Why meter-wide dikes at oceanic spreading centers?

Numerical models show that maximum dike width at oceanic spreading centers should scale with axial lithospheric thickness if the pre-diking horizontal stress is close to the Andersonian normal faulting stress and the stress is fully released in one dike intrusion. Dikes at slow-spreading ridges could be over 5 m wide and maximum dike width should decrease with increasing plate spreading rate. However, data from ophiolites and tectonic windows into recently active spreading ridges show that mean dike width ranges from 0.5 m to 1.5 m, and does not clearly correlate with plate spreading rate. Dike width is reduced if either the pre-diking horizontal stress difference is lower than the faulting stress or the stress is not fully released by a dike. Partial stress release during a dike intrusion is the more plausible explanation, and is also consistent with the fact that dikes intrude in episodes at Iceland and Afar. Partial stress release can result from limited magma supply when a crustal magma chamber acts as a closed source during dike intrusions. Limited magma supply sets the upper limit on the width of dikes, and multiple dike intrusions in an episode may be required to fully release the axial lithospheric tectonic stress. The observation of dikes that are wider than a few meters (such as the recent event in Afar) indicates that large tectonic stress and large magma supply sometimes exist.