By Malcolm Peirson

From the beginning of the railway age locomotive engineers were concerned with increasing the power and efficiency of steam locomotives. In the late nineteenth century and the beginning of the twentieth century two of the methods developed to achieve this were the use of compounding to achieve thermal efficiency, and boosters to enable more power. Although the compound locomotives of the London & North Western Railway (LNWR), along with those of the Midland Railway, are probably the best remembered and received the most limelight, it should be noted that compound locomotives were developed on the Great Eastern Railway (GER), the North British Railway (NBR), the Great Northern Railway (GNR), the Great Central Railway (GCR), and the North Eastern Railway (NER). In fact, the NER made use of compounds for a longer period and with more consistent success than any other pre-grouping British railway.

The following sections of this article are an attempt to explain the theory behind compounding, and to describe the development of the former on the pre-grouping constituents of the LNER, and the latter on the LNER itself.

Compound Steam Engines

In a simple (or 'single-expansion') steam engine, the high-pressure steam enters the cylinder through a cut-off valve and moves the piston down the cylinder. The first problem occurs because full boiler pressure cannot be applied to the face of the piston throughout its entire pressure stroke. If this were done then steam would be wasted, as at the end of the stroke the crank pin is directly behind the driving axle. Pushing on the crank pin in this position does nothing to rotate the drive wheel and all the force does is try to bend the connecting rod and shear the crank pin. The cut-off valve is used to cut the intake of the steam early and save this steam. The piston moves down the cylinder, and when it is at about 25%-33% of its stroke, the cut-off valve shuts and the steam expands, pushing the piston to the end of its stroke. The exhaust valve then opens and expels the depleted steam to the atmosphere (Fig. 1 shows a schematic of a cylinder). The second problem is the fact that, as steam expands in a high-pressure engine, its temperature drops and because no heat is released from the system, steam enters the cylinder at a high temperature and leaves at a low temperature (this is known as adiabatic expansion). This in turn causes a cycle of heating and cooling of the cylinder with every stroke, which is a source of inefficiency.

Fig. 1, Simple Expansion Cylinder (CC-BY-SA2.5,2.0,1.0 GNU)

In 1805 a British engineer by the name of Arthur Woolf patented a method to lessen the magnitude of this heating and cooling with what was called the Woolf High Pressure Compound Engine. In the compound engine, high-pressure steam from the boiler first expands in one or two high-pressure (HP) cylinders and then enters one or more subsequent low-pressure (LP) cylinders. The complete expansion of the steam occurs across multiple cylinders and, as there is less expansion in each cylinder, less heat is lost by the steam in each. This reduces the magnitude of cylinder heating and cooling, increasing the efficiency of the engine. For railway locomotive applications the main benefit sought from compounding is economy in fuel and water consumption plus high power/weight ratio due to temperature and pressure drop taking place over a longer cycle. Additional perceived advantages include more even torque and in many cases superior riding qualities with consequent less wear on the track.

Thus, the cylinders on a compound steam engine can be said to work in 'series' as opposed to the normal arrangement of a 'simple' engine where they work in 'parallel'. However, in order to balance piston thrusts of a compound, the HP:LP cylinder volume ratio has to be carefully determined, usually by increasing the LP cylinder diameter and/or by increasing the length of the piston stroke. In non-condensing engines the HP:LP volume ratio is 1:2.25. The correct ratio between the HP and the LP cylinders has always been one the important problems confronting designers. As for the arrangement of the cylinders, particularly applied to British locomotives, in a compound locomotive, they can be arranged most commonly two ways:

  • Cross-compound (2, 3, or 4 cylinders) - the cylinders are side-by-side.
  • Tandem compound (2 cylinders) - the cylinders are end-to-end, driving a common connecting rod.

The eternal problem, however, with compounds is starting. For all cylinders to take their weight, it is advisable to have some way of short-circuiting the HP cylinders and getting steam at a reduced pressure directly to the LP cylinder(s); hence many of the patented compounding systems are associated with starting arrangements. There were various configurations and engineers associated with them but these would need an article in themselves, and for this reason this article will deal with those associated with the compound locomotives of the LNER constituents.

