Railway Series Studies

The rebuilding of the
NWR's number 4

Written by Simon Alexander Martin & Sean O'Connor

In our last essay, subtitled The Gresley A0 Pacific Locomotive and the Origins of the North Western Railway’s Number 4, we found substantial evidence to the claim, that the character Gordon in the Reverend W.Awdry’s The Railway Series books was in fact written to be an unmade Gresley Pacific design for the Great Northern Railway. The last essay dealt with the origins of the design (factual) and the origins of Gordon, if taken to be the lone A0 Pacific (fictional). In this essay, again a collaboration with Sean O’Connor, I intend to show comprehensively the ways in which Gordon was rebuilt in 1939 at Crewe, and why certain aspects of the descriptions in The Island of Sodor (TIOS) and Reading Between the Lines (RBTL), may have been misinterpreted in the past.

Simon Alexander Martin

The A0 Pacific and its fictional life
with the Great Northern Railway

Past the design's original emergence in 1915, and the emergence of the GNR's number 1470, Great Northern, in 1922, there is not much to go on with regards the “what if” that is the Gresley A0 Pacific design.

We cannot say for certain, one hundred percent, that this would have happened then, or later, and in what form, and so on, with regards a locomotive that, to all intents and purposes, was never built, but we can make educated guesses based on the history of the designer of the locomotive, one Herbert Nigel Gresley, and the traditional railway and its engineering practises with regards the GNR and later the LNER. With that in mind, we should at this point restate a few facts we remember from the last paper, with a view to establishing the A0's physical attributes at the time of its fictional disposal in 1923.

Therefore, the A0 Pacific design is replicated here, in its original 1915 design, and with its statistics, as a reminder:

Technical Information

A0 Pacific — As Built (1917-20)

If built, the locomotive would have had some Ivatt and Gresley features, but would not, in 1917, been completely unique. A four cylindered Pacific, The Great Bear, was built for the Great Western Railway (GWR) seven years previous to this design, in 1908. Number 111 for the GWR, the locomotive continued in service until 1924 when it was “rebuilt” as a member of the GWR's “Castle” Class locomotives. The use of four 15in cylinders (remarkably the same as the A0 Design, of seven years later), essentially those of the GWR “Star” class, coupled with a heavier locomotive, using a new boiler which Cecil J.Allen in his book, British Pacific Locomotives, described as unable to “raise steam proportionate to its size”, is an indication of what a Pacific design needs first and foremost: a boiler with the ability to create steam and lots of it.

The comparison between this and the aptly christened A0 Pacific design is thus: one was an enlargement of a successful class of locomotives: the Star Class for the GWR was a very successful 4-6-0 design, however was unsuited to being modified into a Pacific design. For one, while the interchangeability of parts was desirable, the Star components were not up to the task of powering such a heavy machine (97¼ tonnes). As an aside, the enlarging of a locomotive to make a Pacific locomotive was also attempted by Raven with his North Eastern Railway (NER) Pacific design.

In the case of the A0 Pacific, a similar thing can be seen in its stats and design components. The wheelbase is most favourable to a Pacific design: large driving wheels (6ft 8in) with a cartazzi arrangement rear pony truck are two of the crucial components of the original design that Gresley did take forward with his A1 Pacific design. Where the design would fall short of expectations, is in its boiler size (a true GNR design: parallel boiler with small smokebox) and its unfavourable front end arrangement, four cylinders incorporating that used by the large boilered Ivatt Atlantics, with an accentric crank to the middle driving wheel. The speculative statistics produced by the Doncaster drawing office in 1917 – the use of the 170 Psi boiler and the 15in cylinders, would have made a locomotive barely more powerful than the Atlantics it was designed to replace.

It would be in these two crucial areas that the A0 Pacific, if built, would have seen major modifications. The rebuilding of some of the Raven Pacifics, with a Gresley Taper Boiler (and Gresley standard pattern cab), shows how the original A0 design could have been rebuilt with the 180 Psi boiler:

Rebuilt Raven Pacific, #2404 
Photograph attributed to the LNER Encyclopaedia Website,
reproduced for educational use only.

