Sir Nigel Gresley - The L.N.E.R.’s First C.M.E.

I don’t believe Mr Andrews has demonstrated that. I think there’s a fundamental misunderstanding of how the dynamometer car works, which is why my approach is different to his.

He highlighted a portion of the dynamometer roll’s one second interval and has showed us an odd tick and stated it shows a variance and is proof which means the roll was speeding up and slowing down.

If this was true, you would expect it to only happen on a few one second intervals throughout the paper, right? Or to only be on a few in a pattern, throughout the paper. Right?

The trouble is, it’s on every one second interval mark, which are also identical in length to each other. The slight tick at the top (and the bottom) of each one second sine wave is the marker for…one unit of one second.

The LNER assistants on the train understood that, and that’s why they use this as their marker for their 5 second intervals on every one second mark at the end of the five second interval sections.

Why is this? The dynamometer car records time regularly, not irregularly, and measures distance travelled against that.

So you have time in seconds measured entirely regularly and distance dependent on actual distance travelled.

This proves unequivocally that the roll was passing through the rollers at a constant speed, and not an irregular one. It in fact disproves David Andrews theory.

If the roll was in fact speeding up and slowing down in any way, you would expect to find a variance in the length of the recorded one second interval by way of a regularly recorded difference in length of one second interval on the paper itself.

Which I can confirm, there is no variance: they are all identical to each other throughout the entire roll. Which means it was travelling at a constant speed through the rollers, as I have always suspected.

The dynamometer car was always calibrated to record time primarily. Distance was measured in quarter miles, with the machine set up to mark each quarter mile and then the full mile slightly differently. Which is why you see three identical quarter mile patterns and for every fourth quarter mile, you see a slightly different pattern applied.

The distance between each quarter mile is the time taken and that is why the dynamometer roll is set up to record one second intervals. When you compare the lengths between the quarter miles taken, of course the distance between them is irregular. It is meant to be irregular. It is time recorded regularly with distance recorded against it based on actual distance travelled.

This is why I am now convinced that Mallard achieved 126mph outright.

Why am I so confident in this? I have the whole roll digitised in front of me and can report on it fully to you.

 

I'm not sure you're understanding of what he's saying is the same as mine. We don't have a sine wave trace on the paper. The one second "tick" trace approximates to a square wave. Its not actually a square wave because its a mechanical device, so we get a rise on the wave form.
My understanding is that the reason he has highlighting the one second trace was to demonstrate that claiming a speed on the basis of a single 'tick' is unreliable because of the width of the actual line and the relatively slow rise of the wave form. I'm inclined to think that the "spike" is more an artifact of an electromechanical mechanism rather than a deliberate design feature, but its just as useful (and limited) either way. Anyway, I submit that part of his paper is reasonably uncontroversial and I don't believe its where he's proposing an inaccuracy in the mechanism.


No, to my mind where sine waves come in is with the other chart, this one of instantaneous speed against one mile average speed. Now I have to say I am not altogether comfortable with this chart. Graphing instantaneous speed against the average makes it a bit more readable, but it adds a level of indirection I am not altogether happy about. Is it a coincidence that his pulse every mile matches the mile he is averaging against? I'm just a bit unhappy about that now I come to look at it in detail. Anyway this is where the sine wave comes in, and I've drawn a sine wave in blue over the top to clarify it a bit. What he's claiming is that as a result of an inaccuracy, presumably in the mechanism somewhere, there's an underlying pulse where the measured speed is varying consistently over each mile. If he's correct in this then this underlying pulse would be visible in the entire trace from start to finish.



Now as I've said I'm a little unhappy about this chart. So is there anything we can do to ensure this is not an artifact of how he is interpreting the data rather than something that's really there? The first would be to look at the mechanism itself and the gearing, if it all still exists or a working drawing does. Are there in fact any shafts or anything else that rotates once per mile? If there is indeed a shaft in the paper drive mechanism that rotates once per mile that's something of a smoking gun. If there isn't then we need to look more carefully. How would I do it? Well the first thing I would do is to move away from the actual speed record area of the trace and try and find a section where the train is moving at a reasonably steady speed - if this variation does exist it will exist over the whole trace and if I can find a spot where speed doesn't vary too much it will make interpretation easier. Over about 5 miles of trace I would measure the ticks as well as I could. Ideally one would measure single ticks, but I don't believe its possible to do that accurately enough. I think I'd probably do say 4 seconds and 10 seconds, and do the 10 ticks twice, overlapping so they are 5 seconds apart. Now chart the actual measurements for all three, in real time. What should happen is that there are three lines that almost overlap, but not quite, and the 4 tick line will be spikier than the 10 tick line. If Mr Andrews interpretation is correct then we should see a similar undulation in all three plots that matches a sine wave drawn over it in the same way I've drawn one over his chart. If on the other hand no such consistent undulation is visible then I would conclude that what we are looking at is an artifact of his interpretation rather than a real phenomenom. This isn't my area of expertise though, there may be better ways of achieving the same thing.

