Steam speed records including City of Truro and Mallard

How often does one have to say this. There was NEVER a 130mph Saint claim.

I've written this up as best as I can here.

http://devboats.co.uk/gwdrawings/2903myth.php

The beginning and end of it is that they set out to run the locomotive over 100mph, and claimed they succeeded in this. That is the only actual claim Collett made. They stated that they obtained a figure of 120mph - which is of course a *very* round number when it comes to stopwatch timings - but not in any way that they had confidence in. The stuff about 130 or 135, which seems only to be repeated by those seeking to debunk the non existent claims, comes from signal box timings which at best are plus/minus 35mph, and could easily be plus/minus 100.

 

I think Jim’s erudite analysis shows how the timings can fit the result desired.

I do think this is the wrong way round, though.

I think you should be starting from looking at the train itself and asking some basic questions about if the locomotive had the potential to achieve that speed in the first place.

Almost every single article I have seen on CoT’s speed run does a lot to analyse the timings, but doesn’t look at the boiler, evaporative rate, weight of the train, and any basic calcs for rolling resistance or wind resistance.

And the big thing that has become apparent from my research (and really ably shown up thread on an excellent diagram showing where steam development sort of got to around 90mph and remained pretty static for a long time) is that issues like these are consistently underplayed by enthusiasts throughout railway history. That’s before we get to signalling, gradients, and the like. Throw those in and I think you have significant grounds against achieving 100mph in a lot of cases.

So for me, I think there’s a significant amount of theoretical work that can easily through into question CoT’s 100mph outside of the timings published by Rous Marten. And for me, that balance of “did it/didn’t it” pushes me to “it definitely didn’t”.

Absolutely agree with you there Tom.

Sorry Jim, I wasn’t trying to make light of your work. I was absolutely being facetious/mischievous about the claim.

May I say, your write up there is excellent (I have in fact cited it in my paper).

Your write up does prove my point though - a lot of room for interpretation and a lot of room for error from what is still a secondary source at the end of the day.

 

I'm sure @Jimc can answer for himself. But I think it is a bit disingenuous to assume that the "timings [have been fitted] to the results desired". Rather, my reading is that the timings are taken at face value, and from that you get a range of speeds which are consistent with those times.

By contrast, I think it is you that is guilty of "trying to fit the data to get the results desired". Your strongly held view is that locos of that era couldn't exceed about 90mph. It seems to me that you are then taking that as a starting point to discount the data - but if so, you have to say not just that it is objectively wrong, but why such an error might have occurred. It's axiomatic in science that you can't simply ignore data that doesn't fit your theory: you have to either admit your theory is wrong, or else explain why the data are wrong. So you have a theory ("CoT definitely didn't achieve 100mph") and some data (which show that, within error bars, it may well have done and at the least got very close). So either your theory is wrong, or the data are wrong - but if so, you have to explain what the sources of error are in the data that make it systematically err on the side of being too fast.,

It is not sufficient to make a claim that Rous-Marten's data are wrong on the basis of power. Firstly, and less importantly, because you don't actually know what power CoT produced, nor what was needed to haul that train at that speed down that gradient. But more significantly, because we come back to the mismatch of theory and data. The power calculation is essentially your theoretical underpinning of the view that CoT didn't reach 100mph; therefore you cannot use that in itself to disprove the data - it becomes a circular argument.

The TL,DR: There is nothing wrong with putting forward a theory about the actual speed attained by CoT based on power calculations - but since that theory doesn't fit the known data points, you have to independently explain why those data points must be erroneous. Otherwise you are essentially saying "my theory says CoT didn't do 100mph, and therefore my theory proves that the data are wrong" - ignoring the fact that it could be the theory that is wrong and not the data.

Tom

 

Forgive me but I am not accusing Jim of trying to fit the data to get the results desired. I just want to make that explicitly clear.

I’ve not ever said that Tom, I think. If I have, quote me and I will happily correct myself.

I do think getting to 90mph is significantly more difficult for locomotives of that era than later steam locomotives.

I have used the word “theoretical” Tom. It’s right there in my paragraph. I haven't claimed it disproves it - I have said, for me, the balance of probability is on it didn’t do it.

It’s not circular. I’m starting from a different starting point. The first question is: is CoT developed enough as a steam locomotive design to mechanically achieve 100mph? If so, great, look at the timings and then from there you do the analysis of Rous Marten’s work.

If no, then you go to 90mph. You introduce the other factors. Does it look likely? If so, great, look at the timings. Can we see a potential for Rous-Marten being wrong, or right? How reliable is his account? Is there anything like this that exists on the same line?

