Technically, yes. Practically... Well, let us say there are 20,000 genes in your typical phrag (Could be 50,000...). In order to get your grandparent species back, you would need to segregate all 20,000 back together. Probability is not my strong point (somebody else do the math), but I suspect that is mind bogglingly small chance, on the order of once in all the atoms in the universe.
Now, it isn't quite that bad, many of those 20,000 genes would segregate in sets (large or small), so that increases your probability a great deal, to perhaps 'no chance in hell' level. Of course we have recombination to think of, so that decreases the probability again... Regardless, it isn't really a very likely scenario.
Now, if we wanted to incorporate a trait from besseae into schlimii, say its red color, we could make the cross, and then back cross to schlimii the best Hanne Popow (the red one that looks most like schlimii). If we did that for several generations, we might get a red flowered plant with all the characteristics of schlimii.