I've seen studies where they split up morphology by Gardner excellent, good, average, and poor where poor is usually defined as: 1BC, 2BC, 3BC, 4BC, 5BC, 6BC, 1CB, 2CB, 3CB, 4CB, 5CB, 6CB, 1CC, 2CC, 3CC, 4CC, 5CC, 6CC,1BB, and 2BB, and they still have decent success rates. So I think the 'why they freeze these' is because there is proof that sometimes they work, even if the rate is lower than the other gradings! If there's nothing else to transfer, some chance is better than none surely? x