I have analyzed the box car data in the 1948–1952 conductor’s book I mentioned in my previous post on this topic (analyzing refrigerator cars, at http://modelingthesp.blogspot.com/2011/03/modeling-freight-traffic-coast-line.html). The box car data only extend to April of 1951 and are a mix of loads and empties.
The sample contains 139 cars. Of these, 40 are SP cars (and 10 are T&NO cars), making the home road (SP) 29 percent of the total. If the T&NO cars are combined (and certainly SP and T&NO freely shared box cars), the SP system total of 50 cars amounts to 36 percent.
This provides one answer to a persistent question modelers ask: what proportion of home road cars should I model? I noted in an earlier post touching on this topic (available at http://modelingthesp.blogspot.com/2010/12/choosing-model-car-fleet-2.html) that I expected, on the basis of photographic evidence, to have about a third of my cars should be home road cars. That is in fact quite close to the 36 percent just stated, but only with the inclusion of T&NO cars. For SP cars alone, it is below 30 percent.
The remainder of the box cars in this sample are dominated by a few roads. These include PRR (11 cars), GN (8 cars), Southern (7 cars), and NP (6 cars). Following those four railroads are IC and NYC (5 cars each), and SLSF (4 cars). A surprising note is struck by CP with 3 cars and CN with one, more than might have been expected; and at least two of the CP cars were being loaded at Salinas for delivery elsewhere in the U.S.
Minor parts of this sample were ATSF, B&O, CB&Q, MP, and Wabash, with 3 cars each; and C&NW, MILW, SAL, and UP, with 2 cars each. Finally, single cars were listed for C&EI, D&RGW, EJ&E, Erie, GM&O, GTW, MKT, NC&StL, L&N, P&LE, RI, SP&S, SSW, T&P, WM, and WP.
Are there surprises here? A few. Two roads with substantial box car fleets, ACL and C&O, do not appear. It is a little surprising that NYC cars are less than half of the PRR total, though the two roads had similar-size fleets of box cars, and roads with large box car fleets, such as B&O and Milwaukee, are lightly represented.
What about the well-known Gilbert-Nelson hypothesis? I would hesitate to draw much of a conclusion on the basis of only 139 cars, but certainly the biggest box car fleets all show up with significant contributions, even if not necessarily in rank order with fleet size. As just mentioned, there are some surprises. In percentage terms, PRR and NYC each owned about 9 percent of the national box cars in 1953, and indeed PRR cars are 8 percent of this Coast Line sample; but NYC cars are less than 4 percent. The B&O fleet was about 4 percent of the national total in 1953, and here we observe about 2 percent B&O cars. Those same relative values, 4 vs. 2 percent, apply also to Milwaukee Road.
There are also some roads which are, in Gilbert-Nelson terms, over-represented. These include GN, about 3 percent of the national fleet but nearly 6 percent of this sample; and SLSF, with 1.8 percent of the national box cars but almost 3 percent of this sample. But again, with a small sample, distortions are inevitable. One car more or less from some of these roads would greatly alter their standing.
Modeling lessons here, as with many things, are in the eye of the beholder. For me, the main observation from this sample is the proportion of home-road box cars, around a third. That’s something I can and will use in my fleet planning and operation.