Pre-Grouping Compound Railway Locomotives

In 1850, the first recorded compound railway locomotive was constructed for the GER, then the Eastern Counties. Although the patent (13029) was taken out in the name of James Samuel, the company's engineer, the idea appears to have come from one John Nicholson, a driver on the line. Nicholson's system, which was employed on two locomotives, converted from "simple" working, differed from the general principle subsequently adopted in locomotive compounding in that it made use of what was termed 'continuous expansion', the steam being finally exhausted in both the high and low pressure cylinders. In this system, the two cylinders alternated as high and low pressure, with the changeover occurring half way through each stroke. Two locomotives, one passenger and one goods, were converted to the system but no further examples followed. One of these engines had two cylinders of equal size, while in the other the area of the low- pressure piston was approximately 2.3 times that of the high-pressure pistons.

However, the real history of locomotive compounding does not begin until 1876, when the French engineer, A. Mallet introduced a compound locomotive on the Bayonne - Biarritz Railway. Mallet's work inspired August Friedrich Wilhelm von Borries, a German engineer whose work on compound locos was done in collaboration with the British engineer Thomas William Worsdell, and together these two men obtained several British patents.

For ten years T.W. Worsdell had worked as Works Manager at Crewe and witnessed Webb's work on the LNWR with compounding. Webb had applied compounding to everything from small tank engines to 0-8-0 four-cylinder freight engines, but had achieved little success owing to a combination of badly proportioned cylinders and inefficient valve gear (lack of coupling rods on the earlier express locomotives was also a hindrance). When T.W. Worsdell took over as Locomotive Superintendent on the GER in 1881 he decided to try compounding himself, the system that he devised using only two cylinders. The T.W. Worsdell system was very similar to that being developed concurrently by von Borries, the principal differences being in the starting valve arrangements. Accordingly, the system was jointly patented by them. The characteristic of this system was that it made use of only two cylinders, and was thus especially suitable when inside cylinders were used.

Fig. 2, GER No. 230 (M.Peirson)

In 1884 T.W. Worsdell built an experimental 2-cylinder compound 4-4-0 locomotive for the GER, No. 230 (Fig. 2). It was identical to a previous class 2-4-0 (G14), except that it was fitted with a leading bogie to provide room for the larger low-pressure cylinder and - like most compound engines - the boiler pressure was higher than normal, at 160 lbs. psi. As compounds, the high pressure cylinder was 18 in. x 24 in., and the low pressure cylinder, 26 in. x 24 in. The cylinders had a common stroke of 24 inches, with the valves on top, driven by Joy valve gear, as on the G14s. Driving wheel diameters were 7 ft. with bogie wheels of 3 ft. 1 in. A grate area of 17.3 sq. ft. gave a heating surface of 1,200 sq. ft. The weight of the engine was 43 tons 15 cwt, with tender weighing (full) 32 tons 8 cwt (2,755 gallons). The G16s had a total wheelbase of 42 ft. 8.5 in., and the length over the buffers was 50 ft. 10.5 in. The prototype engine was ordered to 'letter account' G16, which became the class designation in the usual way. After a period of testing and refinement, ten more locomotives were ordered, numbered 700-709, although only one of them had been delivered when T.W. Worsdell resigned to join the NER, where he went on to perfect the system further. In service, the G16s were 'good pullers', but tended to become sluggish when the motion was notched up. Overall, they showed a reduction of 14% in coal consumption compared with the G14 2-4-0s. However, against this economy had to be set the greater maintenance costs of the higher pressure boilers. When James Holden reduced the pressure to 150 lbs the fuel economy advantage fell to only two percent. Thus, in 1892 the G16s were regarded as surplus to requirements and Nos. 700-709 were placed on the 'duplicate list' as 0700-0709. However, at the same time, No. 230 was rebuilt with G14-type simple-expansion cylinders, and the remainder were similarly altered by the end of the year. Two years later, all eleven engines were given new boilers, and they continued to give useful service until withdrawn in 1902-1904.

Fig. 3, Rebuilt NBR No. 223 (E.Cameron and A.Rogers)

