1. Malcolm has been calculating the length of track available cumulatively on an annual basis from 1832 within a 40km radius of each of our six estates. So far the regressions of rent against track, controlling for wheat and cattle prices for the period 1832-1899, have been excellent. Today I was writing up the mathematical model for the paper and I am not sure that just length of track is enough. How about if there was a very long track but only one station? Of course that didn't happen, but we need to have a variable that we can prove mathematically. While we'll stick with track lengths for the moment, let's consider other options. We want a measure of connectivity: how easy was it for a farmer to get his cattle on a train? How about number of stations? From this I thought about network theory.....and nodes. I found there is a useful add-in to Excel called NodeXl that will draw a graph of connections and also give us a statistic of how connected a network is. So we could update the network year by year and then use the annual statistics of connectivity as the variables in the regression. We could get names of stations and years when they opened from railway timetables, the famous Bradshaw.
I tried this for Norfolk. Here is the map of modern stations in Norfolk and a bit of the Nodexl output. There is obviously a lot I need to learn about Nodexl but it does have the advantage of having some math theory to back up the output stats.
|Nodexl output of some of the map. I got bored before I put in all the stations|
|Modern rail network|
Holkham Hall is close to Fakenham towards the north of the map. When I was a kid we went on family holidays at Hunstanton on the coast to the north-west. And we went on the railways!
3. I think I've found somewhere equally interested in Victorian railways, found his blog by chance this morning. Beautifully laid out and all the sources meticulously included.