Since the impact factor indicates the citations of a respected journal, then with increasing number of journals shouldn't the citations be distributed somehow.. resulting in a decrease of impact factors over time!
Also, should the total number of journals be a factor in adjustment?