You are placing a lot of emphasis on this particular time keeper’s “data” but this is a log based on a man, with a stop watch, looking out of a window at mileposts that may or may not be in the right place.

This man has a historical reputation for dubious opinions and on several occasions made claims that were later discredited. You have take the whole story on balance and make a judgement accordingly, not just take the log at face value.

For me - and I will make clear here, for me - I don’t have the absolute faith in Rous Marten’s data (can we actually call it data? We only have some published versions of his log in secondary sources and no primary record of the event. Is not the other work done by other writers who have produced speed graphs closer to data than this?) nor him and that’s important.

Tom, I have said “for me” throughout that whole spiel you’ve quoted.

I haven’t said “I am definitely right and everyone else is wrong” - I have said “for me”.

Everyone is welcome to disagree with me!

 

I think I've largely said all I have to say on the City of Truro run, and its tedious to repeat it. However if one seeks to discard Rous-Marten's numbers on the basis of power calculations then there are a number of issues.

One is that there are a number of other records of the 6'8" Dean/Churchward 4-4-0s running in the 90s. If one is to discard Rous-Marten's figures then one must either discard all those other records too, or else consider that the mail train run wasn't as exceptional as every contemporary source considers it was. One must also discard the independent timing by the postal worker Kennedy.

Another is that in order to produce theoretical speed based power calculations one must make a considerable number of estimates and assumptions. All those estimates and assumptions are subject to error bars just as much as the stopwatch timings are. So really one should present all the variation possible from those estimates and assumptions, and therefore all the error bars implicit in those calculations.

Occams's razor is of course purely a philosophical tool, and there is no science about it. When constructing hypotheses, however, it may well be useful to test them against it.

 

Exactly. We don’t know the power CoT could produce, nor what is needed for the train, nor precisely what the local gradients and curves were. So just as with timing evidence, you put error bars on your analysis, so you must with a power-based calculation. I suspect those error bars might be quite wide given the lack of data about the drag co-efficient of a GWR clerestory carriage …

I think in an ideal world, you might end up with something like “based on this detailed power analysis, we estimate that CoT might have run at 95mph +/- 6 mph. And based on stopwatch timings, the estimate is it ran at 101mph +/- 4mph”. At which point you have have two entirely independent calculations that are consistent with each other. Whereas as soon as you do a power calculation where the upper bound does not meet the lower bound of the stopwatch calculation, you have some explaining to do.

Tom

 

Hi Jim,

Nothing to disagree with there. And yes, of course we should look at the other locomotives running in the 90s. That I am emphasising there as I think you will understand what I am getting at.

Ultimately a lot of this is going to be theoretical, a lot of assumptions will have to be made, and it will all be in the write up. I haven’t shied away from that, nor have I have tried to present it as anything else.

You will note, I hope, that everything I have said today was couched in the terms “for me”.

The big thing I would like you and Tom to take away from my writings today is that we should at least be consistent in our approach to prove 100mph running as much as we should be consistent in our approach to disprove 126mph.

I’m not really seeing that where I’m being asked for CoT to take the stopwatch timings as read, but not question the machinery or the man behind it, but for Mallard I’m being asked to question the dyno roll, the timings, the machine, the gradients, the mileposts and much more besides.

Where we don’t have physical evidence theoretical modelling gives us an idea of the probability, not an outright result and ultimately that’s all I am pointing towards. And yes, I agree with you, we should always test the modelling.

I hope that makes it clearer, I feel like I have been repeating myself for most of this afternoon.

 

Absolutely, I agree with this. I don’t think I have disagreed with this approach at all? I think this is precisely what I have been saying?
The only difference being - I am saying for myself - it may make me think a particular way. And others will disagree or agree with me, and that’s fine!

Precisely what I am getting at. Thank you.

 

