Edward Thompson: Wartime C.M.E. Discussion

Hi Jon,

I touch on this in the limitations section of the book. In fact it looks like Springbok was finished on 19 Dec 1942 - the year end figures are as above.

The question - put simply - is how is that 40,000 miles possible?

We’re talking 3 weeks to do all that mileage that some classes didn’t manage in a year.

We potentially have an error in the original data, and it should read “4000” or “400” miles.

Which is my personal view. Even 4000 miles looks quite significant given the number of days left in 1942 to do work!

This isn’t the finished graph by any means but I’ve got to be careful in how I use the data - if you show what is recorded and then clarify the potential errors, you are showing the primary evidence as it actually is, and not what you think it is.

 

Setting aside the fact that the appeal of most books on railways are aimed at a small audience, my understanding (I haven't, admittedly, read all this thread) is that the aim of this book is to present the thesis that Thomson and his work have been much misrepresented over the years. I'm not sure how you expect @S.A.C. Martin to do this without including 'masses of figures and graphs'. Surely, without data, the reader will not be able to test the validity or otherwise of Simon's arguments.

 

So that has to go in discussion and interpretation along with a whole host of other things. Data is nothing without a good story.

I guess what you might be able to show is that the B1s went to the top of the mileage and availability league table in their class, as you would hope and expect with any new design, stayed on top line work doing 200 miles/day for the first ten years of their life, then were gradually cascaded.

Obviously the existence of the B1s may have affected the mileage of the B12s and B17s not to mention the scrappage dates of the GN 4-4-0s etc. Maybe there is room for discussion of the range of work the B1s covered in the first half of their lives and what locos they displaced in the fleet. In some ways that is more interesting than the equivalent for the A2s and L1s because of the very wide range of locations and duties they covered.

 

Indeed, I heard a very interesting podcast yesterday from "More or Less" entitled "Maths without numbers", which focuses on the point that telling the story matters more than the numbers.

 

I was taught a long, long time ago not to include data either as a table or a graph in a report without putting it into context and explaining what message you wanted the reader to draw from it.

 

Ralph - let's be fair. You were never interested in buying a copy of the book.

 

Success - I have found the error. I went back to the original document.



So the mileage to year end was in fact 701 - that changes the graphs quite substantially:





This makes far, far more sense. When going through the data and recording it in the spreadsheet originally I must have used the original B1 (Robinson) data and spliced it with the class B thereafter. No harm done. I will check everything else in the same way now, of course.

 

Remarkably catty from someone meant to be a moderator.

 

Surely, as has been discussed many time before, at the end of the day a moderator is just an ordinary forum member who has as much right as any member to express their thoughts and opinions.
Oh, and thank god the moderators have agreed to take on the responsibility (unpaid) because without them we'd be stuffed.

 

If Mr Thompson had been able to visit Denmark in the 1940s, I wonder what opinion he would have of the Pacifics that your country purchased from Sweden?

https://en.wikipedia.org/wiki/SJ_F_(steam_locomotive)

As they had 4 cylinders driving on the middle coupled axle, nearest British equivalent would have been the 4-cylinder Pacific that Mr Gresley designed, but did not build for the Great Northern Rly in 1915.

 

Right to express an opinion doesn't give one an automatic right to be rude.

 

A quibble maybe, but as one who suffers from colour blindness, I find the ones used - especially for B1, B7 and B17 - are rather indistinguishable. Is it possible to leave some gaps in the lines and/or perhaps put some dashes in to make them clearer, please?

 

Yes, happy to make such amendments.

 

He would have termed them pure scrap and they were bougth from Sweden for scrap price.They had compound cylinders with the ugliest crossed ports that could be cast.
Constructed 1914 and state of art.
Chapelon did the ugly act 1929 and we bought them being surplus after electrification 1937.
After occupation 1940 we needed more locomotives and asked the germans for steel and steel casings and built 25 more to the original swedish drawings no changes allowed.
I believe that Thompson would have felt even more contempt for Danish state Railway management than LNER/Gresley.
The A4 cylinders were formidable and post Chapelon,and DSB build to prewar(one) drawings?.We have to import all coal.

WP_20180926_006[1] by Hermod posted Sep 26, 2018 at 2:30 PM

 

Apologies for - ah - "Spamming" - !!! - but I just wanted to come back to this as it's an incredibly important point where the B1 is concerned (thank you for raising it).

So our dataset is vast, covers 5 years worth of data, across 6500 locomotive classes, which vary in wheelbase, size, power, etc etc and then of course, the whole dataset is in fact split between sections of the LNER, as shown below:



Now happily we already have total stats (WHOLE LINE means the whole of the LNER - as you can see above, 21+11+4 gives 36 LNER A1 (A10) Pacifics) which means we already have average mileages plus averages for the days at shops/sheds/in use/not in use, etc.

The availability formula however works as follows:



We developed this formula in this very thread a couple of years back, in fact. It looks good on paper - and it is good - for showing an average availability where a class numbers 2 examples or more, and the classes' numbers remain consistent.

However - whole line averages are what we are looking at. We can with a more specific study go into individual areas in more detail - but the nationwide averages help to identify trends in classes I feel.

Where we have a further issue - and I identify this in the book - is that where rebuilding's are concerned (see P2s), or where classes go out of existence or conversely, into existence - you may get a no. of locos to year's end being lower or higher than expected and their mileages/days may not be recorded in the same way: so the numbers can get skewed in a number of ways.

Case in point - Thane of Fife comes out of the 1940s very chuffed with itself: it's the only A2/2 which we can identify almost every year on its own, therefore its work is its own and it is excellent.

However...for the same years, some of the A2 data has more than one A2 sub class in it. So for example - in 1946, Thane of Fife has its own row, the other A2s (Five A2/2s, four A2/1s and two A2/3s) are grouped together into different areas of the country, and we can't differentiate between them, which would be useful.

We also have an issue whereby the formula assumes a locomotive has available to it approx 311 days a year (because each year, for various reasons, England and Scotland's working days change in total), whereas the reality is - if a locomotive is newly built in 1942 (e.g. B1) and in fact has only 12 days in which to do work - then the formula under-represents its actual availability figure in the spreadsheet.

So for the B1, the formula looks like this:

4 / 311 = 1%

Where if we adjusted for the number of working days it - actually - had available to it, when released to traffic:

4 / 12 = 33%

Which provides an obvious increase in activity.

So the big question is: what is fairest? Do I adjust the formula to take all of this into account? And how far do I take it? If we do it for all the new locomotives - then we're treating them fairly, so far as being able to do work is concerned. That being the case - we have to do it on a case by case basis.

Which means - for example - that Great Northern, currently with 46% in 1945, due to being released to traffic in September 1945, would actually have an availability of 87% when we adjust her formula accordingly for the days she could physically work after overhaul.

For the record, I am looking to adjust the figures on that basis for all of the Thompson designs concerned, in order to be consistent. I will make that clear in the book.

 

If you normalise, you need to do so across the patch - which also requires that for withdrawals, you adjust for the shortening of the year occasioned by withdrawal.

My question to you would be to ask whether this makes a material difference to the overall story, or is a statistical quirk that needs to be noted and understood, but is of limited real world importance. Approaching this as a historian seeking to explain policy as it was being made at the time, I would be inclined to stick with the figures that were being used to help manage the LNER, on the basis that they most nearly represent the understanding of those making decisions at the time.

 

A really good point. I need to think about this in the overall presentation of the book.

 

Despite his intemperate style, I also suggest you take into account the opinions of those like Ralph who aren't huge fans of statistics when considering this approach - you're writing a work of railway history, not a submission to the society of actuaries!