Talk:2024 Talerddig train collision

Unverified details

Please can we NOT add unverified details to the article. I particularly have in mind the numbers of the units involved and the fact that conditions similar to the 2021 Salisbury rail crash may be a factor. If and when these are reported in mainstream media, they can be added. There is WP:NORUSH to add stuff to the article. Mjroots (talk) 07:21, 22 October 2024 (UTC)

Would you not suggest that we could use RTT to verify the numbers of the units, in which case it would be 158841 (https://www.realtimetrains.co.uk/service/gb-nr:L40013/2024-10-21/detailed) and 158824 (https://www.realtimetrains.co.uk/service/gb-nr:L40361/2024-10-21/detailed) Vanmanyo (talk) 15:34, 22 October 2024 (UTC)

RTT only provides details seven days previously, so in a week the reference will fail WP:V. As @Mjroots says, there really is no rush to provide the unit numbers along with a quality reference. We'll get there eventually. 10mmsocket (talk) 16:11, 22 October 2024 (UTC)

@Vanmanyo: if all else fails, the information will be in the RAIB final report. You will notice that the photo chosen is of one of the units involved, ready for the caption to be changed when we have a reliable reference for the fact. Mjroots (talk) 16:17, 22 October 2024 (UTC)

The points you and @10mmsocket make are completely valid I was just wondering why it had not been done. I agree though to wait and ensure we have the correct unit numbers. Vanmanyo (talk) 16:25, 22 October 2024 (UTC)

Digging through RTT (which can be archived) is borderline original research and is not guaranteed to be correct. For unit allocations it merely reflects the data in the relevant system, which might be what was planned to be allocated, what was actually allocated, or a mix of both. The exact units are really not that important for Wikipedia's purposes. Thryduulf (talk) 16:29, 22 October 2024 (UTC)

The two RTT pages have several interesting features - they show the trains as "1J25 1831 Shrewsbury to Talerddig" and "1S71 1909 Machynlleth to Talerddig", as if Talerddig (where there has been no station since 1965) was always going to be the ultimate destination; and they also show "This service was cancelled between Talerddig and ... due to leaves on the line (QH)." That is going to give much fuel to those who ridicule "leaves on the line" as a pathetic excuse. For the train from Machynlleth, why is the headcode 1S71? Surely the S denotes an inter-regional train to the Scottish Region from another Region? --Redrose64 🌹 (talk) 20:55, 22 October 2024 (UTC)

That's how RTT always handles partially cancelled trains, showing the ending as the point in the system (whether station or something else) closest to the place the service actually terminated.

Separately, Spirit of Railways is continually adding the unit numbers based on unreliable sources (and on the most recent occasion also original research). I've reverted them three times, hopefully it won't be required a fourth time but I'm not certain they understand. Thryduulf (talk) 21:10, 22 October 2024 (UTC)

I understand that the RTT sources will become out of date, and respectfully agree that they are not a good source, and won't be using them anymore. However I fail to understand why the usage of an image of the exact same 158 unit 14 years ago is allowed, however a photo of the exact same unit taken 3 weeks ago is not allowed? If it's a matter of the wording I used in the edit I understand, and understand changing it, but if it is entirely based on the image, I remain confused. Spirit of Railways (talk) 21:27, 22 October 2024 (UTC)

@Spirit of Railways: - picture quality is generally the major factor when choosing images. How recent the photo is, or who took/uploaded it to Commons does not come into it. Mjroots (talk) 03:46, 23 October 2024 (UTC)

The RTT links still work as of a couple of minutes ago. So much for the one-week life. --Redrose64 🌹 (talk) 21:00, 11 December 2024 (UTC)

Interesting. Possibly preserved as evidence? However the units have now been confirmed from other sources, so it's irrelevant apart from a bit of history. — Voice of Clam (talk) 21:43, 11 December 2024 (UTC)

The oldest working RTT report in my browser history is this one from 2 July 2024 (23 weeks ago). It's a train which apparently ran to timetable with no delays of more than half a minute, and indeed a couple of early arrivals. --Redrose64 🌹 (talk) 23:02, 11 December 2024 (UTC)