Different individuals will have different preconceived notions that they would like to be correct, but we should all try to be open-minded. A lot more data is available for analysis about Mallard's record run than about CoT's, but both should be subject to the best analysis that can be achieved.
Certainly CoT went faster than anyone was used to, and R-M must have been excited as he gripped his stopwatch. (One stopwatch or two?) We can choose to believe that he did his best and that the error bars on his quarter-mile times are only those that are inevitable from the 1/5 second resolution, human reaction time, and possible errors in the locations of the quarter mile posts. Or we can choose to believe that he got carried away and massaged the raw data. If the latter, his numbers are worthless, though the apparently independent data from the other timer remain to be explained. If R-M's numbers are those that he honestly measured, then the calculated speeds for the whole descent (up to the point when steam was shut off and the brakes applied) are subject to considerably smaller error bars than the peak speed over a single quarter mile and (assuming no sudden and inexplicable changes in the power developed in the cylinders) a best-fit (with error bars) smooth speed curve can be drawn (and of course has been before now). That best-fit curve is surely the best measure that we can ever have of how fast CoT actually ran.
For Mallard, we have the time and distance markers on the paper, and from those can be calculated the apparent speed over each second, each five-second interval, or each quarter mile. (I haven't seen an answer to Hermod's question (post #749) of where the quarter-mile speeds at the bottom of that picture came from.) As with CoT a best fit curve can be drawn, but with many more data points with smaller individual random error bars. Andrews has done that and reported an apparent systematic error with a one-mile period. If genuine that surely must have been caused by something in the transmission between the ninth wheel and the distance-marking mechanism. But Simon sees no such systematic error. Regardless of what any of us would like to believe, that discrepancy in their analyses of the same data definitely deserves explanation.

 

Correct and it was light engine, fresh out of Swindon.

 

I don't think I've quite said that. I think with the mail train run I take the timings as read, but subject those to a whole bunch of caveats and error bars which leads me to the conclusion that I consider an actual speed of 100 - 101 mph most likely, but it's not possible to rule out just under 100.

With Mallard's run, whilst I find Mr Andrews' analysis quite compelling, and worthy of further study, if I have a position at all, because I have never looked into the subject deeply, its that I am basically with Gresley that the data as recorded means that 125 over 5 seconds is well attested, but I have doubts about whether an instant speed of 126 is really valid or whether its just a measurement artifact. And again I am taking the timings as read. What I will say is one wishes to claim the last little increment in speed that Gresley would not claim then one must consider every possible factor in order to make a case.

 

Most of the brake force on a fitted passenger train is from the coaches not the loco so Mallards test train had sufficient brake power, the German test train did not.

 

There are two major reservations wrt COT IMHO I.e. CRM’s timings, 1/5 th second stopwatch and
did COT have the power.
CRM’s successive quarters exit the tunnel were 11.00, 10.6, 10.2 ( on 1/80 gradient )
10.0, 9.8, 9.4. ( on 1/86 gradient ) and 9.2 and 8.8 ( on 1/90 gradient )
Simplistically if one draws a curve recognising the error bar of the stop watch there
is one odd reading i.e the 8.8. but up to 9.2 all appears consistent.

The average speed over the 1/80 was 85 mph (A) over the 1/86 95mph( B) and over the 1/90 100 mph
(C) The loco plus consist 92 + 148 tons I.e. 240 total

HP as a result of the gradient donation ( A) 1520: (B) 1580. (C) 1590
For (A). R in lbs/ton for loco and consist 31 and 17
i.e.IHP = 650 + 670 = 1320
for (B ). = R = 36 and 20.5 lbs/ton
i.e. IHP = = 840 + 770 = 1619
for (C) = R = 41 and 23
I.e. IHP = 1005 + 907 = 1912.
I have deliberately taken a high figure for the consist as AFAIK there is neither detailed knowledge of the
specific stock or its resistance. My figures may be up to 20 % high.

COT had a Swindon No.4 boiler. Based on available performance data for the Cities and the same boiler
on the 43xx 2-6-0s a maximum IHP for 3440 of 1300-1400 appears realistic, but these figures are at
medium speed so allowing for shape of HP/mph curves use 1100

On the 1/80 the HP available for acceleration is 1520 - 1320 + 1100 = 1300
On the 1/86 158o - 1619 + 1100 = 1060
On the 1/90 1590 - 1912 + 1100 = 780.

Potential errors in above : work offered by gravity, gradient profile, train weight and speed.
+/- 2%
Loco resistance +/- 5%
Consist 15% uplift included to mask ignorance.
Overall +/- 10%
therefore on the first 3/4 of a mile available HP for acceleration 1170-1430
on the next 3/4 mile at 1/86 available HP for acceleration 950-1170
on the final 1/2 mile at 1/90 available HP for acceleration 700-860

These figures suggest that power, especially if there was a South Westerly trade
wind was not a limitation.

if we were to assume 98mph was obtained on the 1/90 then possibly
WDBG = 1500-1600
Loco IHP = 937: Consist 890
i.e. it requires 1700-1900 HP to maintain 98mph on the level with
a 92 ton loco plus 148 tons consist. There is available 1100 plus
gravity 1500.

