Although much experimentation centered on the possibility of using the choice between a C1 or a C2 glider to represent a bit, the way the Cyclic Tag System has been formulated is to use C2 gliders exclusively, the relative spacing between them playing the role of a bit. Actually, it turns out even more complicated than that; it is the relative spacing between two pairs of C2's, the spacing within pairs being constant.
C2 gliders have both an active and a passive role. The passive form must allow all EBar gliders free passage as they wend their way west. Since the encounter requires contact between the jog in the EBar and the T6 of the C2 (with a marginal displacenment of 2), the spacing of the C2's must conform to this relationship, and can only vary by including one of more extra EBar unit cells.
Figure 6 shows the four C2-EBar collisions. Besides the soliton, there is another near-soliton, in which a C1 is left behind after the encounter rather than a C2. Additionally, the EBar converts into an F. Had the C1 been more important, this might have been a mechanism for flipping between the two pairs.
The third collision in the figure is typical of C-E collisions, in which a threefold of B gliders is released, but with a spacing typical of such encounters. Its effect on additional C's has been investigated in detail, but it takes no further part in the present discussion.
The remaining collision, complicated though it is, actually constitutes the mechanism by which predicates are implemented in the Cyclic Tag System. The BBar5, on striking the next C2, converts into an EBar of a type which is ignored by the western shower, but which gets it out of the way. The three B's which are released in accompaniment with the BBar play a transient role in this transformation, and so disappear from sight.
It is the surviving A which reacts back with some earlier residue and the forerunner of an EBar packet, in a series of steps, to alternatively generate the D1 - A trimer leapfrog which erases arriving EBar's, or sets up a tricky filter which allows a third of them to pass. They can't be allowed to pass directly because of the parity rules, but it is acceptable to delete a pair of them while passing a third.