First collision between (passenger?) trains since 1991

The BBC are reporting this as the first collision between trains (or maybe just passenger trains) in Wales since the Severn Tunnel rail accident in 1991.[1] Should this be added to the article? Thryduulf (talk) 19:27, 22 October 2024 (UTC)

I would say why not? Because rail accidents are few and far between. They don't occur every day, week or even month in the UK so I'm surprised to see that we didn't have an accident between trains for over 30 years in Wales (population 3m vs 67m). The last one was in August 2020, just after the Stonehaven one. JuniperChill (talk) 20:17, 22 October 2024 (UTC)

I've added it as a see also for now, someone can work it into the prose if they wish. Thryduulf (talk) 21:15, 22 October 2024 (UTC)

Abermule 1921

I am very much reminded of the Abermule train collision which occurred 103 years ago on the same line, but 18 miles to the east; it also involved a head-on collision between two passenger trains on a single track between passing loops. Shall we put it in a "See also" section? --Redrose64 🌹 (talk) 21:02, 22 October 2024 (UTC)

Yes, that's definitely a good see also. I'll add it. Thryduulf (talk) 21:11, 22 October 2024 (UTC)

I removed the word "similar" from the Abermule blurb. Mismanagement of the token system was not involved here. Added Salisbury, in which low adhesion was a major factor. Mjroots (talk) 04:07, 23 October 2024 (UTC)

It's similar as in it also involved a head-on collision between two passenger trains on a single track between passing loops. I didn't mention the token system. --Redrose64 🌹 (talk) 20:58, 23 October 2024 (UTC)

Describing it as similar (which to be clear the article currently doesn't) would be original research I think. We'd need a source to make that connection. Opolito (talk) 21:26, 23 October 2024 (UTC)

@Redrose64: - therein lies the difficulty. Similarity in type of event, but from completely different causes. Whereas with Salisbury, a different type of collision, but from (apparently, and subject to confirmation) a similar cause. Mjroots (talk) 06:25, 26 October 2024 (UTC)

Image

As the question of which image to use has been raised a few times, either on this talk page or by changing the image in the article, we should discuss this. Image quality should be the main factor. As we know (but can't yet verify) the identities of the two units involved, it is natural that we should use an image of them. For 158 824, Commons has ten images availabe at C:Category:British Rail Class 158 158824. File:158824 at Crewe.jpg is the best image, and is used in the article. File:ATW 158824.jpg was used in the article. I reverted its addition on quality grounds - it is very dark. However, it is a more recent image. As for 158 841, C:Category:British Rail Class 158 158841 only has four images. File:158841 Wrexham General 130115.jpg is possibly the best of them. Once the identities of both units involved are verified, we can use two images in the infobox. Mjroots (talk) 04:27, 23 October 2024 (UTC)

File:158824 at Llandudno Junction Sep 28th 2024.jpg has also been used in the article. Again, fails on quality although it does show the actual livery carried at the time of the accident. This photo wasn't in the correct category at the time of my earlier post. Mjroots (talk) 04:57, 23 October 2024 (UTC)

As the identities of the units have been made public, I've added images of both to the infobox. Mjroots (talk) 06:21, 26 October 2024 (UTC)

Article name

I may be opening a can of worms here that it may be better to discuss in general elsewhere, but is there any convention for naming this sort of article? With this and the three articles in the see also section we have

• [year] [place] collision {edit - this article has now moved to [year] [place] train collision}
• [place] train collision
• [place] rail accident
• [year] [place] rail crash

Should we aim for some consistency, or is there some fine distinction between accidents, crashes and collisions? — Voice of Clam (talk) 10:04, 23 October 2024 (UTC)

@Voice of Clam Salisbury has the year to distinguish it from the 1906 crash. None of the others need disambiguation. Awkward42 (talk) [the alternate account of Thryduulf (talk)] 11:50, 23 October 2024 (UTC) Awkward42 (talk) [the alternate account of Thryduulf (talk)] 11:50, 23 October 2024 (UTC)

As noted at the Talerddig railway station article, there was a previous collision near Talerddig. Year disambiguator is justified. Mjroots (talk) 12:08, 23 October 2024 (UTC)