In 1885, Matthew Holmes of the NBR rebuilt the Wheatley 4-4-0 number 224 (famous as the locomotive involved in the Tay Bridge disaster of 1879) using a system of tandem compounding patented by a relative of his, W. H. Nisbet (Fig. 3). A tandem compound has each pair of high and low pressure cylinders driving a common crosshead, connecting rod and crank, but unlike other compounds, the cylinders are mounted fore and aft of each other (Fig. 4 shows a simplified diagram of a tandem compound system). The rear wall of the forward cylinder is usually the forward wall of the rear cylinder. The piston rod of the rear cylinder is connected to the crosshead in the usual way, but the forward cylinder may have its piston rod, or rods, in either of two forms. Either the piston rod of the rear cylinder is extended forwards to also carry the forward piston, or, if the forward cylinder be the low pressure cylinder (and thus larger in diameter than the high pressure cylinder behind), it may have two long piston rods which pass above and below, or to either side, of the high pressure cylinder in order to reach the common crosshead. The Nisbet system used two high pressure cylinders of 13 in. by 24 in. fitted well to the front of the smokebox. These shared a piston rod with the 20 in. by 24 in. low pressure cylinders set in the conventional position below the firebox with the shared piston rod acting on a single crosshead and connecting rod. From the connecting rod two parallel links reached up to two parallel sets of Joy valve gear; one on each side, the set of valve gear inboard of the connecting rods being linked to the low pressure cylinders and the set outboard being linked to the high pressure cylinders. Both sets of Joy valve gear were linked to two reversing levers side-by-side in the cab. The point of the system seems to have been to allow the driver to use high-pressure steam expansively in the small high-pressure cylinders while allowing a relatively late cut-off in the low-pressure, larger cylinders sharing the same piston rod. In other words it would have allowed the driver to notch up the high-pressure cylinders, while leaving the low-pressure cylinders in something nearer to full gear. This would have overcome the phenomenon known as 'throttling' where compounds tended to become less free-running as the speed increased. A last interesting point was that a small by-pass valve in the smokebox side, controlled by a rod on the left side of the boiler, admitted live steam to the low pressure cylinders at starting, allowing condensate to be expelled as the locomotive moved off. To allow for the upward angular motion of the Joy gear valve rod, the 4 ft. 0 in. boiler was pitched high, at 7 ft. 5.5 in. above rail level. The tandem compound experiment was tried uniquely on No. 224 for two years from 1885 and was reported as a success at the time, but by the end of 1887 the locomotive was rebuilt as a simple by removing the front cylinders.

Fig. 4, Tandem compound (M.Peirson)

1885 also saw T.W. Worsdell appointed to the position of Locomotive Superintendent of the NER, and he at once began turning out two-cylinder compounds that excelled in the very feature that was so deficient in the G16s of the GER - they were supremely fast. The first compound locomotive was completed at Gateshead in the autumn of 1886, and was a two-cylinder design using the Worsdell-von Borries system (Fig. 5). The locomotive was an express goods engine, the first of the celebrated "C" Class of 0-6-0 (LNER J21) with a high pressure cylinder of 18 in. diameter and a low pressure cylinder of 26 in. diameter, with the stroke of both being 24 in (Fig. 6). In the Worsdell-von Borries system of two cylinder compounds, exhaust from the high pressure cylinder was fed through a large pipe in the smokebox that acted as a receiver before being used in the low pressure cylinder. One potential problem was starting when the high pressure cylinder was on dead centre, but that was solved with a driver-operated valve that allowed steam to pass directly into the low pressure cylinder. The coupled wheels of the locomotive were 5 ft. 1.25 in and Joy valve gear was used, actuating slide valves with 1.125 in. lap, 0.1875 in. lead, and a travel of 5.5 in. in full gear. The boiler had a tube heating surface of 1,026 sq. ft.; the firebox had a grate area of 17.23 sq. ft. and a heating surface of 110 sq. ft. The boiler pressure was 160 lb. per sq. in. Only one compound goods engine was built in 1886, and while trials were conducted ten simple 0-6-0s, Class C1, were built. These had two cylinders 18 in. by 24 in. but were otherwise the same as the compound. After testing had proved the success of the design a further 171 compound versions were constructed between 1887 and 1892.

Fig. 5, Worsdell-von Borries System (M.Peirson)

When Wilson Worsdell took over from his brother in 1890 he showed a dislike of compounding and Joy valve gear, consequently when a further batch of the "C" Class was built in 1894 - 5 they were non-compound. Gradually all the compounds were rebuilt as simples with only 31 still running as compounds in 1910, and the last one being rebuilt in 1913. Now classified as just "C", the LNER inherited these fine locomotives and gave them the designation J21 and many survived (as simples) into Nationalization (1948), the last being withdrawn in 1962.