In this instance, the front end arrangement of the Raven Pacifics was left untouched. In the A0 Pacific design, it is unlikely, given Gresley's work with the K3 and 02 locomotives, that the four cylinder layout would have been left untouched. Given that we know from our previous paper that the A0 design is capable of being fitted with a Gresley taper boiler, and a new cylinder block, in a rebuild prior to 1922 and after 1917 we can safely state that the A0 could have been rebuilt with both the 180 Psi boiler and the Gresley/Holcroft conjugated valve gear.

It would produce this locomotive design, a second draft of the A0 Pacific in 1920:

Picture accredited to Simon Alexander Martin,
used with kind permission of its copyright owners.

A brief note – this drawing, and its brother draft, have been replicated from “Indian Ink” cloth drawings. The quality for the scan is excellent and shows where the original A0 drawing was modified to accommodate the proportions of the newer aspects of Gresley's engineering for the GNR.

The design itself differs very little from the original A0 design: it appears at this stage, in reality, Gresley had decided upon the proportions of his Pacific design, including the use of a new taper boiler and smokebox design – this drawing comes from one of two (different) original “Indian Ink” drawings of the further drafts made to the A0 Pacific: its changes above the running plate and below showing where edits were made, prior to what we now know as the final design of 1922, resulting in the A1 Pacific locomotive, Great Northern. The second drawing is replicated later in this paper, however let us examine more closely the design at hand, for the moment.

The most obvious change lies in the boiler: virtually identical to the standard A1 Pacific boiler, and, like the rebuilt Raven Pacifics, incorporating the boiler onto its frames most snugly. Further changes include the Gresley and not Ivatt pattern smokebox, and the cab has been modified slightly to accommodate a single side window. The use of three cylinders is a further change from the original design drawings, though not, the Gresley/Holcroft conjugated motion at this stage. However, it is stated that Gordon has indeed the Gresley three cylinder motion before being sold to the NWR in TIOS; we can then state with conviction that in fictional terms the A0 would have been rebuilt with the 3 cylinder conjugated motion in a further rebuild before the dispatch of the A0 – latterly known as Gordon, to the North Western Railway (NWR).

And this would be where the story of the A0 Pacific before its sale to the NWR would end. It would look startlingly like an A1 Pacific, and mechanically in any event, be virtually identical to a standard Gresley A1 Pacific. However, there is more to discuss.

The description of Gordon in TIOS throws up an interesting set of points, as to the nature of the 1939 rebuild. Here it is, reproduced from the previous essay:

“He was built at Doncaster as an experimental prototype for Mr Nigel Gresley's 4-6-2 Pacific for the Great Northern Railway. Inevitably there were faults that needed correction so Gordon was kept hush hush and apart from test runs was never put into regular traffic or given a GNR number. He was used experimentally till all defects had been cured and the first batch of Pacifics had appeared in 1922/23. In 1923 therefore Gordon was no longer needed and was sold to the NWR along with a spare boiler and firebox.”

“Gordon's present form is interesting. He is a Gresley/Stanier hybrid. Above the running plate he is Gresley, below it he is Stanier. This is the result of a heavy overhaul at Crewe in 1939. Gresley's conjugated valve gear had been giving endless trouble, so Topham Hatt persuaded Mr Stanier to substitute a 2 cylinder chassis of his own devising instead. Between chassis and boiler Gordon's running plate is a law unto itself. Personally we prefer its flowing curves to either Gresley or Stanier angularities.”

We have deliberately sectioned the quotation into two sections to illustrate a point. There is a popular opinion that the second paragraph of the above quotation talks specifically and entirely about Gordon's rebuild in 1939. That is not the case. Only partially does the quotation speak at any length about Gordon's rebuild in 1939, and it refers only to the valve gear, mistakenly labelled as “chassis” when referring to the front end arrangement.

On that point, however, there is president for this description, found in many locomotive manuals around the country, referring specifically to the valve gear used and number of cylinders, and not the frame/wheel arrangement as the full, technical “chassis”. At any rate, Gordon's rebuild involves just the cylinder and valve gear arrangement.

Coming onto the final technical aspect of the design; the only time Gordon's running plate is mentioned is to say, off hand, that it “is a law unto itself”. That might have been true, if it were not for this 3 cylindered Gresley Locomotive:

Gresley Prototype 02 Locomotive, number 461 in 1918.
Photograph accredited to Mike Peirson, and the LNER Encyclopaedia Website.

