To the left is a left bundle branch block (LBBB) morphology at cycle length ~ 800 ms with a PRI ~ 160 ms. Same thing is to the right with normal QRS duration.
It can be said that the LBBB in this case is not a fixed phenomenon. There is normalization after a PVC (coming from the LV - RBBB morphology).
Persistence of BBB at lower cycle length and termination by a PVC
The persistence of BBB at lower cycle length (here ~ 800 ms) can be due concealed transseptal conduction. Concealed simply means you cannot see it in surface ECG but can be recognized from its effect on the subsequent impulse. The LBB here is not fixed so something is making the LBBB.
In this tracing we may not know the initiating event of the LBBB but it continued. Its perpetuation may be due to impulse coming from the right bundle that penetrated (transeptal-retrograde) and collided with the incoming (antegrade) impulse of the left bundle.
The abrupt recovery was due to the PVC that excited the left bundle early and allowed more time for it to recover ("peel back refractoriness").
Fisch C., Zipes DP and McHenry PL. 1973. Rate Deependent Aberancy. Circ 48:714-724 (http://circ.ahajournals.org/content/48/4/714.ful l.pdf+html)
Fisch C and Knoebel SB. 2000. Electrocardiography of Clinical Arrhythmia. New York. Futura Publishing Co.
Neiger JS and Trohman RG. Differential Diagnosis of Tachycardia with a Typical LBBB Morphology. WJC 3(5):127-134