I've been working through compiling more data from the 1% Waybill Analyses from my era. I find this sort of thing quite interesting. I've posted links in the right column for several of them hosted at Hathitrust. If you are using the mobile version of the site, click 'view web version' at the bottom of the page to see that column.
I think it's important to understand what this data is, and what it isn't.
First off, there are a lot of books/documents that were generated from this data. The links I provided include the state to state traffic of carloads for each commodity class. These are the classes from the AAR Freight Commodity Classification book I mentioned in a prior post.
There are territorial studies along with others, but my interest is in the origin of traffic to CT, and the destination of traffic from CT.
- The data is for carload traffic only. It doesn't include L.C.L traffic. I do not know if this also excludes cars loaded for multiple destinations, but I suspect it does (I'll get to that in a moment).
- It's 1% of the waybills for a given period. Class I railroads provided data for any waybill numbered "1" or ending in "01" starting in late 1946. They also weighted the numbers to account for differences in waybill practices among roads. So the data has been massaged.
- It also indicates that they did not report data from states where there are 3 or fewer industries shipping a given commodity. These carloads are included in the totals for the major categories. I have noticed these discrepancies when comparing the totals, and have opted to ignore these extra carloads because I don't have any way of knowing in which subcategory they belong.
- They state the margin of error varies depending on the number of loads. For example, if there are 100 carloads reported, there's an expectation of about 10% deviation, but with 400 loads the data is more accurate, with an expected deviation of 5%.
To put it a different way, the fewer loads of a commodity, the less accurate the data is. The number of carloads reported in the CT data is frequently less than 5. Since this is 1%, it represents less than 500 carloads. However, it's quite possible that an unusual shipment was made, and that happened to be the one waybill recorded. That one load would now be considered 100 loads.
Less frequent loads may have been missed entirely as well. Despite these potential issues, the data itself is quite useful.
To give you an idea of the sort of thing that I think you can derive from the data, here's an interesting example.
This is the entire list of car movements for 739 Luggage and handbags, NOS (not otherwise specified)
- Samsonite, a major manufacturer of luggage, doesn't seem to originate any rail traffic from Denver or PA.
- For the luggage/handbag commodity it appears CT and VA dominate the market. But all I can find are a couple of dozen small handbag companies in CT, and in both cases they are averaging around 2 cars/week for the entire group of commodities, not just a single company.
- Hardware shipments from CT are lower than expected, and to fewer destinations than expected.
- These commodities ship in less-than-carload more than I expected.
- The carload shipments are going to distribution centers or major wholesalers, who then ship primarily in less-than-carload lots.
- Warsaw, NY (via O&W NYC Can LV)
- Charlton Depot, MA (NYC (B))
- Batavia, NY (NYC)
- Churchville, NY (NYC)
- Gasport, NY (NYC)
- Gasport again
- Medina, NY (NYC)
- Huntington, NY (LI)
- Waterville, ME (Sid D&H Mcv B&M Portland MEC)
- Dover, NH (Sid D&H Mcv B&M)
- Barre, VT (Sid D&H RPT CVT)
- St Albans, VT (Sid D&H RPT CVT)
- Bellows Falls, VT (Sid D&H Mcv B&M)
- South Vernon, MA (Sid D&H Mcv B&M)
- Owego, NY (Sid D&H Bing ERIE)
- Lebanon, NH (Sid D&H Mcv B&M)