I can't remember the source, but I believe the central return rail was adopted to avoid the need for impedance bonds (common on other systems) to allow the negative return to continue through a succession of signal blocks while containing the signal circuit within the block; thus the signal circuit and the current-return circuit are isolated from each other. From what I've observed, in territory where both commuter rail and the Underground share the same track, a central return rail is installed, since that's what the Underground trains are designed for, while impedance bonds are in place to permit "railroad" trains to use one of the running rails for current return, which is what they are equipped for. I have to assume that on most systems it is felt that impedance bonds are cheaper than a fourth rail. Another factor may be that if the Metropolitan and the Metropolitan District Railways, both originally steam-powered, were still using Manual Block when they were electrified, there would have been no signal circuits to consider and when Automatic Block was installed later, putting in a fourth rail was easier than re-doing the electrification--but I don't know. FWIW.