Friday, March 4, 2011

Modeling freight traffic: Coast Line, 1953-Part 6

The conductor train book which I discussed in the Part 5 post of this thread holds many more potential insights, beyond the ones already cited. In my previous post using these data (here is the link:, I noted that the surprising finding that only about three-fourths of the refrigerator cars were PFE cars, but that this reflects conditions at peak harvest time. Most of the year PFE was able to cover nearly 100 percent of its car needs with its own fleet.
     I was interested to see whether a factual approach to car fleet modeling was reasonable, using the fleet data from the Official Railway Equipment Register (or ORER) as a pattern. In a previous post entitled “Choosing a model car fleet” (linked at I essentially proposed proportioning my fleet of model PFE cars for the Coast Line, according to the proportions found for the entire PFE fleet in 1953 in the ORER. These conductor’s data for one part of the Coast Line can essentially test the reasonableness of such an approach.
     In other words, I asked the question, do the frequencies of the PFE car classes in this conductor’s train book match the frequencies actually present in the fleet at that time? Since there are over 800 cars present, it seemed possible that this test would show something interesting.
     The results I found are shown in the table below. I have used 1951 as a compromise date, since the conductor's data span 1948 to 1952. In the column marked “Actual,” I list the percentage of the total fleet represented by each car number group. In the column marked “Observed,” I list the percentage of that number group in the sample in the conductor’s train book.

Note that both actual and observed percentages nearly add to 100 percent.
     To me, the most striking thing in these results is that the observed percentages do track the actual car fleet percentages pretty well. There are some discrepancies which make sense in terms of the time period. During 1948 to 1952, the R-40-26 class was arriving (delivered during 1950-1951), so is underrepresented in early years of the data; the WP cars and the R-30-12 and -13 cars were rapidly being scrapped or rebuilt and so are underrepresented in the later years. But overall, this fleet-based approach does represent the data from this conductor’s work.
     I also plan to examine the box car data in the sample to see if it tracks the Gilbert-Nelson proposal, though the sample is far smaller than these PFE reefers. There are a few other interesting details, such as the evident change from Blackburn beet racks in the early sugar beet shipping, to composite drop-bottom gondolas in the later years, and I will touch on those results also.
Tony Thompson


  1. This is impressive in that only 6 classes of cars varied more than 1% from the predicted findings. I think you have really hit on a good tool. It will be interesting to see how the other classes of cars match.
    Thanks for mentioning the Blackburn Sugar Beet Racks, I am still working with Doug Junda of Prowest Scale Models here in Colorado to produce a resin model of the Blackburns.

    Tom VanWormer
    Monument CO

  2. Sort the list by expected percentage, descending, and see if the observed percentages fall into the same order.

  3. Thanks, Tom. I have a scratchbuilt Blackburn rack car, a model built by Pat Bray, and I will include a photo of it when I talk about my roster plans for gondolas (including sugar beet traffic, of course).

    Genma, this is a good suggestion. When this is done, the percentages do largely fall in synchrony with the expected data, absent the time discrepancies I mentioned in my post, such as the R-40-26 deliveries. I continue to marvel at how closely several of the classes do match.
    Tony Thompson