This engine is the prototype Gresley O2 freight locomotive. It has a 2-8-0 wheel arrangement and a variation on Gresley's 3 cylinder conjugated valve gear (the outer cylinders being to a different design than the production 02 locomotives, known as “Tangos”). Of note is its unusual and possibly unique outer cylinders, and a large Ivatt cab we believe is shared with the original A0 design (and, like the A0 to A1 design here, found to be wanting by engine crews and modified significantly for the production variant O2 locomotives).

It has been a commonly held belief that the running plate which Gordon has is a Sudrian invention, simply because it is described as unique – and some people have gone on record to state that it is a Hatt product, despite neither TIOS nor RBTL stating anything near the like. After extensive study we now feel we can say otherwise. With the prototype O2 locomotive receiving an incredibly similar design, it is not unthinkable – given that Gresley experimented so much with Ivatt Atlantics, his K3 locomotives and even with the illustrated number 461 – that this was a running plate design clearly favoured by Gresley.

That this progressed to the running plate that the character Gordon currently runs with was for a while mere speculation until we looked closely at the “final” draft of the A0 closely:

Picture accredited to Simon Alexander Martin,
used with kind permission of its copyright owners.

This drawing was a variant on the other Indian Ink drawing edit. It is marked for “disposal” in 1941 in the bottom left corner. This clearly shows the design of the locomotive in question, in a transition phase: it was designed with a running plate similar to the O2, while its brother drawing was re-drafted with a running plate closer to that found on the final design: the Gresley A1 Pacific.

It is our belief that through natural and logical design progression, Gresley translated his findings from his other prototype designs to the locomotive now known as Gordon. The fact that such a running plate was used on a Gresley locomotive (which had a long life of nearly thirty years in its original form before being scrapped) is proof enough that Gordon's running plate was never a Sudrian invention so much as an engineering evolution by its original designer.

This design of running plate, being on Gordon as sold, also goes a long way to logically explaining the design of the rebuilt Gordon's running plate, as is seen commonly today. The only problem that stands in our way is that the description of Gordon's plate as “unique” would suggest that a plate having been used on now two prototype Gresley designs, is not quite unique. This makes them contradictory to what we are explicitly told – or does it?

For the Reverend W.Awdry, at the time of writing the TIOS volume, the running plate which Gordon carries would indeed have been unique to Gordon (the book was published in 1987, forty years after 461 was scrapped). The unique nature of plate is retained: it remains a law unto itself for the precise reason that it is unique. It is not, however, a Hatt design, but a Gresley feature on an experimental locomotive.

In that respect, the engine known from 1923 as Gordon would be a very different engine to the locomotive that we would surmise was built in 1917 to the original plans. The major changes to the A0 locomotive would have taken place in 1920. The modifications would have been primarily centred around the replacement of the boiler, smokebox, and front end cylinder arrangement, as per the 1920 plans, and it is also not impossible to imagine that the standard Gresley pattern cab would have also been trialled on Gordon as it was with other prototype locomotives such as the K3s.

The choice of one running plate over the other is simple: the O2's running plate was in service for nearly two years at the point of “build” for Gordon the A0 Pacific, and may therefore have been implemented as a matter of course, before the final design – the production A1s – started to emerge from Doncaster in 1922. In essence, and very much like the prototype O2 and its production variants, the O2/1s, Gordon would have been identical mechanically and for the most aesthetic part, to the production A1 Pacifics when he was eventually sold (as described in IOS and RBTL) in 1923. That is only half the story, then, of the locomotive named Gordon.

Gordon and the 1939 Rebuild at Crewe

These are the facts we know about the lone A0 Pacific, after its purchase by the NWR in 1923. The locomotive gave around seventeen years service on NWR metals. However a locomotive crisis (described at length in TIOS and RBTL) at the time of purchase gave little room for regular maintenance and as such wear and tear began to show on the locomotive, nowhere moreso than the temperamental Gresley Conjugated valve gear.

As on all the Gresley locomotives built with the conjugated valve gear, it began to give trouble with mistiming valves and this if not properly maintained could have disastorous consequences. It is not exactly known what specifically called for a complete overhaul of the locomotive and the complete rebuild of its front end, but we feel that, as in the TIOS description, this is due in no small part to problems found with the conjugated motion.

To this end, the A0 – by now, NWR #4, was sent to the London, Midland & Scottish (LMS) works at Crewe. According to both Gordon's and Henry's descriptions in TIOS and RBTL, Sir Topham Hatt had a long standing friendship with Sir William Stanier and in a quiet favour was able to get the loco rebuilt to a standard suitable for the needs of the NWR.