 

Looking forward to that, Simon.

It would be good to go through all the moving parts. How confident are we that the quarter miles were not actually 439 yards apart, for example?

Isn't the language of probabilities better? 99% confident it did 125, 60% confident it did 126 or whatever?

 

Okay, I think we need to be clear on the terminology and what's being said. My apologies for not being clearer.

When I say "quarter mile" in relation to the graph paper, that is a physical measurement of the actual distance covered, and is not in any way linked to the posts on the ground by the track.

As far as I'm concerned, the posts on the track can go hang. We don't need them. They may well be inaccurate. They may actually be inaccurate. They are only relevant if you care about where 126mph might have been recorded (for the record, I believe that the Gresley Society sign is in the right area for 126mph being achieved and can therefore stay put!)

The dynamometer roll records actual distance covered by the train. The ninth wheel under the dynamometer car measures distance and does this by knowing the number of rotations it does.

Yes, you're quite right. That's why I'm 100% certain it did 126mph. I am 90% certain it went faster than that. That is based on wanting to do a peer review at a later date.

I agree Jim, and that's why I had another look. But his approach is flawed because he is coming at it from a distance perspective and not a time based one. The dynamometer car was designed to provide this manner of recording one second intervals. That allows for accurate interpretation of the graph.

Incidentally, it does the same small tick for the full mile recording. It's not an accident, it's not an issue with the electromagnetic pen, it's a designed in feature for measuring the data the dynamometer car records.

Precisely what I have done and as I have said above, it's not an error or an issue with the equipment. It's a designed in feature of the recording mechanism.

Jim, I'm going to send you a PM. That may be helpful.

 

Based on the thread linked below I believe Courier is David Andrews even though his profile shows another name.

He contributed an article about Mallard’s speed record to BackTrack Magazine. @S.A.C. Martin looked forward to reading it in his lunchtime

Some readers did not seem to appreciate the idea of a historian re-examining such a cherished piece of history; after the thread petered out Courier stopped contributing to this forum. Whether his latest theory turns out to be right or wrong I’m very glad Simon’s still here.

https://www.national-preservation.com/threads/mallards-maximum-speed.1191983/

 

Interesting if they do turn out to be the same person. The articles differ to each other. I note I disagreed with the article published at the time too.

I think it’s great he went out and did the work. I’ve got a different point of view but that’s fine, we are effectively peer reviewing one another with our work and it gives people more information to make their own minds up.

I’ve been playing around with it more today and trying to produce something better than a two decimal place measurement per quarter mile but truth be told I think I’m on a hiding to nothing with that.

There’s a point where applying digital methods to analogue methods doesn’t make sense anymore and I think I’ve reached that limit given the evidence.

I remain convinced 126mph was sustained, for my own piece of mind I’m very glad I waited to review the roll before I sent my book off for publishing, it’s allowed me to write a piece on it that feels fair and balanced to me based on the evidence I’ve reviewed.

I agree with David Andrews on one very important point. Sustaining 124mph for nearly a mile was an achievement too, much overlooked.

It does occur to me that of all the A4s, Mallard went the fastest but the A4 that probably achieved the most was Silver Link.

That she didn’t survive the cutters torch is something I suspect LNER enthusiasts will long regret…

 

I'm a bit busy at the moment and haven't had time to read Dave Andrews paper and have only skimmed what has been said on here so it might already have been said. However. looking at the trace and the so called 'tick' shown in Jimc's post, that looks simply to be pen overshoot, which is quite common on graphs created in this way. I'm very doubtful that it is a design feature. The paper is travelling at a uniform speed and the pen takes a finite time to move so the line created is never truly vertical. The pen obeys the laws of physics and does not stop instantaneously so actually overshoots fractionally before returning to its intended position. Even lightweight pens do this and I doubt that you could class those on the dynomometer car lightweight. That is why you get a tick at each change of position. I spent a lot of my professional life looking at speed - distance and time graphs (not on railways) and this was quite common.

