Re: Freight car Distribution (off list)


Tim: I have, with a lot of help from Allen Rueter, Matt Herson, and others, built a NP database for 1969 that's whopping big now (and growing)---13419 freight car records, ca. 150 trains spread throughout 1969. Obviously it's post-steam era but certainly can help inform G-N and alternative hypotheses.

There are some obviously skewed deviations from G-N predictions in NP traffic flow, but they're all explainable based on recognized exceptions or geography. In a nutshell I'd say from the NP's point of view, there is the CBQ (way overrepresented), the GN (way underrepresented), and all others (conform more or less roughly to G-N proportions). Some other car types, in particular reefers, are very severely skewed away from national fleet proportions. The upshot is that G-N strictly pertains to such a small subset of the overall freight car fleet that Like Al Brown, I question if it's worth all this energetic argument.

I wonder why Mike A. would use the Monon train records to validate the rarified G-N hypothesis and buy a fleet accordingly, when the has in his hands a direct empirical template for the fleet he could be aiming for--with all the assigned-service, traffic routing and car-pooling agreements and multiple other local particularities built in. I use my data more like Clark Probst does. If a certain car series shows up in the database--especially multiple times--it gets priority for modeling. If a car doesn't show up in the data, it requires a fairly well-reasoned stretch (or photo evidence) to rationalize including it in my model fleet.

Chris Frissell
Polson, MT

--- In, Tim O'Connor <timboconnor@...> wrote:

Jeff Aley wrote

The "proximity" part of the hypothesis may also be affected by business relationships
between the railroads (some of which are friendly, and others downright hostile).
Jeff, I really doubt very much that any such skew to "friendliness" exists.
Certainly car assignments, local traffic requirements, and tendency of freight
cars to travel moderate distances (average box car shipments in the 1950's under
600 miles) alone can account for skewing of the averages without invoking some
"friendliness" factor. At the same time, in cases like B&O vs PRR, whose
territories were practically identical, the percentages may be skewed (e.g.
fewer than expected B&O cars on PRR, fewere PRR on B&O) simply because cars
loaded on those roads -tended- not to be delivered to a parallel competitor.
Of course they could be interchanged, but the skew would be away from it.

Tim O'Connor

Join to automatically receive all group messages.