Intervehicle communication enables vehicles to exchange messages within a limited broadcast range and thus self-organize into dynamical and geographically embedded wireless ad hoc networks. We study the longitudinal hopping mode in which messages are transported using equipped vehicles driving in the same direction as a relay. Given a finite communication range, we investigate the conditions where messages can percolate through the network, i.e., a linked chain of relay vehicles exists between the sender and receiver. We simulate message propagation in different traffic scenarios and for different fractions of equipped vehicles. Simulations are done with both, modeled and empirical traffic data. These results are used to test the limits of applicability of an analytical model assuming a Poissonian distance distribution between the relays. We found a good agreement for homogeneous traffic scenarios and sufficiently low percentages of equipped vehicles. For higher percentages, the observed connectivity was higher than that of the model while in stop-and-go traffic situations it was lower. We explain these results in terms of correlations of the distances between the relay vehicles. Finally, we introduce variable transmission ranges and found that this additional stochastic component generally increased connectivity compared to a deterministic transmission with the same mean.