 

Hi Steve. I have the full dynamometer roll in front of me and assuredly it’s designed in.

Not to undermine your view but there’s a clear difference between something which is a pen flick - which might not happen every time - and something which the LNER’s own assistants clearly used to measure against.

It’s also used on the measurements for quarter and full mile distances, but not on drawbar horsepower or steam applied, which are recorded differently.

With respect I have the advantage of the full roll in front of me and can say that with confidence.

The NER and then LNER were not messing around with their datum points, that is for certain!

 

There's a difference between "designed in" and "something which the LNER’s own assistants clearly used to measure against."

If the little spike is caused by the inertia of the pen - as @Steve suggests - it would occur pretty consistently; after all, the weight of the pen is constant and the laws of physics occur every time it changes direction. Given that, it would be a useful marker for measuring, occurring regularly and with a sharp peak to measure from. But that is different from "designed in". For that to be true, you'd need to identify a piece of mechanism that was explicitly designed to create that little spike - which would seem unlikely.

To take an example from a different field. In my research career ("High-valent half-sandwich hydrazido and alkoxylimido compounds of the early transition metals" - look it up at a good bookseller near you) we used to routinely run proton-NMR spectra to help identify the structures of new compounds. You’d dissolve your sample in per-deutero benzene, C6D6, and routinely the biggest spike you saw was at 7.16ppm, regardless of what your compound was. If you were a bit clumsy, you'd also get a peak at about 0.2ppm, again regardless of your sample. Neither was "by design" but were instead caused by impurities: the 7.16ppm peak was the residual proton in C6D5H which came as a trace impurity in C6D6; and the 0.2ppm peak was caused by the methyl groups in the silicone grease used for sealing glassware: you got it if you were too clumsy to avoid contaminating your samples with grease.

Neither of those peaks occurred "by design", but both were "something which we used to measure against", since they provided two known calibration values at either end of the spectrum (which for the compounds we were interested in typically gave values between about 0 and 8ppm). So the point is, a regular blip can be simultaneously "something used to measure against" but also "not there by design".

Tom

 

Don't get me wrong, I'm enjoying reading the scientific debate here as I always do, but what are we actually trying to prove here? What is the end goal? I'm assuming Simon's book on Gresley has a similar purpose to his book on Thompson - albeit starting from a different default point of view - namely an attempt at a data-led approach to appraising how successful a CME they were. A Gresley-designed loco achieving the steam record is undoubtedly an important part of that story. But is whether it did 125.5mph or briefly 126.0mph really going to make any difference to the overall judgment of the man and his achievements? How precise do you want to get, and at what point does increasing precision become completely worthless?

For determining actual records, 3, 4, even 5 significant figures are of course important. For general judgments though, 2 significant figures are enough to go "Bloody hell that was quick, impressive design". Especially in the context of other data being used like the engine availability stats and all the rest of it. Simon's graphed it up and we look at trends, and the discussion was around making sure we're comparing the right things and processing the data in the right way, not whether A3s had an availability of 84% or 85%.

Just my four penneth worth, as I say, I do find the debate interesting, I'm just not sure what the point of it is

 

But a pen overshoot will happen every time. Assuming its driven by something like a solenoid. The pen will accelerate up or down but when it reaches the point where the solenoid is trying to put it its inertia will put the flick in. That's just how the mechanism works, and also there is no downside. The overshoot mark is there, consistent and dead handy to measure against. Trying to eliminate the pen overshoot would have no advantages and quite probably a considerable number of disadvantages. Consider: would any sane practical engineer firstly design a mechanism that eliminates pen overshoot, and then add a second mechanism that puts it back in. What might happen, I suppose could be that if the pen overshoot was desired and not clear enough then you might do something to exaggerate it - a bit of extra weight on the pen arm ought to do the job nicely. But I'm somewhat bemused about why you are bothered about this.

The trouble too with this very technical discussion on the mechanism is that its largely pointless without the mechanism or its drawings to study. Take Mr Andrews' every mile variation (which is quite a different matter to this pen overshoot business and potentially much more controversial). For that to exist and be an artifact of the mechanism then there probably needs to be a component in the drive train rotating at one revolution per mile (unless there is some subtlety I'm not clever enough to imagine). If there's nothing in the drive train rotating at that speed (and I'm not sure why there should be) then his theory is in trouble unless he has very good evidence. On the other hand if there is a visible one per mile wave pattern in the signal noise then something must be causing it. I've assumed it would have to be paper speed, but that's not a safe assumption. Just because I can't imagine another cause doesn't mean there isn't one.