Fig. 6, Class C 0-6-0 (author's collection)

At the same time as the construction of the "C" Class, T.W. Worsdell built a tank engine equivalent, of which the boiler, wheels, cylinders and motion were identical to those of the "C" Class. These tank engines were of the 0-6-2 type; 51 designated Class "B", were built as compounds, and 11, Class "B1", as simples. During trials, the compound locos proved superior to the simple versions, although this dropped significantly for short runs and shunting (Fig. 7 shows No. 855 ex-works & Fig. 8 shows No. 1319 at Eston). They were used on the shorter-distance mineral workings and, after rebuilding as simples by W. Worsdell between 1902 and 1912, they could be seen on passenger workings. After Grouping they were designated N8 by the LNER, the last being withdrawn by British Railways in 1956.

Fig. 7, Class B No. 855 ex-works (author's collection) Fig. 8, Class B No. 1319 at Eston (author's collection)

Due to the importance of freight traffic on the NER, T.W. Worsdell focused on quantity-production of the 0-6-0s as a standard before work on passenger engines had passed an experimental stage, the first of which emerged from Gateshead works in 1886 as No. 1324 ("D" Class). This was a 2-4-0 with coupled wheels of 6 ft. 8in. diameter (Fig. 9), but while the cylinders and motion were the same as those of the Class "C" goods, the boiler provided a larger tube and firebox heating surface, with a total of 1,323 sq. ft., as opposed to the 1,136 sq. ft. of the "C" Class. 1324 had a boiler pressure of 175 lb. and was a fast and free runner, however, the locomotive proved to be unsteady at high speed, probably due to the cylinders doing an unequal amount of work and thereby setting up lateral forces. Experiments with 1324 led to the development in 1887 of the "F" Class 4-4-0 (LNER D22), of which two batches were built at Gateshead; ten were compounds ("F" Class) with cylinders the same as the "C" Class, and ten simples ("F1" Class) had two cylinders 18 in. dia. by 24 in. stroke (Fig. 10 shows No. 779 at Edinburgh Waverley). There was some show of comparing the compounds with the simples, but as the compounds were provided with 175 lb. boiler pressure and the simples with 140 lb., the results were obvious.

Fig. 9, NER Class D (author's collection) Fig. 10, NER No. 779 at Edinburgh Waverley (author's collection)

To obtain a smooth action and free-running it was essential for the work done in the two cylinders to be as nearly equal as possible. To do this T.W. Worsdell calculated that the cylinder volume of low pressure to high pressure should be about 2.3 to 1, but this would have resulted in a larger low pressure cylinder than could be conveniently accommodated. T.W. Worsdell got around this by using cylinder diameters of 18 in. and 26 in. (same as his GER compounds), but on the NER engines the Joy valve gear was arranged to give a longer cut-off in the low pressure cylinder to compensate for the low pressure cylinder having to be smaller than was theoretically correct. Typical figures were: 50 per cent. H.P. and 73 per cent. L.P.; 70 per cent. H.P. and 84 per cent. L.P. As an aside here: it was noted that despite their free running the NER compounds had a harsh and noisy beat. However, it was with one of the "F" Class compounds in the race to Edinburgh in 1888 that Bob Nicholson made what was then the record time from Newcastle to Edinburgh - 124.4 miles in 126 min. with a load of 100 tons. The compounds of this class were converted to simples in 1911 and the class became NER Class D22, the last of which being withdrawn in 1935.

Fig. 11, NER Class D 2-4-0 No. 340 (author's collection)

1888 also saw a significant event for not just the NER, but also for British locomotive development in general. Walter M. Smith, the Chief Draughtsman for the NER, turned his thoughts to piston valves as a means of avoiding engine failures due to broken slide valves, and set out to produce a valve that would be free in operation, but would provide a means of allowing trapped water to escape without damaging the piston. These were first tried on a second engine of the "D" Class 2-4-0 compound, No. 340 (Fig. 11), which was distinguishable from No. 1324 by virtue of a long extended front to cover the tail rods of the pistons. No. 340 had 2 in. spring loaded relief valves fitted at either end of each cylinder for the purpose of allowing any trapped water to escape; but it was found that in practice sometimes, with less experienced drivers, the relief valves did not open quickly enough and cylinder covers were broken, and so Smith designed an improved type of valve in which relief from excessive pressure was obtained by use of segmental piston rings. Both "D" Class 2-4-0 compounds, 1324 and 340 were modified with bogies to conform to the "F" Class locomotives (although they were in effect a bogie version of their original builds). Smith had at one time served under S. W. Johnson,/a>, on the Edinburgh and Glasgow Railway, and this led to collaboration between Derby and Gateshead in the development of Smith's piston valves. It may also have provided the inspiration for the next class of compound express locomotive on the NER, also in 1888.

