Übrigens habe ich bemerkt, dass du diese Daten von OSM kopierst.
Das Problem an diesen Daten ist, dass diese unter der ODbL lizensiert sind und dementsprechend eigentlich so nicht in OHM (dessen Daten CC0 sind) importiert werden sollte.

Claiming that "OHM is a single hack" is still not a good excuse to falsify data.

After looking at both relations, I've realised that they represent the states in which the tracks go to Wiesbaden (before 1906 when the old station existed and past 1906 where the tracks follow the current relation).

In any case, what I just did was to apply a proper fix, to add a start_date for each relation (and also add a chronology relation because it is a major route which changed over time quite a lot alongside end_date for the older relation because that became invalid).

The problem with your change is that this is a hack (or as we OSMers call it, tagging for the data consumer) and will only lead to later issues thanks to the misrepresented data instead of fixing it straight up.

Think about it: Someone could create an OHM version of ÖPNKarte and users will notice that weird route which spans from Frankfurt to Wiesbaden. Sooner or later, someone will notice that this relation doesn't refer to any PT route but rather a railway relation which happen to be tagged as PT relation and will change it to route=railway (not to mention we're back to square run about an eternal route which didn't exist in the 18th century and earlier).

I've changed it back to route=railway because it isn't a train route: route=train is a PUBLIC TRANSPORTATION route (one which usually is served by a mainline train) but not all railways are served by public transportation.

Was ich meinte ist, dass ich die Gleise der Hessischen Ludwigsbahn in Mainz eingetragen habe.