Everything was in COT’s favour, comparatively high HP/ton weight for
the loco, a small tender and a 1/80 to 1/90 gradient. The big unknown is
how much internal loco resistances increase and front end efficiency
decreases at speeds in excess of 90mph. The above calcs suggest there is
considerable wriggle room.

So I think, within the limitations of a 1/5 th second stop watch and
assuming 3440 capable of at least 900IHP at 90mph plus speeds
in the upper nineties, perhaps even possibly 100 mph is feasible.

Not proof but……

Michael Rowe

p.s if you move the event slightly closer toTaunton as per John Heaton in
the RM the final result is not dissimilar.

The late Professor Tuplin upset Mr.C.J.Allen when he asserted that a
A train could maintain 100;mph on a 1/100 gradient ( I think he was
using the Johansen formula but he had a point ) CJA took exception,
perhaps what he should have said is but you still need the power to
get there unless you have a very very long stretch of favourable gradient.

 

I think the forward dome is covering the top feed apparatus, a common feature of overseas designs. It was normal on the German standard classes built during the inter-war period, although dropped from the wartime-built Kriegsloks and absent from post-war Bundesbahn designs. British designers seemed generally averse to having a top-fed dome in addition to a steam-collector dome, but Lawson Billinton was an exception:

https://railway-photography.smugmug...-/LBSCR-LBBillinton-K-Class-2-6-0/i-ppp3js2/A

The cylinder layout of the Class 05, and of the German standard 3-cylinder Pacifics of Classes 01.10 and 03.10, with divided drive and 3 separate valve gears, seems analagous to the later Peppercorn Classes A1 & A2. The earlier (1917 design) Saxon Pacifics were closer to the Gresley model in having unified drive and conjugated valve gear. But although the Saxon engines survived in East Germany until the 1960s, they were a small class and appear to have had no influence on later German designs.

 

Re: Your point no4- the wind tunnel models apparently stacked one upon another. Is it possible that the photo depicts one single wind tunnel model sitting atop a mirror?

 

I think it’s a good starter for ten Michael, but I’d put the horsepower required to achieve 90/100mph much higher.

Horsepower required to overcome air resistance looks like this, according to Bert Spencers work.



But that’s just head on resistance above, that’s not taking into account a real life scenario (so arguably, ideal conditions).

The ease with which an A4 Pacific can do 110+ and did so on so many occasions pre Second World War (and post Second World War) compared to the absolute flogging Scotsman and Papyrus had to achieve 100 and 108mph pre Second World War probably shows best the importance of streamlining in a railway setting.

Now, the above graph is for an unstreamlined Gresley Pacific against a streamlined one. Think about what would be required for a much, much smaller 4-4-0 with many more areas of drag, smaller cylinders, much smaller boiler and much smaller reserves of power, together with a likely very draggy train of parcel/passenger stock behind.

 

No, there’s wires holding them up. It really is quite bizarre looking. It’s on page 25 of the book.

I’ve just realised I can google translate the caption, which says the following:



Never heard of this approach before. Absolutely fascinating!

 

So the reason I said Thompson initially was having the three gears in the same plane (but then remembered that the centre cylinder is set forward, whereas this isn’t) - so in short, I agree with you, it is closer to the Peppercorn engines.

I will go and edit that.

 

Simon,

I have used loco resistance figures ( this includes air, mechanical and rail ) based on established
quadratic equations, which were derived by various people over the past fifty years, primarily
by graph ‘fitting’ of various available data ( eg ex Rugby, Swindon, Doncaster, on the road etc.

Mr Spencer’s work whilst interesting, was arguably placed in perspective wrt non stream -lined
locos by Messrs Ell and Andres. I endeavoured to explain this previously when you referred to
the Spenser curves.

Regarding the consist, I stated that I am not aware of either the nature of the 1904 stock
or the resistance thereof. I have therefore taken known resistance values ( actually with
a7.5mph headwind at 45 degrees ) obtained on mixed stock in the mid 50sand increased
this by 20%.

if anything I suspect my estimates give slightly high figures for the 240 tons
at 90mph on the level.

Simon, to be brutal, I have to say based on your previous failure to understand
the difference between DBHP and IHP ( despite several attempts by me to
explain why I was calculating power outputs ) your continual quoting the
‘Spenser’ graph shows a similar failure to understand the constituents of
loco resistance.

Once again I find myself regretting having posted. Yes of course I recognise
it was no better than an estimate ( the available data offers no more ) but
for someone who apparently does not understand the nomenclature, let alone
the assumptions made, to declare it a “good starter” beggars belief.

Michael Rowe