 

I dunno, I think the idea that the deliberate shape of the 1 second interval readings is somehow just a by product of the way the pen is supposed to work doesn’t hold water when you realise the paper roll has five separate lines drawn on it by different parts of the overall equipment (and some are designed to produce a different pattern than the others in order to show what they’re recording - truly, the roll is a marvel to read).

I’m bemused that we’re suggesting that the NER and then LNER, probably the most scientific of the railways (LMS accepted for their efforts at the National Physics Laboratory) didn’t spend quite a lot of time refining how and what they recorded. I think the use of the flick, if thats what we want to call it, to book end the one second intervals makes the reading of it all the easier - the handwritten notes and markings on them really confirm to me how the LNER assistants went about their work. There’s also some brilliant mathematics recorded on it, which I had to replicate by using a spreadsheet and found the assistant’s answer was only something like 0.03 out, which for someone doing this without a calculator is something else. Truly humbling!

And that matters - the quality of the workmanship, the scientific led approach, the human interface with the machine - seeing the speed record roll in its entirety has given me a new found respect for the men operating the dynamometer car and those who originally built it. To run it requires more than a grasp of maths and physics and understanding what you’re reading all play into what was undoubtedly a very well looked over and much loved bit of machinery. The dynamometer is, frankly, one of the best parts of the Mallard story and it’s fascinating to analyse.

A very good question and the answer is that I wanted to analyse the data myself to see if what I have read and been told for decades was largely correct - and it was. The last nagging doubts I had about 126mph are now answered. I’m satisfied with what I’ve seen, whether there is any follow up largely depends on how far I take it now.

I think enthusiasts are wont to discuss and think over things. I am grateful for the additional commentary the last few days.

Look, ultimately there’s always going to be questions to answer for. And we can keep going, for certain. Let’s say I do my study of the mechanism, I look at the internal workings, what will be the next question? How quick did the ink dry…?

For my part, I think I can put this thread to bed now, with the book submitted for final layout and printing.

So, if you’ve been reading from the start of this and possibly the Thompson thread too - thanks for taking part, everything has added to the debate and knowledge gathering. Hopefully this thread will serve as a good reference for debate in future.

 

Sounds am interesting read, though Late TM is more my kinda thing.
I will admit on occasion using the grease peak to reference my spectrum if aromatic region was too messy.

Sent from my moto g(7) play using Tapatalk

 

Here’s some very good pictures of the equipment, including lots of pens, inside the dynamometer car.
https://collection.sciencemuseumgro...tern-railway-dynamometer-car-railway-carriage

I wonder when it was last used.

 

Great link, thank you for sharing. Note this photograph, showing all of the different pens and what they recorded on the paper.



I do think the ingenious of the people who worked on these things can be seen in the craftsmanship of it all. Really beautifully made.

 

Jimc has explained why you wil lget pen overshoot far better than I did and asked the important question how the tick was designed into the system if you think that it was. You have only to look at the Science Museum photos linked by Miff to see that the pens and assoicate solenoids were not exactly lightweight and would have significant inertia. Elimination of the overshoot would have required significant damping and led to a curved trace more akin to a sine wave and, in reality, it is not a problem. The 'tick' is convenient for measuring purposes but so would any repeated change of line direction such as appears at the other end of the square trace.

 

Look at the photo above Steve. You’ll get your answer!

 

For the scientifically and engineering hard of thinking, please could you clarify.


Sent from my iPhone using Tapatalk

 

I did but I don’t know the question it answers.

 

I'm confused as well. To support the assertion that the blip in the trace is there by design, you've got to simultaneously identify some part of the mechanism to remove the inertial movement of the pen, and then something else to deliberately reinstate it! That seems unlikely to me.

The Dynamometer didn't come out of nowhere. The NER car wasn't the first, and development of precision recording machinery in other fields had developed in the Victorian era. So the designers would have been well aware of the quirks of such things, and would have been well aware that you don't get sharp corners in traces. Bear in mind as well that the Dynamometer car would have been in regular use for a wide variety of measurements. It was hardly designed to answer a question about whether a loco went for a fraction of a mile at 125.9 or 126.1mph, because that's basically irrelevant in commercial terms.

Tom