An aversion to reversions
Posted by Thomas Young in category

at 22:15 on 29/01/2023

---

I quite enjoyed the challenges of adapting the vagaries of rolling stock numbering and classification to fit the strictures of a database structure, and I think by and large it has worked quite well. However, there are a couple of aspects that I am not completely happy with, mainly related to the three Rs of Renumberings, Re-uses and Reversions. The latter are particularly problematic, which is what I am writing about today.

A reversion is when an item of rolling stock is renumbered back to a number it carried earlier in its career. Fortunately, reversions are a relatively rare event - the LTSV-RD database currently lists just 540 instances, which represents about 1 for every 500 numbers. Quite early on I decided that a reverted number would have to be treated as a distinct entity from the original number. This was to enable the renumbering table to work out chains of changes without getting stuck in a loop, and to allow the on and off dates for each instance of each number to be recorded (where known). I also decided that reversions after an item's mainline career (ie in preservation) would be ignored. This was quite easy to implement. Taking loco 37165 as an example, the base record for this number had an ID (unique identifier) of LT-37165, and displayed as '37165'. In 1988 37165 was renumbered to 37374, but it reverted to 37165 early the following year. Thus a second record with an ID of LT-37165_R was added, this displaying as '37165 (R)'. 37165 was a rather extreme example because it was then renumbered back to 37374 before reverting to 37165 for a second time (all of which took place in 1989). So the database has a third record added, with an ID of LT-37165_R2 and a display value of '37165 (R2)'.

All well and good, but there were some undesirable implications. Firstly, if you were adding a sighting or a listing of 37165, it would be easy not to notice these renumberings and to 'link' the sighting to the wrong instance of the number. For example, the number 37165 was 'live' from 1974 until 1988, but it might then show a sighting in 2019. To try and reduce the chance of this happening, I have recently been amending the display values for numbers that were reverted (or re-used). The first instance of 37165 will now display as '37165 (1)', the idea being that if (when adding a sighting) you see that a number has a (1) after it, you will know that the number also has later records. You could then check that you have selected the right one.

The second issue is harder to solve, and it is to do with the fragmentation of information. On LTSV-RD, if you look at the detail page for 37165, it will only show items relating to the first incarnation of that number, likewise for the second and third instance. This may seem OK for data like sightings and listings, but what about photos? If I add a photo of 37165 taken in 1992, it will show on the page for '37165 (R2)' but not the other two. As a partial work-around, I have added a 'Similar numbers' section, which will contain links to the other instances (plus to detail pages for any 37165s that occur in other number series), but I think there must be a more elegant solution.

Heritage coaches are the worst culprits for reversions, and they present a few issues of their own, mostly stemming from two events that happened back in the 1980s. Firstly, hundreds of coaches were given new numbers to avoid possible confusion with locomotive numbers. I'll come back to this in a minute. Secondly, it was decided that privately-owned coaches that were allowed to run on the mainline would be allocated new numbers in the private-owner series. The latter was principally a paper exercise. For example, Mk1 coach 3132 was now officially PULR99191, this number being carried on a small yellow plate on the solebars. In most cases, the bodyside numbers remained as they were. A decade later the railways were privatised, meaning that all coaches were now privately-owned and the previous distinction became irrelevant. The majority of heritage coaches were then simply returned to their previous mainline numbers (which they had probably carried all along). So, would it be possible to simply ignore the PO numberings? Not really, for a couple of reasons. Firstly, some of the coaches had adopted their PO numbers as their bodyside numbers and indeed a lot of the West Coast Railways fleet still displays numbers in the 99xxx series. Secondly, I like to record the official information where possible (adding notes where required to highlight any differences from the reality) and (for example) 3132 was 'officially' renumbered PULR99191 for a time. Which means that I have added '3132 (R)' as a distinct record from '3132', with the drawbacks as mentioned above.

Going back to the mass renumberings of the 1980s, many coaches were later preserved, and some were restored to their original (or earlier) identities. Any such cases that were then mainline registered will have been added to LTSV-RD as a reverted number. However, following my own policy, examples that are confined to heritage lines are not added. So, you might see 14007 on the North Yorkshire Moors Railway, and you will be able to log a sighting against '14007 (R)'. However, if you see '14023' on the Mid-Hants Railway, you will only find one entry for this number, and it will show a status of 'renumbered'. I had proposed that any sightings of preserved stock should be logged against the final mainline number (17023 in this case), but I now think that is asking too much. There are many cases where coaches had several later numbers (including in departmental and internal user series) and I can't blame users for 'saying what they see' rather than following through chains of renumberings. And whichever number the sighting is logged against, the data is again being fragmented.

Sometimes I find that explaining a problem gives new insight into potential solutions, which is partly why I have rambled on so much here. Or perhaps somebody out there will have some bright ideas.

Actually, something has just occurred to me! Assuming that the data in the renumberings table is accurate, the database should be able to gather information about related numbers. It already does this to work out the renumbering chains anyway. Perhaps I could make it so that, if you were looking at the detail page for coach number 3132, the photos, sightings and listings sub-sections would include any records from later (or earlier) numbers. It wouldn't really matter that the detail page for PULR99191 and 3132 (R) would show the same records. Hmm, that might work. I'll give it a go, and I'm glad we've had this chat!