Todd Sullivan wrote "A question: How did you determine the number of interchanges for each load? I understand the numbers of originating ("Orig") and terminating ("Term") loads, but I didn't think you included data on the number of times a load was interchanged (which, I assume, means an inter-railroad movement)."
The answer is that I inferred it from more detailed information inside the report. Each year's ICC Freight Commodity Statistics Report consisted of a set of master sheets, with detail on each Class 1 carrier. Each carrier reports (on an annual basis) the following (in tons):
a) traffic originated on line and terminated on line
b) traffic originated on line and delivered to another carrier
c) traffic received from another carrier and terminated on line
d) traffic received from another carrier and delivered to another (third) carrier
So the sum of a and b are the total originated by the reporting railroad and the sum of a and c are the total terminated by the reporting railroad. Here's an example of raw data I had already transcribed from 1957 for bananas, fresh. (For brevity this table shows only a few of the 110 Class 1 Carriers; had to transcribe the total list to get the grand total in the bottom line):
The total tons (last column) are obtained by sum of the first four columns and the total row at the bottom is obtained by the sum over all of the carriers (including those not shown in this brief snippet). The fourth numeric column, tons received from connecting roads is actually what we are after, the total number of tons of pure interchange (or overhead) traffic. It was my hypothesis that the number of tons in this column would be greater for commodities that would be diverted more frequently.
The trouble is that there are 261 commodity classes and each commodity class has 4 raw numbers to transcribe for each of 110 Class 1 carriers. So I've been plugging away at transcribing and QAing this for a while now and it's not done yet for all commodities for any single year. And when I was looking at Tony's post I wanted to see if perishables as a group had more interchanges compared to other commodity classes. So I took a shortcut and transcribed one of the summary tables after I had convinced myself that there was a correlation between the total number of tons (or carloads) handled and the total number of tons (or carloads) of bridge traffic. It was kind of a kludgy demonstration, but that's how the ratio was computed.
I figure I'll complete all of the transcriptions in a year or so and then I can look at the detailed data more analytically, but anyway it was a start. Eventually I hope to use this type of information to pattern my bridge traffic (I have a lot of overhead traffic to deal with relative to locally originated or terminated traffic). Hope this wasn't too confusing and if you want to take a look at a couple of spreadsheets to see some of the details please drop me a line off list and we can discuss in more detail.