An experiment in optimal transport original data source is:

https://commons.wikimedia.org/wiki/London_Underground_geographic_maps/CSV

I cut and pasted this source to produce the three csv files, and then
London.R was used to produce London.Rda which has a somewhat cleaned up
and expanded version of the data.  The main addition is the column of
distances for the links.  The file tube.R  contains a function to
compute pathways from origin to destinations.  There are several issues

    o  some links are used by more than one line so it appears that 
	there are more line transfers than necessary.
    o  it would be nice if there were similar data sets for other metro systems
    o  if there were better line info for links one could penalize line changes