@Mjroots as that doesn't have an article, I don't think it needs more than a hat note. Awkward42 (talk) [the alternate account of Thryduulf (talk)] 15:07, 23 October 2024 (UTC)

@Thryduulf: it could possibly have an article. The 2023 Voorschoten train crash article was created at that title. The creation of the 1926 Voorschoten train crash followed later. Mjroots (talk) 15:12, 23 October 2024 (UTC)

The 1894 collision has an accident report. Looking at this report, and correlating it with other works, it seems that Talerddig is the summit of the line, with a level stretch 20 chains long, from milepost 61 to milepost 61+1⁄4; to the west it falls at 1 in 56 for 66 chains and then at 1 in 56 for 2 miles 20 chains. The passing loop is partly on the level stretch and partly in the 1 in 56, so if there was low adhesion the train could fail to stop on the falling gradient westbound and overshoot the loop exit. --Redrose64 🌹 (talk) 22:49, 23 October 2024 (UTC)

@Mjroots: I deliberately used the sentence "The downhill gradients onward to Machynlleth are less severe.", because these gradients are not relevant to the 2024 accident. In fact, it's not a consistent 1 in 80 - the slope varies a lot, the steepest being 1 in 60, the easiest non-level gradient being 1 in 628, and there is even an uphill stretch at 1 in 132 approaching Machynlleth. The gradients that I gave details for are in the three miles between Talerddig and Llanbrynmair, within which the accident occurred. --Redrose64 🌹 (talk) 14:44, 14 November 2024 (UTC)

@Redrose64: - I've clarified that the 1 in 80 is immediately after the loop. If you wish to add to that, feel free. Mjroots (talk) 18:47, 14 November 2024 (UTC)

This is worse. As with my original edit, the gradient immediately after Talerddig loop is 1 in 56 for about three-quarters of a mile; then 1 in 52 for about two and a quarter miles; then level for about a quarter mile at Llanbrynmair; then short stretches of 1 in 60; 1 in 142; 1 in 91; 1 in 355; 1 in 119; etc. but no 1 in 80 that I can see anywhere in the fourteen miles between Talerddig and Machynlleth. The accident happened somewhere in the vicinity of the change from 1 in 56 to 1 in 52, so what happened beyound Llanbrynmair is not relevant. However, going the other way, towards Caersws, there is a falling gradient of 1 in 80 for about 70 chains, easing to 1 in 462, then 1 in 907. But this is where 1J25 had already come from, so again is not relevant. --Redrose64 🌹 (talk) 19:16, 14 November 2024 (UTC)

Rail states "Westbound trains climb a 1-in-80 gradient as they approach the loop, which is level, before descending at 1-in-56 the other side." I'll not press the issue for now. If Rail are incorrect then we can ignore them. Expect the gradients will be mentioned in the RAIB interim/final report and will be correct. Mjroots (talk) 06:39, 15 November 2024 (UTC)

Rail is correct (presumably - I haven't read it) in that the 1:80 climb is on the approach to the loop in the westbound direction. However your edit stated it was beyond the loop in the direction of Machynlleth. — Voice of Clam (talk) 07:53, 15 November 2024 (UTC)

There's an on-line gradient profile and details here. — Voice of Clam (talk) 08:00, 15 November 2024 (UTC)

Good find. Clicking on the cyan "Gradients" button I find that the gradients correspond with the sources that I used earlier, but the distances are much more precise. I've extracted the gradient data:

