We clarify some aspects related to critical phenomena in 1D coupled map
lattices. In particular, we explain why a model studied by Chat&eacute and
Manneville did not seem to be in the same universality class as directed
percolation, contrary to first expectations. We claim that this model is
actually in this universality class, but that scaling is extremely late. We
present a slightly modified model which seems to be in this universality class
with respect to <em>some</em> properties but not all. In contrast to directed
percolation and to the model studied by Chat&eacute and
Manneville, our model shows a very complex sequence of singularities vaguely
reminiscent of the richness of bifurcations in 1D maps.