Though I have not attempted to turn an entire ORER into a database, as someone who has attempted to trace the history of particular categories of cars through several decades using both ORERs and other data, I can relate to much of what you described. Have you input that data by hand or found any way to scan it in?
My first efforts with tracking cars in ORERs (attempting to track the CPR's flat car fleet matched to photos for each number series) ended in a quagmire of incomplete data where whole number series disappeared over a couple of years and no clear explanation was evident. Fleet re-numberings in the early years didn't help either. I've spent some time attempting a more sophisticated approach on refrigerator cars, but have only worked through half of the relevant time period yet.
A good friend of mine has suggested that the answer for identifying a group or class of cars is to link a number of the attributes. We've attempted to start something like that for tank cars we can document as having served in our local and approx. modelling era. So original lot numbers, builder identification, built date, dimensions, materials, etc can each be used with car number series as ways of identifying cars. But as you have observed - these signifiers often change as the years go by. I've found it helps to be able to find the first entry for a series of cars, as it can be very difficult to sort out series made up of a number of originally distinct series.
So I continue to find these efforts are extremely time consuming. I'll be very interested to see what you have done when you're ready Dave.
From: Dave Nelson
Sent: Tuesday, August 19, 2008 10:55 PM
Subject: RE: [STMFC] ORER inaccuracy
Rob, I have 99% of the 4/50 ORER in a relational database (what's still
missing are many of the entries in the little boxes at the end)... and at
least half of another ORER from 1948. The second one was added precisely
for the purpose you have in mind.
That said, there is are a few "little" issues here that I've never been able
to bring myself to resolve: First, how to tell it's the same car from year
to year. Obviously if the car series is the same and the dimensions are the
same, they're the same car recorded in two different years. Where it gets
hard is when there is a renumbering. With that it's no longer possible to
link identical cars without having an overall data concept which, for better
or worse, one could call "Car Class". Some railroads did maintain such
information. Sort of. Most didn't. And of course one could also aggregate
"Car Class" into something that cuts across railroads, such as "Design". So
far, bad, but not too bad. Where it gets really hard is the concept of
version management. When a car class has an interior lining change and the
dimensions are now slightly different... Is it the same "Car Class" or not?
And if it is the same, is there a need to record Rev A and Rev B? How much
change do you accept before you say it's no longer a Rev but a new Class?
Worst of all is doing the first pass thru all the data to make the intial
assignment of a car class.
What I have now is an ability to link by car series, w/o regard to minor
changes in the actual car series value. And I also have a first pass at car
class based on identical values of certain key dimesions.
Here's what remains open (and why I haven't pushed forward): How does car
class relate to Acquisition (e.g., the New Date)? Builder? Clearly one Car
Class can have multiple Builders and multiple Acquisition dates. And that
information is usually tied to an Original Car Series value... Which may or
may not be the value I've got in my database. And that any one current Car
Series may actually be an aggregation of multiple Original Car Series.
So when it's all laid out on paper, it looks (sort of) like this:
1) One Car Design can have one to many Car Classes, spread across multiple
2) One Car Class for each railroad can have one to many Original Car Series.
3) One Original Car Series can have zero to many Sub Series, based on
Acquisition Dates, probably one per Builder, but perhaps more.
4) One Original Car series can have one or more ORER-Date Car Series,
ordered by date.
5) One ORER-Date Car Series can have one to many lines on the page.
6) One ORER-Date Car Series lines can have zero to many other ORER-Date Car
Series lines (to record break outs and consolidations).
And what I have is ORER data by ORER-Date Car Series lines. That's line 6.
What I'm not keen on doing is the work to populate lines 1 thru 5.
If you want to discuss it further, I'm open to talk about it.
P.S. As far as inaccuracy goes, there are errors in the ORER but not too
many. Things like one dimension being greater than another when you know
that's just not possibly correct (e.g., Interior > Exterior). Having it all
in a database made queries for such oddities rather easy.
From: STMFC@... [mailto:STMFC@...] On Behalf Of Rob
Given the assertion (and I doubt any will argue it) that the ORER lists
would include equipment that was not really available for service - whether
retired or in storage or .... etc - is there a realistic research approach
one could adopt to try to identify what parts of a given railway's fleet
were no longer in service?
I wonder if anyone has spent any time developing a useful approach for
sorting out this kind of question? I'd like to hear what approaches you
taken when sorting out the history of a given group of freight cars....