Gradients between Caersws and Machynlleth

Miles	Chains	Yards	km	Gradient Change	Length	Station
53.15	12	0263	85.54	Level	820.77 m	CAERSWS
53.66	53	1161	86.36	+1:211 (+0.47%)	1158.73 m
54.38	30	0668	87.52	+1:71 (+1.40%)	804.67 m	Pontdolgoch
54.88	70	1548	88.32	+1:199 (+0.50%)	917.33 m
55.45	36	0792	89.24	+1:152 (+0.65%)	1142.63 m
56.16	13	0281	90.38	+1:127 (+0.78%)	643.74 m
56.56	45	0985	91.02	+1:100 (+1%)	1335.76 m
57.39	31	0686	92.36	Level	740.30 m
57.85	68	1496	93.10	+1:171 (+0.58%)	949.51 m
58.44	35	0774	94.05	+1:128 (+0.78%)	901.23 m
59.00	0	0000	94.95	Level	434.52 m	Carno
59.27	22	0475	95.39	+1:149 (+0.67%)	514.99 m
59.59	47	1038	95.90	+1:907 (+0.11%)	531.08 m
59.92	74	1619	96.43	+1:462 (+0.21%)	820.77 m
60.43	34	0756	97.25	+1:80 (+1.25%)	1303.57 m
61.24	19	0422	98.56	Level	321.87 m	Talerddig
61.44	35	0774	98.88	-1:56 (-1.78%)	1416.22 m
62.32	26	0563	100.29	-1:52 (-1.92%)	3556.65 m
64.53	42	0932	103.85	Level	659.83 m	Llanbrynmair
64.94	75	1654	104.51	-1:60 (-1.66%)	1223.10 m
65.70	56	1232	105.73	-1:142 (-0.70%)	434.52 m
65.97	78	1707	106.17	-1:91 (-1.09%)	692.02 m
66.40	32	0704	106.86	-1:355 (-0.28%)	1287.48 m
67.02	2	0035	107.86	-1:119 (-0.84%)	450.62 m
67.48	38	0844	108.60	-1:78 (-1.28%)	997.79 m
68.10	8	0175	109.60	-1:113 (-0.88%)	418.43 m	Commins Coch Halt
68.36	29	0633	110.01	-1:163 (-0.61%)	4087.73 m	Cemmes Road
70.09	7	0158	112.80	-1:105 (-0.95%)	-418.43 m
70.64	51	1126	113.68	-1:256 (-0.39%)	1384.04 m
71.05	4	0087	114.34	-1:90 (-1.11%)	675.92 m
71.92	74	1619	115.74	-1:628 (-0.15%)	2108.24 m
73.23	18	0404	117.85	-1:152 (-0.65%)	337.96 m
73.44	35	0774	118.19	-1:275 (-0.36%)	804.67 m
73.94	75	1654	118.99	Level	1464.50 m
74.85	68	1495	120.46	+1:132 (+0.75%)	885.14 m
75.04	3	0070	120.76	Level	-273.59 m	MACHYNLLETH

Positive gradients are ascending; negative are descending. So it's clear that 1J25 had already climbed the 1 in 80 up to Talerddig, had run through the level stretch there and was descending the 1 in 56 that followed when it collided with 1S71. So Rail is correct, and this edit is not. --Redrose64 🌹 (talk) 14:14, 15 November 2024 (UTC)

Q for Redrose64 - the RDT is essentially correct, is it not? Mjroots (talk) 06:43, 16 November 2024 (UTC)

Yes; it's orientated with east at the top, as is the list of gradients that I included above. It lacks a little black triangle ( (lGIP) or similar) to denote the summit, but as the summit wasn't a single point but a level stretch of 16 chains, I don't know the best position for such a triangle. --Redrose64 🌹 (talk) 13:15, 16 November 2024 (UTC)

@Redrose64: I think I've now got the gradients correct. Mjroots (talk) 16:54, 16 November 2024 (UTC)

Comment from a signalling engineer: the gradients as declared by gradient charts may well differ from those on signalling diagrams. Not only do the actual gradients depend on the track tamping but the signalling may well use average gradients as that is what important to braking distance used to determine train protection required, not every tiny up or down — Preceding unsigned comment added by 2A00:23C6:2006:6F01:A530:17EB:D446:4354 (talk) 17:57, 23 November 2024 (UTC)

Thanks, but Wikipedia has a policy of verifiability: is the information published in a reliable third-party source? --Redrose64 🌹 (talk) 18:32, 23 November 2024 (UTC)

The relavant documetation depends on the many Networkrail specifications, and RSSB publications plus codes of practice used locally - many are confidential or access only by licence. To understand them and mine them out over the last 50 years has given me an IRSE Licence and professional status. However I was trying to head off excess pedanantry at the pass as very unhelpful.

