Yes, that's the more formal equivalent formulation. For n > 1/(y-x), the sequence of rationals ... ,i/n, (i+1)/n, ... must land in (x,y).