Alternative High Speed Steam?

You disagree with me in the first sentence and then basically make my case for me that we need more evidence.

I'll let myself out!

 

I think one of the things that folk don't make enough allowance for is that any recorded speed is not really a single figure, but a range, and the width of the range depends on the granularity of the equipment. Rous Marten was using a 1/5 second stopwatch, so he didn't really record 102.3mph, he recorded between 101.1 and 103.5, and the actual velocity could be anywhere between those two figures. And then if the actual time was, say 8.9 seconds, then the stopwatch would be equally capable of recording 9.0 or 8.8 seconds. And human error, of course, could equally throw up a 1/5 second error. Rous Marten knew all this, and used a complex setup with multiple watches so one cross checked the other - except that the driver braked and he never got the second reading below 9 seconds which would have settled things as well as could be managed with that technology. Its possible to draw this up graphically. In this chart the darkest gray band is the speed range if you assume no errors in the timing. The paler bands are 1/5 second and 2/5 second errors. Because of his cross checks 1/5 second errors aren't very likely in the earlier figures, but must be accepted as quite possible in the last figure. And I have drawn two entirely hypothetical lines to hazard a best fit as to what the real speed might have been. The yellow line is a good fit assuming all measurements are correct. You'll see it comes out almost on the edge of the dark gray band. This suggests the real speed was at the bottom of the range, around 101. The blue line, on the other hand, is the most pessimistic interpretation, and represents the last number having a 1/5 second error, so the time for the last quarter mile should have been 9 seconds. By taking a very loose fit, including the assumption of 1/5 second errors further down, its possible to produce a feasible, if somewhat unlikely, fit that comes out below 100mph. I haven't thrown any proper analysis at this, but I would think a speed in the 100 - 101 range most likely, but a speed below 100mph isn't impossible. The other thing that's evident is that on both line fits the train is still accelerating, so if the driver hadn't braked...


... we could be talking about the great Wellington mail train accident.

Of course if you don't accept the published figures all this is moot.

 

Is there another source of variation in how accurately surveyed the 1/4 mile markers are? Which supports your point about a range, but just widens the possibilities of that range? Even 1 yard of error would be about 0.25mph - you'd think over a long distance they should be correct, but we are typically talking about timing over only one or two quarters. Again, the error could be in either direction. (I seem to recall there's a 1/4 mile near Salisbury that was about a hundred yards short, but that was by design ...)

Similarly - is the track curved at that point? As a fag packet calculation, on a double track line curving at, say, 30 chains radius, the outside curve has a radius about 10 feet bigger than the inside - approximately 0.5% longer than the inside. That's enough to make 0.5mph difference in speed: which way depends on whether the surveyed "1/4 mile" mileposts are inside the curve or outside, and where the loco is. The most favourable situation would be with accurate 1/4 mile markers on the outside of a curve but the loco travelling on the inside; least favourable in the opposite situation.

Tom

 

Hopefully you are not going for a speed record around a 30 chain radius curve?

 

Very sure footed them Great Western engines…


Reasons why CoT could not have travelled at 100mph? And….Go!

(Lack of perfectly calibrated evidence is not a valid reason @S.A.C. Martin )

 

The blue line would also corespond to a decreasing rate of acceleration, which would be expected.

 

What’s the radius on the approach to Crewe?

My sense is that at times there was quite a lot of hard running on curves in the late Victorian and Edwardian era, that was just about tenable with small engines but became increasingly less so as weights and centres of gravity went up.

Tom

 

I've looked up Rous-Martens' original report of the London to Brighton record; it is in the Railway Magazine for September 1903. It was a special train made up of Pullman cars, and he reports the maximum of 90 mph at the bottom of the 1-in-264 descent through Haywards Heath, and comments that this speed was corroborated by an independent observer and also by Mr Billinton's recording apparatus of which I know no details.

 

Your argument seems to be that, because speeds above certain levels would have been exceptional (true) and because there is no reliable evidence of their happening (also true), therefore they never did happen.

 

I think it's not unfair to say that we should treat those claims with some scepticism? How can we sit here and say, without any concrete evidence, that something definitely happened?

The onus is on those making the claim to prove their claim, not the other way around.

 

This is an interesting post Jim. The thing is, you're applying a scientific analysis to a method of recording that is very unscientific and open to error. I don't think you can assume no errors in timing - human reaction time/stopwatch/placement of posts/etc introduce a wider range of variables than a dynamometer car.

Do we have Rous Marten's full log available to analyse what he recorded?

 

Which was a possible (probable?) cause of the Salisbury derailment.