The overrun at the block marker or signal depends on average gradient over the braking distance approaching the signal. The length of excess overrun depend on the gradient after the signal. Basic Newtonian mechanics 2A00:23C6:2006:6F01:7DA2:EFDF:303D:D54F (talk) 21:37, 23 November 2024 (UTC)

Irrelevant information?

(Splitting off from above topic as it has nothing to do with the article name) - are we in danger of adding too much irrelevant detail here? The fact that east-bound trains were banked up from Machynlleth in the days of steam serves little purpose other than to indicate it was steep, which has already been stated by the gradient. I agree that we don't need to specify gradients beyond Llanbrynmawr - in fact I would go so far as to say that we only need to give the first gradient of 1:56 falling from the west end of the loop. Thoughts? — Voice of Clam (talk) 22:30, 14 November 2024 (UTC)

Did you know nomination

The following is an archived discussion of the DYK nomination of the article below. Please do not modify this page. Subsequent comments should be made on the appropriate discussion page (such as this nomination's talk page, the article's talk page or Wikipedia talk:Did you know), unless there is consensus to re-open the discussion at this page. No further edits should be made to this page.

The result was: promoted by Crisco 1492 talk 14:35, 12 November 2024 (UTC)

(

• Comment or view
• Article history

)

• ... that the 2024 Talerddig train collision was the first collision in Wales between trains since 1991?

• Source: https://www.bbc.co.uk/news/articles/c4glg9gzrdjo "The last train crash in Wales happened in 1991 inside the Severn Tunnel which allows travel between England and Wales."

• Reviewed: Template:Did you know nominations/Final Fantasy XIV: Dawntrail
• Comment: I thought I was the first one to create this train accident that just happened in Wales, as seen over here

Created by Mjroots (talk). Number of QPQs required: 1. Nominator has 7 past nominations.

JuniperChill (talk) 20:55, 23 October 2024 (UTC).

• General eligibility:
• New enough:
• Long enough:
• Other problems:

Policy: Article is sourced, neutral, and free of copyright problems
Adequate sourcing: Neutral: Free of copyright violations, plagiarism, and close paraphrasing:

Hook eligibility:

• Cited:
• Interesting:
• Other problems:

QPQ: Done.

Overall: The merge issue will need to be resolved before this can be approved — MaxnaCarta  ( 💬 • 📝 ) 12:09, 26 October 2024 (UTC)

@MaxnaCarta: that merge template is about history merging, not the one you would expect. JuniperChill (talk) 12:16, 26 October 2024 (UTC)

@JuniperChill: ah, I see. Is that template not a barrier to passing? I wouldn’t have thought it’s okay for an article to appear on the main page while it has some sort of call to action template. Also, is your username referencing the Juniper Inn from Hotel Hell? — MaxnaCarta  ( 💬 • 📝 ) 13:25, 26 October 2024 (UTC)

@MaxnaCarta: The reason why a history merge is being presented is because this page was (possibly) cut and pasted from Draft:Talerddig rail accident. And that will be accepted/declined well before it hits the main page anyway. And with regards where my username came from, I got it from my favourite plant. I also wanted to add a little chill to my username. I never watched Hotel Hell or even heard of it, but heard Gordon Ramsay. JuniperChill (talk) 16:51, 26 October 2024 (UTC)

@JuniperChill: Consider the antepenultimate version of the draft, i.e. that before it was blanked and redirected, and also consider the version of the article as it stood at the same moment. There are significant differences; it is clear to me that there was no cut-and-paste. They were started independently, both in mainspace, although the one that is presently in draftspace was moved there by Discospinster (talk · contribs). I don't think that a histmerge would be useful. --Redrose64 🌹 (talk) 17:35, 26 October 2024 (UTC)

Redrose64 is correct. There was no cut and paste involved, therefore there is no need for a history merge. I've removed the template. Mjroots (talk) 09:43, 27 October 2024 (UTC)

That's good to hear. Now we'll have to wait for the original reviewer to approve this. I would also note (as seen from my comment at the top) that I also created my own version of this crash, but since I realised one was already made, I redirected that. That's why I normally do DYKs on articles I created and don't really do drive-by nominations (in fact, most DYKs are self-noms). JuniperChill (talk) 11:01, 27 October 2024 (UTC)