Herein lies some confusion about what exactly the nature of the rebuild is as described in TIOS:

“Gordon's present form is interesting. He is a Gresley/Stanier hybrid. Above the running plate he is Gresley, below it he is Stanier. This is the result of a heavy overhaul at Crewe in 1939. Gresley's conjugated valve gear had been giving endless trouble, so Topham Hatt persuaded Mr Stanier to substitute a 2 cylinder chassisof his own devising instead.”

The important word here is chassis. This is a common misconception made by even the most professional authors on locos. Most people would take chassis to mean the frames, wheels and cylinders. What the word chassis has been used to mean here, as is quite common, is the front end arrangement (cylinders and valve gear). It would not be economical, as has been suggested previously, to completely replace the frames/wheels/cartazzi arrangement/cylinders/ and front bogie. Why? Engineers would not rebuild a locomotive this drastically. This sort of rebuild (coupled with the removal of the right wheel tender!) would qualify as building a new engine from scratch in all cases. If the frames, cylinders, wheels and bogie arrangements are beyond repair, it would be cheaper and more desirable for Topham Hatt to buy a new engine instead of Gordon. That is the not the case described. A decent real life counterpart in the rebuilding stakes would be the LMS' “Patriot” class of locomotives – their original frames, wheels, and cylinders were utilised in building the new Patriots from their original “Claughton” forms. While it was economical (and desirable) to replace the boilers, cabs and later cylinders, the frames, driving wheels and front bogie were left virtually untouched from the original design.

Taking that into consideration, the front end arrangement of Gordon was replaced with a 2 cylinder design (as described in TIOS), which we have concluded would be a new and unique casting since no suitable LMS donors were available (most Stanier 2 cylinder machines were mixed traffic or similar and were not built to the same tolerances required in an express passenger locomotive). These new cylinders would be bored out to 20x28in. This is deliberately similar to those seen in later years on the “Brittania” class locomotives, an class of locomotive which could be described as Gordon's a modern day counterpart. The Brittania class were two cylinder Pacifics, whose cylinders were designed to operate at 250psi as opposed to the much lower 180psi of the standard Gresley A1 boiler.

However, we feel that in addition to this modification in the cylinder area, the original Gresley taper boiler would be overhauled and rebuilt to operate at 220psi, giving a tractive effort similar in size to a 220psi standard Gresley A1 (with 3 cylinders).

This modification to the original A1 boiler had been done in real life on two Gresley A1 Pacifics, whilst still retaining the distinctive round dome. These locomotives and their case histories are explained at length by Cecil J.Allen in his book British Pacific Locomotives. The locomotives in question were 4480 Enterprise & 2544 Lemberg. Further, both these engines would retain their standard Gresley A1 look until rebuilt as A3 Pacifics some ten years later. It is interesting to note that an almost original Gresley A1 Pacific continued in service with the LNER and latterly British Railways until 1949! Again, Cecil J.Allen recounts this in his book: the locomotive in question being the last modified, 2567 Sir Visto.

The original modifications being the case in 1927, it does not seem unreasonable to suggest that a similar overhaul would be conducted by Stanier in 1939. In actuality, this kind of modification is essential for making the locomotive reliable and powerful.

This overhaul made Gordon a much better locomotive in maintenance terms for the NWR. A 2 cylinder simple does not require inspection chests (no third cylinder), oil reservoirs are easier to access, and the locomotive itself in service would be not dissimilar to the Brittanias in many ways – high tractive effort for its relative size, with a few advantages over its British Railways counterpart.

The Brittania boiler worked at a higher pressure, but required more work to boil water efficiently, and especially at the firing end of the equation. The wide firebox of the Gresley taper boiler, coupled with its round top design, has proven time and again to be an incredibly efficient steam raiser, and it is testament to Gresley's original design that its attributes were perpetuated on both Thompson and Peppercorn derivatives. It is safe to assume that the combination of a reliable front end arrangement and a superior boiler, would make the rebuilt Gordon a machine that the NWR could be rightly proud of. The tractive effort would be lower than the Brittania, but (as has been shown) the balancing of a Gresley locomotive with the taper boiler would produce a machine with better adhesive weight – good for starting and maintaining trains at high speed.

