Steam speed records including City of Truro and Mallard

I don't believe that racing was the cause of the Salisbury accident.

There was a brief outbreak of racing between the GWR and the LSWR about two years earlier in April 1904; however, there had been subsequent very firm management action by the LSWR, starting with Drummond, which had stopped that occurring. Any racing that did happen seems to have lasted no more than four weeks, i.e. for four trains. (The specials ran once per week). Drummond had made it clear that any driver running more than five minutes ahead of schedule risked losing his job, and in the 1906 case, Driver Robbins was clearly well aware of that and had remarked on it before leaving Templecombe. Moreover, the train then proceeded to lose 4 minutes on schedule in the first twenty miles of running from Templecombe, which is hardly indicative of a crew out to break records.

The most likely reason for the Salisbury crash is that the driver momentarily fell asleep at a crucial period approaching Salisbury, probably a consequence of the hours and shift pattern he had worked in the run up to the fateful day. Those working hours were laid out in some detail in Major Pringle's report into the Salisbury accident.

With regard Rous-Marten, although on many occasions he may have published detailed logs, he was also not averse at times to writing dramatic reports not backed up by detailed timings. In particular, after the 1906 crash, he seemed to wage a one-man campaign to show that excessive speed was not the cause of the Salisbury accident. As "proof" he made reference to a fast run from 23 April 1904, during the very brief flurry of "racing". However, the published description in 1904 was vague or backed by timings that clearly were not internally consistent. For example, "our London Correspondent" in the Western Daily Mercury - probably Rous Marten writing anonymously - says "The train dashed through Clapham Junction at 7.59 and drew up with a flourish at Platform No. 5 in Waterloo Station exactly as the hands of great clock pointed to 8.3". The observant will note that one time is from a stopwatch and the next from a station clock, so any errors in calibration between the two over such a short distance can have very big impact on the speeds run. Clearly in that particular case, he was writing for dramatic effect rather than scientific rigour.

Subsequently, after the 1906 accident, Rous-Marten went on the record to prove that speed through Salisbury was not the cause of the accident, and as evidence he once again went back to the 23 April 1904 run and gave further details of the running in the vicinity of Salisbury - he mentions the train running at "75mph" for the half mile approaching Salisbury, then "at full speed" and "without appreciable slowing" through the station, and at "60mph" at the Tunnel on the exit to the station, but without detailed timings to back that assertion up. By contrast, it is important to realise that prior to that run, Drummond had issued an edict to run at no more than 30mph through Salisbury only three days earlier; and the Outdoor Running Superintendent was on the footplate for the special, so it seems inconceivable that the crew on that occasion ran beyond those speeds, within the limits of accuracy of running on locos with no speedometers.

Rous-Marten seemed somewhat alone in trying to demonstrate that speed wasn't the cause of the Salisbury derailment, even going so far as to speculate that the loco may have suffered a failure of a tyre or crank axle. (No such damage was found). He put forward the 23 April 1904 run as "proof" that it was possible to run through Salisbury at high speed, but with no actual proof of the actual speed on that run. I think the end-to-end times (Plymouth - Templecombe and Templecombe to Waterloo) can be considered as robust - at least to the nearest minute over about four hours elapsed - for those trains during the very brief period of racing, but the point speeds at particular places on the journey are much more debatable.

Finally, it is worth mentioning this comment from The Railway Engineer, published in the aftermath of the Salisbury accident:

"Only a few years back when a train left a station late it arrived at the next equally late. But a band of self-appointed railway experts arose and found it profitable to make 'logs of runs' for the daily and other papers, and they began, in the language of The Jungle to 'speed up' one railway after another.​

The arch speed expert, Mr Rous-Marten, was soon in print with the object of showing that speed was not to blame [at Salisbury], but that appearances pointed to some 'hidden flaw in a wheel or axle or in the permanent way' and to denounce those who dared to express a different opinion as 'irresponsible and ill informed'".
​

Tom

 

Rous Marten gets criticised frequently by writers.

I personally don’t put much, if any stock, in him.

 

I don't know how reliable that nice book S N Pike Mile by Mile on Britain's Railways is, but that says 10 mph.

 

No, and Maj Pringle did calculations that an L12 (involved in the accident) would tip over at 67mph. The loco involved on 23 April 1904 was a T9 with a lower centre of gravity, and might just have scraped round at the speed Rous-Marten implied to have occurred, but it seems inconceivable that the driver would have run at that speed when Drummond had issued an explicit 30mph limit only three days before and he had one of the company’s senior officers on the footplate!

It seems fairly clear to me that the Salisbury accident can best be explained as a consequence of driver tiredness, probably caused by a working pattern that wouldn’t be tolerated now.The loco was going too fast for the curve, but not because that was habitual practice, but rather as a consequence of a tragic error.