Fig. 12, NER Class I No. 1329 at Doncaster (author's collection)

New motive power was needed for the Leeds and Scarborough expresses, and because the road was easy and the trains light it was decided that a well-designed 'single' would be ideal. The "I" Class engines (Fig. 12 shows No. 1329 at Doncaster), with driving wheels 7 ft. in diameter had the standard NER compound front end, with high pressure cylinders 18 in. diameter, low pressure 26 in. dia. and both having a stroke of 24 in.; the boiler was the same as that of the "B" and "C" Classes, and the pressure was 175 lb. per sq. in. Both high and low pressure cylinders were located on the same horizontal centre-line with the valves located immediately above the cylinders. The wall of the low pressure cylinder extended through an aperture in the frame on the right hand side of the engine. Ten of these engines were built at Gateshead: Nos. 1329 and 1330 in 1888; Nos. 1326 - 8 in 1889; Nos. 1527 - 1531 in 1890.

Fig. 13, NER Class J No. 1517 (author's collection)

Shortly following the "I" Class engines, which were intended for secondary main line work, the second T.W. Worsdell 4-2-2 series, the "J" Class, were built to handle the heaviest of the East Coast expresses (Fig. 13 shows No. 1517). The driving wheels were 7 ft. 7.250 in. dia., with much bigger cylinders than previous NER compounds, being 20 in. dia. high pressure and 28 in. dia. low pressure, both with a stroke of 24 in. The tube heating surface was not larger than the "I" Class but the grate area was increased from 17.2 to 20.7 sq. ft. with the grate sloping throughout. There had to be some ingenuity in the design to accommodate the huge 28 in. dia. low pressure cylinders, but this was done by placing the cylinders at a higher level, and inclined. However, this brought the cylinders so near to the underside of the boiler that there was no room for the steam chest, and so both high and low pressure valve chests were placed outside the frames. The receiver was arranged very neatly in a curved tube within the narrow smokebox. Joy valve gear was used, and after experimentation the following details were settled upon:

High Pressure Low Pressure
Lap: 1.125in 1.125in
Travel (full gear): 4.375in 5.750in
Lead: 0.1875in 0.1875in
Exhaust clearance: 0.125in 0.125in

The tenders of these locomotives were also the largest yet built for the NER, and had a capacity of 3,940 gallons of water. Their life as compounds was, however, very short as trouble arose with the outside steam chests, and broken valves, while the complicated arrangement of rocker gear from the 'Joy' motion to the outside valve spindles was another weakness. Within six years of construction all ten engines of the class had been converted to two-cylinder simples; the "I" Class receiving the same treatment soon after. Of interest here is that notes have been found that passengers experienced a rather jerky journey behind the compound singles despite them being thought of as among the finest British express locomotives of their time. However, both classes did not survive into the grouping, the last one ("J" Class 1520) being cut up at Percy Main in 1920.

The retirement of T.W. Worsdell, in 1890, and the appointment of his younger brother Wilson to succeed him saw three major points of change and development: the Worsdell-von Borries compound system was abandoned; Stephenson's link motion was used on all new engines, and the use of piston valves was developed to a high degree of reliability. Yet even though W. Worsdell disliked the use of compounding, there were three cases of further selected development.

The NER had, for some reason, always preferred tank engines with a single driving wheel to pull inspection saloons for railway officials. The original Aerolite was originally built as a 2-2-2WT (well tank) in 1851, rebuilt in 1869, 1886, and 1892. The final rebuild in 1892 by W. Worsdell was into a 2-2-4T with, strangely enough, a Worsdell-von Borries two cylinder compound arrangement. This engine was used by the Chief Assistant Mechanical Engineer A. C. Stamer until 1933 when it was prepared for preservation in York Museum, and is now the sole remaining example of an LNER pre-grouping compound locomotive.