The 2 cylinder design adopted for Gordon, and described here, would allow Topham Hatt to substitute a measure of speed, not needed on a sixty mile main line, for an increase in adhesive weight which is what he needed most of all.

The outside valve gear, originally Gresley Walschearts, would be replaced with a form of the Stephenson link design, a design Stanier was rather fond of for his 2 cylinder machines (however we emphasise NOT simply lifted from an existing design – like the cylinders, most likely a unique design to fit the requirements of the locomotive).

There would also be a new form of motion bracket - similar again in style to the Stanier machines, but by no means lifted from such designs. The actual details of both the valve gear and motion bracket are frustratingly lacking. We may only offer the view that designs close to the Stanier ideals would be in keeping with Gordon's design and of course, the practicalities of locomotive engineering.

Both of these form the new valve gear which RBTL states was a “Hatt” design. TIOS actually states that “... Topham Hatt persuaded Mr Stanier to substitute a 2 cylinder chassis of his own devising instead”. This has been misinterpreted in RBTL. Given English Grammar, the subject of the sentence is in fact due to the phrase “persuaded Mr Stanier”, making it Stanier's Chassis, and not Hatt's that is described (We have, however, for the purposes of the comparison table, labeled it as “Hatt/Stanier” to appease everyone. In reality since information on the new type of valve gear is limited to descriptions in TIOS and RBTL, the valve gear's designer is of little relevance in comparison to the job it has to do. In that respect, the design of the valve gear may be interpreted through illustrative purposes as opposed textual).

Aesthetically Gordon would remain in the 1939 rebuild, outwardly close to its 1923 form in other respects. The running plate would retain the curved Gresley design. The Gresley cab being a very successful design would also be retained.

One final change would come in the shape of the tender. The standard 8 wheel LNER tender would be replaced with a 6 wheel Fowler tender: a tender which has become the standard on Sodor.

It is not specifically known why this was done, but a common belief is that it was to reduce the overall length of the locomotive, as per the requirements of the NWR's turntables, and other such equipment. Indeed, when the 8 wheel tenders were introduced on the LNER, the total wheelbase length of the Gresley A1s was 60ft 10in, and therefore could not be turned on anything less than a 65ft long turntable (for which, the LNER had to build new and larger turntables).

All these changes put together in the 1939 rebuild produced the fine express locomotive known as Gordon, which can still be seen giving good service on NWR metals to this day.

N.B For the interests of comparison, we have included at the end of the paper a technical comparison table, which states the main differences between the 1917, 1920 and 1939 forms of the Gresley A0 Pacific as “built”, “rebuilt” and then “rebuilt at Crewe“.


With dedicated research, careful observation and working through the logical development of a real, factual design, we believe that Gordon's 1939 rebuild can be explained by logical development of real life engineering. As with the first paper we feel we can state that the A0 Pacific design of 1915 is Gordon as the Reverend W. Awdry first intended, and not only that, but that way in which the locomotive was rebuilt was wholly more logical and less drastic than has been suggested by any misinterpretations that have gone before.

Simon Alexander Martin & Sean O'Connor (Sept. 2009)

Technical Comparison

* Ivatt Pattern to a Gresley Design.
** Curved Running Plate found on Gresley O2 Prototype, number 461.


Allen, Cecil J, British Pacific Locomotives (London: Ian Allen Ltd, 3rd Edition 1990, originally published 1962).

Allen, Cecil J, Locomotive Practice and Performance in the Twentieth Century (Cambridge: W. Heffer & Sons Ltd, 2nd edition 1950).

Awdry, Christopher, Sodor: Reading Betwen the Lines (Oundle: Sodor Enterprises, 2005).

Awdry, Rev. Wilbert; Awdry, George, The Island of Sodor - Its People, History and Railways (London: Kaye and Ward, 1987).

Glover, Graham, British Locomotive Design (London: George Allen and Unwin Ltd, 1967).


Special Thanks and Acknowledgements

We wish to express our thanks and acknowledgements to the National Railway Museum, York, for the use of the technical drawings from their archives.

We would like to thank in particular the NRM's former Curator of Railway Vehicles, Jim Rees, for his help in the research stage, and wish him well at Beamish.

We would also like to thank the following people for their helpful critiques and support: Ryan Hagan, and Gavin Rose.

We'd like to thank in particular Martin Clutterbuck for hosting our research on his wonderful site. Thanks Martin!