Rous-Marten then tried to imply that, on the basis of an earlier run with a different loco without any robust timing evidence, that high speed was not to blame for the subsequent Salisbury disaster. In hindsight, his comments at the time feel pretty disreputable: in essence he was presenting an assertion about speed as if it were fact, and using that to justify an approach to safety that was very cavalier in the interests of breaking a few records.

In later years there has also grown up a view - in the knowledge that the GWR had the previous day opened the Castle Cary cut off which gave them an unassailable advantage in distance to Plymouth - that the LSWR was out for one last “hurrah” to show what they could do, but the evidence presented in the accident report I think conclusively refutes that. If they were racing, how come the first part of the special only gained one minute on schedule, and the second part had dropped time on the way to Salisbury?

Ultimately it was an accident caused by fatigue, and speculation about racing is not supported by any evidence.

Tom

 

I think Waszak's Rail Centres: Peterborough quotes 15mph but I suppose what we really need is a RCH map? It certainly was a fierce restriction and I think it came about because the GNR and MR wouldn't play nicely with each other, despite later forming a much-celebrated joint operation. The dog-leg was sharp enough to warrant an assisting locomotive to get long trains out, I think - a C12 4-4-2T was often used.

(I'd forgotten Mile By Mile - I have the LNER issue of it for the ECML)

 

According to the graph within the I.Mech Eng paper

At MP 91 Red line, 124.3mph, Blue 124mph.Black 123.7
( within +/- minus errors the same number I.e. 124mph

At MP 90.5 Red 123.5, Blue 123.3 Black 123.9.
( within errors 123.5 mph

At MP 90.25 Red 125.0 Blue 125.0 Black 124.0
( within errors 124.8)

At MP 90 Red 124.2, Blue 124.4 Black 124.0
( within errors 124 mph

Between MP 90.25 and 90,0
Red falls from 125.0 to 124.3, blue rises to 126 and falls back to 124.5 in 440 yards,
black remains at 124

The DBHP ( the power available at the drawbar to haul the carriages falls from
1588 at MP92 at 120mph to 1260 at MP 90 at 124mph.

Sorry Big Al your logic IMHO does not stack up.

Michael Rowe

 

With respect to everyone looking at the Andrews paper - I respect him for sticking his neck out, but my contention has been, and always will be, that he was interpreting the data wrong.

Fundamentally he did not fully understand how the dynamometer car worked.

Even when I used a conservative estimate of the speeds and times recorded on the graph paper, by taking the actual quarter, half, three quarter and full mile records instead of the (after the event) notations on the graph paper, you would still end up with instantaneous speeds recorded that are higher than those given by the L.N.E.R. team, who did a very specific method of analysing the data (five second intervals for average speeds) as opposed to taking the distance, and measuring the time recorded, to get the instantaneous speeds. All the on board team cared about was making sure that it had achieved above 124mph: which it did.

 

Referring to Salisbury - anyone who chances upon a copy of "Salisbury 1906 - An Answer to the Enigma? 3rd Edition" by the late Norman Pattenden, don't hesitate, buy it! It is a South Western Circle Monograph, but may still be available to non-members at the stand that attends various model railway shows. The 2nd edition was produced for the Institute of Rail Operators as a textbook example of accident enquiry. And, yes, a microsleep was his conclusion.
Pat

 

I agree completely and that is borne out by the dynamometer roll when you analyse the data recorded on it.

 

Are the quarter mile speed in lower rigth corner from five second averaging done by Gresley and his mery men and written on roll?

Mallard by Hermod posted Dec 5, 2023 at 10:03 AM

 

Gentlemen, talking in whole numbers of MPH is inevitably misleading. If you are going to seek instantaneous speeds - something whose validity I am entirely in agreement with Gresley about- then I submit that really you need to be presenting the speed at each point including at least one figure of decimals and with error bars showing the limitations of the measurement method.

 

Hi Jim - I agree - and that’s what I have been working on for some time.

 

But people will focus on the precision and ignore the error bars … (so “126.1 +/- 2” will inevitably be reported as “126.1” or maybe even “at least 126.1” totally ignoring the down side of the error bar).

Tom

 

People will always over-simplify, as this discussion has illustrated. If I see a value of 100 quoted with an estimated error of +/-5%, I will interpret that as 100 - and if I'm quoting it and don't want my reader to anchor on "100", I'll say "between 95 & 105".

 

All very interesting stuff. But does it actually matter?

Mallard reached something around 125 to 126 mph, travelling downhill. The German record holder reached about 124 mph on level track. Which is better?

Of course there is also the issue that the German contender was one of a tiny class of basically experimental prototypes, whereas Mallard was one of a large class of locomotives in successful everyday use for many years.

We don’t know what would have happened to German steam development if wartime conditions had not put an abrupt end to the speed-record story.

 

The other question though is how fast would it be possible to run with steam traction regularly without special preparation