The Sectional Appendix Online - sectionalappendix.com

[GLC]

Hello folks,
If you've ever found yourself wanting to be a nerd and look up line speeds, but are put off by the 1000+ page PDFs that Network Rail produces, I've set an LLM the task of scanning the PDFs and making a searchable archive of them. You can find the archive here sectionalappendix.com. As it was done by an AI, there may be errors, but the bits I've looked at personally seem pretty good to me.

Hope someone finds this useful

 

[GraphitProjekt]

As much of an AI sceptic as I normally am, this is good work to be fair. I was just thinking today about making a map of Anglia signalling areas and this already makes the process a lot easier as I can simply type in the next LOR into the search. Although, I had someone honed my skills of scrolling just about the right amount through the Table A section lol.

If I might make one suggestion, perhaps consider making it a little more convenient to scroll through LOR sequence pages, you already can but there's a lot of extra stuff for each picture i have to go past to get to the next page, maybe there's some other place that information could go. I respect that's probably specific to my use-case

 

[zwk500]

The parsing of the diagrams is excellent, although I'd agree with the suggestion that the extra boxes between each page are superfluous. I'd also say the search box and indexing can be tightened up, because typing a station name can return lots of pages that don't actually contain that statoin.

It'd also be really nice to include the Local Instructions and Special Working Arrangements for each LOR, and potentially (eventually) the Table Ds. But it's a very handy tool for those of us who don't have NESA access.

Also if links to the connecting diagrams could be added under the LOR tab that would be really nice.

Sorry if this is quite a long laundry list!

 

[gazr]

Nice site... bookmarked.

 

[55002]

Good that

 

[Horizon22]

Decent work. I agree with others that the interlinking page with the 5 boxes underneath each diagram is redundant.

 

[GLC]

If I might make one suggestion, perhaps consider making it a little more convenient to scroll through LOR sequence pages, you already can but there's a lot of extra stuff for each picture i have to go past to get to the next page, maybe there's some other place that information could go. I respect that's probably specific to my use-case

I've added a sidebar in each LOR, which has links to each diagram there without having to scroll the entire page, and I've collapsed a lot of the text info by default. Let me know if that's what you were thinking of.

The parsing of the diagrams is excellent, although I'd agree with the suggestion that the extra boxes between each page are superfluous. I'd also say the search box and indexing can be tightened up, because typing a station name can return lots of pages that don't actually contain that statoin.

Sorry if this is quite a long laundry list!

I've improved search (it was treating spaces in station names as separate search terms so it was matching a bit too eagerly).

It'd also be really nice to include the Local Instructions and Special Working Arrangements for each LOR, and potentially (eventually) the Table Ds. But it's a very handy tool for those of us who don't have NESA access.

That's a nice idea, the OCR is relatively time consuming without spending a huge amount on tokens, so it'll take me time to work through it but I'll add it to the list

Also if links to the connecting diagrams could be added under the LOR tab that would be really nice.

There should be links to each connecting sequence for each page e.g. https://sectionalappendix.com/scotland/SC009/001 links to SCR001 SEQ014. Am I misunderstanding what you are referring to? It's also possible the LLM didn't pick up the connection clearly, so it might be if there's a specific page you're looking at, I need to nudge the bot to try a bit harder

 

[zwk500]

That's a nice idea, the OCR is relatively time consuming without spending a huge amount on tokens, so it'll take me time to work through it but I'll add it to the list

I believe there's a Git project that has parsed those sections which might save you some work!

There should be links to each connecting sequence for each page e.g. https://sectionalappendix.com/scotland/SC009/001 links to SCR001 SEQ014. Am I misunderstanding what you are referring to? It's also possible the LLM didn't pick up the connection clearly, so it might be if there's a specific page you're looking at, I need to nudge the bot to try a bit harder

Ah, I hadn't noticed the connections box before. However, it's not universally successful, as MD105.001 doesn't link to MD101.026 https://sectionalappendix.com/lnw-south/MD105/001, MD105.003 doesn't link to MD175.001 (although that's not really relevant!), and MD105.007 doesn't link to MD101.029

A quick flick through the first half-dozen or so pages of LN600 also doesn't show anything in the connections box despite numerous connecting lines being shown.

 

[Nicholas Lewis]

Was after speeds on routes the other day and rail data doesnt hold those so this is a great addition.

 

[GLC]

I believe there's a Git project that has parsed those sections which might save you some work!

Ah, I hadn't noticed the connections box before. However, it's not universally successful, as MD105.001 doesn't link to MD101.026 https://sectionalappendix.com/lnw-south/MD105/001, MD105.003 doesn't link to MD175.001 (although that's not really relevant!), and MD105.007 doesn't link to MD101.029

A quick flick through the first half-dozen or so pages of LN600 also doesn't show anything in the connections box despite numerous connecting lines being shown.

I've nudged the AI to re-review those pages, and the rest of the scanned records, so hopefully they should be better now!

== Doublepost prevention - post automatically merged: 2 Oct 2026 ==

I’ve now added loading gauge, cleared TOPS classes, and RA ratings to (hopefully) all pages. The way the info is transcribed doesn’t nearly line up with the diagrams so the AI has done some guesswork to match them up, so expect error etc