In the paper two methods of representing temporal dependencies expressed in natural languages are given. The first one, introduced by H. Reichenbach in 1948, is based on a linear representation of time, with events represented by points on the time scale. The second one is based on a net representation of states, events, and their succession introduced by C. A.Petri in 1962. The main difference between these two approaches consists in accepting by nets (i) partial ordering of events and states rather than their linear ordering, hence accepting their mutual independence, and (ii) a possibility of representing coexisting as well as mutually excluding states, hence accepting different histories in one model. Reichenbach's representation can be viewed as a particular case of the Petri net one. Both representations use graphical means for modeling temporal phenomena.