Date 21 - 23 of 23
Lumber Loads on Flat Cars and in Box Cars
Aley, Jeff A
Very interesting analysis. Can you tell me more about Mr. Ed Ullman's analyses? I have copies of some of the 1% waybill data (copied from The Stanford U. Libraries). I'm interested to know what Mr. Ullman has done with it.
From: STMFC@... [mailto:STMFC@...] On Behalf Of Charles Hostetler
Sent: Saturday, July 14, 2012 6:33 PM
Subject: [STMFC] Re: Lumber Loads on Flat Cars and in Box Cars
--- In STMFC@...<mailto:STMFC%40yahoogroups.com>, "np328" <jcdworkingonthenp@...<mailto:jcdworkingonthenp@...>> wrote:
I was following this thread with mild interest until I saw Jim's comment. Then I got curious as to how much rail traffic equated to "-so much-". Fortunately I had finally received a copy of Ed Ullman's work (American Commodity Flow) that provided an approach to the answer based on his analysis of the ICC's 1 percent waybill sample (1948 through 1950). Those interested can find a state by state breakdown of the destinations of lumber shipments by rail from Oregon, Washington, Louisiana, and Mississippi at:
Now instead of thinking this might be a mildly interesting sideshow I'm thinking its an interesting part of the main story line. Thanks!
Charles Hostetler <cesicjh@...>
--- In STMFC@..., "Aley, Jeff A" <Jeff.A.Aley@...> wrote:
I haven't been able to obtain any of the raw data yet, so some of this is guesswork about his (Dr. Ullman's) analysis. He was primarily an economic geographer, and his speciality was cartography. From the way he talks about data in his book I'm guessing he also had a bit of training in statistics. Here's what I think he did:
1) Categorize the commodities that were shipped into major groupings corresponding roughly to what we would today call commodity codes.
2) Categorize the origin and destination of the shipments by state.
or it could have been that the original data were presented this way and he just took advantage of that format...
3) Multiply the 1% sample by 100.
4) He used some kind of "fill in" rules for commodity flows that were suppressed by the ICCs "disclosure rule".
5) He compared data from several different years to see if there were trends that were distorting his "snapshots" of the data.
He compared his work against at least two other methodologies to see if there was a systematic bias.
But he was primarily a cartographer and like to make maps.
6) So he took all of the data, divided the total number of tons by 10,000 (because that was a convenient number) and plotted a series of maps with one dot in each state for every 10,000 tons of whatever commodity was shipped to that state.
There are over 100 of these maps in the book, and each one was hand-drawn. So I imagine several of his grad students were significant contributors ;)
7) As far as data analysis, the thesis of his book was that (in the 1950s) there was a "Core Area of U.S. and Canada", that the outlying areas (the "hinterlands") traded with the core area, and that the intensity of trade was governed by three principles which he proposed - Complementarity, Intervening Opportunity, and Transferability.
8) He was also a mild critic of the use of the "gravity model" to describe commodity flows, and tried to point out examples where his three principles fit the data better than the gravity model.
Sorry for the length of this, and it's certainly in the weeds and might be verging out of scope. I don't want to have to use my get out of jail free card yet... If you want to see any specific parts of the text or some of the maps drop me a note off-list.
Charles, a big thanks back to you. I was looking at car request numbers by Oregon shippers however once the cars were loaded I was somewhat in the dark as to where from there. I did come across some records of cars waiting to use the load shifter at Laurel, MT yard of the NP and will get this info transcribed after our convention however again, it is nice to know the probable routing of the loads.
Thank you for the recent information you supplied. Jim Dick
|21 - 23 of 23|