Recent deep-tow, Sea Beam, and SeaMARC II studies indicate that ridge axis discontinuities along the East Pacific Rise evolve in several different ways. Where magmatic pulses along the spreading center happen to meet head-on, a low point or “saddle point” occurs along the axial depth profile of the spreading center. Where the magmatic pulses misalign, overlapping spreading centers develop and two distinct evolutionary paths are possible. One spreading center tip may cut through to and link with the adjacent en echelon ridge, chopping off the opposing ridge tip, as proposed by Macdonald and Fox (1983). In addition, either spreading center may cut inside or outside of itself, repeatedly decapitating its own ridge tip. Crack-propagation studies show that the crack-propagation force, G, drops significantly when the ratio of crack overlap to crack offset (L/W) exceeds 3. Applying this relation to spreading centers, we suggest that propagation of the individual spreading center tips may stall when L/W is greater than 3. If linkage has not yet occurred, then the next magmatic pulse that propagates along the ridge may be deflected away from the path of the existing ridge tip, decapitating the ridge tip in a process we call self-decapitation. Subsequent magmatic pulses may be deflected or “derailed” from the existing path of the ridge because freezing of the axial magma chamber near ridge-axis discontinuities creates a core of coherent, unfaulted gabbroic rock along the spreading axis which is strong relative to the intensely faulted lithosphere on either side of the frozen magma chamber. In addition, the local stress field rotates near the discontinuity so that subsequent magma pulses and associated cracking fronts will tend to deflect away from the preexisting path of the ridge. For propagating rifts, which represent first-order changes in plate-boundary geometry, L/W remains <1.5 so that G is maintained near its maximum value and episodes of continued propagation in the same direction are enhanced.