 

Indeed. The loco overturned on a curve with a 15mph limit, probably doing about 60mph from known timings. It was a Drummond L11 which had a boiler pitched 9 inches higher than the more usual types for that service; it was also a rare exception of a train that wasn’t scheduled to stop at Salisbury - after that I believe the stop became mandatory.

Tom

 

Until fairly recently stop watches have been the norm for recording timings of events, whether on the railway or on the sports track. Are you saying that now we have better methods of timing these historical records have no meaning. Dynamometer cars are also not necessarily accurate just like speed cameras on the road. Was it the LMS one that was found to be inaccurate at one stage? I can't remember. To prove accuracy you need an unchanging calibration check before and after the event.

 

No, I am not saying that. I am saying that we need more than "Rous-Marten recorded 100mph therefore it must be true". Where is his full log? Can we see the timings? Can we analyse the likely the required horsepower for CoT to achieve, etc etc.

I am sceptical of the most extreme of timings done with stopwatches because, as highlighted by Jim above, we are missing a corresponding log that could have helped give credence in one case, and in other cases - where are these logs for us to analyse?

Yes, in all the dynamometer car cases calibration was necessary and procedures (which we have documented) were followed. The variables of the dynamometer car output are less wide than that of the stopwatch, I humbly suggest. That's why the dynamometer car was developed.

Modern tech for recording things including speed, distance and time have got better with history. If they hadn't, we would be in trouble.

 

Yes, it was the ex-L&YR car which had not received the proper maintenance, as a result, part of the gear recording the drawbar pull partly seized. Nigel Gresley suspected a fault with data of Royal Scot coal consumption, 'borrowed' it and ran it back to back with the LNER car, and there were discrepancies between the two.

It wasn't related to speed, though. But this same car recorded 114 m.p.h. down from Whitmore with 6220 while all the stopwatches said 112.5 with a possible peak of 113. But the LMS got away with a claim for 114.

 

But that to me highlights the issue. The dynamometer car is undoubtedly going to be more accurate than the men with stopwatches. It is directly recording the results as it happens and has a tangible input to the data collection. The men with stopwatches are relying on visual cues, reaction times, etc.

To get within 2 mph of that Dynamometer is an impressive achievement, of course. But it shows that we can't rely on the stopwatches more than the dynamometer car readings. Which really is the crux of the matter for me. It's where unauthenticated/authenticated claims are split.

 

I think you are confusing accuracy and precision. Using a stopwatch might be accurate but with wide error bars. Whereas a badly calibrated dynamometer might be inaccurate but with narrow error bars.

(i.e. hypothetically if, say, the rolling wheel on a dynamometer was of incorrect diameter, or the clockwork mechanism ran at the wrong speed, the dynamometer might measure speeds very consistently, but get them wrong every time. Whereas the man with a stopwatch might get the speed right, but with a range of +/- several mph.)

The issues with speed records in this country don’t seem to me to be primarily concerned with how they are measured, and you can’t automatically say “dynamometer good, stopwatch bad”. Rather, the issues are (1) speeds are quoted to a level of precision that isn’t justified by the mechanism used and (2) the known values tend to tell you more about who had most confidence going flat out down a steep hill than the underlying performance of the locos concerned.

Tom

 

I don't believe I am Tom, I am saying that the dynamometer car is both a more accurate method of recording and more precise than the man with the stopwatch.

But a dynamometer car gets calibrated. We have the method, we have the procedure, we have the understanding of how it works, we can factor in all of the different variables and factors and we can get to a level of precision that the man with stopwatch cannot.

Ultimately we can review the dynamometer car evidence in a more scientific manner than the man with the stopwatch.

Yes I can! Dynamometer recordings that exist, good, man with stopwatch (without corresponding evidence in the form of the logs/timings etc) bad!

I feel strongly you're missing the point I am making Tom.

How can we possibly have a discussion over the precision with which someone with a stopwatch recorded a log if we haven't got the actual log to analyse?

Evidence - we need the actual logs. Give us the logs, let us look at them, then we can analyse away and say with some confidence whether it makes sense or not.

True of the stopwatch, not of the dynamometer car.

That much we absolutely agree on.

I am going to duck out for the time being I think, the discussion is fascinating but I am not going to be convinced of a timekeeper's stopwatch and method being more accurate/more precise than the dynamometers specifically developed by the railway companies for improving how they collated data on distance/time/speed/drawbar horsepower/etc.

 

Mmm, when you consider how uncontrolled and unscientific rail speed records were compared to almost anything else its a wonder that there is so much hot air - or is the inherently flakey measurement the cause of the hot air? One thing is for sure, go to one of the bodies that authenticate speeds for other vehicles and present them with just about any of the rail speed records and they'd fall about laughing. I suppose too there are all the problems with trying to establish records on a working railway.