We are all good to go, merge issue resolved — MaxnaCarta  ( 💬 • 📝 ) 22:01, 27 October 2024 (UTC)

@MaxnaCarta: You're using the wrong checkmark for DYK. Its supposed to be {{subst:DYKtick}} JuniperChill (talk) 22:03, 29 October 2024 (UTC)

Point worth noting about RAIB report

The content of the RAIB preliminary report - both text and images - are released under the Open Government Licence v3.0, which is equivalent to CC-BY-4.0-SA. That means, with attribution of course, any or all of the content of that report can be incorporated into the article. It is not necessary to paraphrase. I have already lifted one of the two images from the preliminary report. 10mmsocket (talk) 22:40, 5 November 2024 (UTC)

This means that the correct licensing template for such images is c:Template:OGL3, and not c:Template:cc-by-4.0. --Redrose64 🌹 (talk) 23:23, 5 November 2024 (UTC)

Thanks for that. I learned something new! 10mmsocket (talk) 07:26, 6 November 2024 (UTC)

Sander failure

Should we mention the 2010 incident at Stonegate where a train was in service with inoperable sanders? Mjroots (talk) 09:34, 6 November 2024 (UTC)

Do the secondary sources mention it in this context? -- DeFacto (talk). 10:13, 6 November 2024 (UTC)

See section 222 of the RAIB report. It attributes an empty sand hopper, not a blocked sanding system, as one of the causal factors. So in that respect it's slightly different. However failing to stop because of low adhesion compounded by sanding problems means (to me) it's worth a mention. 10mmsocket (talk) 10:22, 6 November 2024 (UTC)

@DeFacto: - not yet, but if we need to wait I would expect it would be picked up in due course. Might be worth keeping an eye on the Rail website. Mjroots (talk) 10:51, 6 November 2024 (UTC)

Rail magazine issue 1022, due to be published 13 November, should be a worthwhile investment. Mjroots (talk) 13:06, 7 November 2024 (UTC)

Which train was which?

I can find sources which state the numbers of the two trains involved, 158824 and 158841, but unless I've missed it I can't find anything that states which unit formed which train. Is this stated anywhere? — Voice of Clam (talk) 17:28, 9 November 2024 (UTC)

Hmmm. You make a really interesting point. Read the RAIB 5 November Update (source #5 in the article) and you can see the service number and confirmation from the pictures that the 1S71 the 19:09 Machynlleth to Shrewsbury is shown as the one sustaining the most damage, but I cannot see any other online-verifiable source which links the unit numbers (confirmed in the Stubbings Rail article on 25 October) to the service numbers. 10mmsocket (talk) 18:28, 9 November 2024 (UTC)

The Railway Magazine (full citation in our article) states involved train 1J25, the 18.31 Shrewsbury to Aberystwyth (operated by No. 158841) and the 1S71, the 19.09 Machynlleth to Shrewsbury (formed of No. 158824). --Redrose64 🌹 (talk) 00:23, 10 November 2024 (UTC)

I've added the relevant info to the infobox. Mjroots (talk) 10:48, 10 November 2024 (UTC)

Previous accident at Talerddig

Seeing that operation of the line in steam days is mentioned, should we mention the previous accident at Talerddig (linked in Article name section above)? Mjroots (talk) 05:37, 14 November 2024 (UTC)

Curiously, the 1894 collision is not described in Kidner, which has a section on accidents (pp. 81-83). Possibly this is omitted because there were no injuries to passengers, and only one member of staff was hurt. --Redrose64 🌹 (talk) 13:04, 14 November 2024 (UTC)

Train reporting number

The westbound train's reporting number is given as 1J25 when first mentioned and then later as 1J21. Do we know which is correct? ChrisKnight2578 (talk) 15:49, 11 December 2024 (UTC)

Fixed. 10mmsocket (talk) 16:48, 11 December 2024 (UTC)

Yep well spotted. A typo on my part that was missed by everyone till today. 1J25 is correct. Thanks 10mmsocket for correcting. — Voice of Clam (talk) 17:46, 11 December 2024 (UTC)