Fig. 14, NER No. 1619 as built (author's collection)

In December 1892, W. Worsdell completed at Gateshead the largest and heaviest express engine yet to be seen in Great Britain. This was the first of the Class M1 4-4-0s (LNER D17) with coupled wheels of 7 ft. 1 in. diameter. Two further engines of this class had been completed when, in May 1893, a further example was built as a two-cylinder compound on the Worsdell-von Borries system for comparative trials. This locomotive, No. 1619 (NER Class M, LNER Class D19), had one 19 in. dia. high pressure, and one 28 in. dia. low pressure cylinder, with the stroke of both being 26 in. However, although the potential cylinder capacity was the same as the other engines of the "M" Class she carried a boiler pressure of 200 lb. per sq. in. against the 180 lb. of the simple engines (Fig. 14 shows No. 1619 as originally built). In 1898, W. M. Smith completely redesigned and rebuilt 1619 as a three cylinder compound with one high pressure cylinder inside the frames and two low pressure cylinders outside (Fig. 15). This arrangement, which was patented by Mr. Smith,, was as logical as F. W. Webb's on the LNWR was the reverse. Instead of two small high pressure cylinders and one enormous low pressure cylinder, the three cylinders in Smith's system were all much the same size. This system was adopted by the Midland Railway, and subsequently on a grand scale by the L.M.S. In its original form, as first used by Mr. Johnson on the Midland, the engines were provided with a changeover valve by which the driver could, if necessary, admit a certain amount of live steam directly to the low pressure cylinders so as to develop increased power on a heavy gradient. This is sometimes referred to as semi-compound or reinforced compound working. On No. 1619, Smith fitted independent valve gear for the high and low pressure cylinders, which called for intelligent work on the driver's part.

Fig. 15, NER No. 1619 rebuilt as 3 cylinder compound (author's collection)

Apart from the arrangement of cylinders - high pressure, 19 in. by 26 in.; low pressure 20 in. by 24 in. - there were certain other interesting details in the design. The high pressure cylinder had a segmental ring piston valve, but the outside cylinders had ordinary slide valves. The boiler was new and differed from other NER boilers by having a much larger fire grate with an area of 23 sq. ft.; the total heating surface was 1,328 sq. ft. and boiler pressure was 200 lb. per sq, in. Another feature was the use of cross water tubes in the firebox to improve the circulation of water and to increase the heating surface; this device was also invented and patented by Mr. Smith. On No. 1619 when first turned out as a three-cylinder compound there was no outward sign of these water tubes, and what was more important, no means of removing the tubes without complete shopping of the locomotive. In August 1900, a new firebox was fitted with outside covers for providing access to these tubes, and at the time of this change a tender with increased coal capacity was attached to the engine, carrying 5 tons instead of 4.5 tons. From that time the total weight of the engine and tender in working order was 94.5 tons. No. 1619 was handled by regular crews who quickly acquired the necessary technique of manipulating the independent reversing gears, and the change valve when necessary. However, W. Worsdell proceeded with some caution as the controls on No. 1619 were decidedly more than enginemen had been expected to understand and operate in the past. This meant that five years were to pass for before the last, and without a doubt, the finest of the NER compounds were to be developed. No. 1619 was reclassified from Class "M", to Class "3CC" (three cylinder compound) in 1914 and survived with the LNER (Class D19) until October 1930.

On the GNR, due to the unsatisfactory performance of his C1 4-4-2 Atlantics, Ivatt built a four cylinder compound version. No. 292 entered service in March 1905 and the design of this locomotive was distinctly odd. While it was superficially very similar to a standard Atlantic, the evident desire for this similarity resulted in the inside cylinders having extremely short connecting rods, only 4 ft. 8 in. between centres. As the piston stroke was 26in. the angularity of these rods was acute, but fortunately the inside cylinders had separate valve gear (Stephenson's), the outside cylinders having Walschaert's. In normal circumstances the low pressure cylinders would have been made at least as large as the cylinders of a standard two-cylinder simple, but in fact on No. 292 their diameter was only 16 in. as against the 19 in. of the standard Atlantic. The high pressure outside cylinders were 13 in. dia., and to achieve a volume ratio of approximately 1: 2 between high pressure and low pressure sides the high pressure stoke was only 20 in. The boiler pressure was 200 lb. per sq. in. although the boiler had actually been designed to withstand 225. This engine lasted the longest of three GNR compound Atlantics (there were a further two), in compound form. It was broken up in 1928 without having received any significant modifications; not even a superheater.

As a result of the poor performance of No. 292, the GNR Board sought a wide range of proposals for designs, including compounds, from outside builders. Many of the designs were dismissed as either being too heavy or too long. The MR compound type offered by R. Stephenson was dismissed by Ivatt as lacking sufficient power, and Oliver Bury persuaded the Board to accept the Vulcan 4-4-2 design, and even this required modification to meet GNR requirements. The second four cylinder compound was built by Vulcan in July 1905 and entered service as No. 1300. The layout was similar to No. 292, with the high pressure cylinders on the outside, a divided drive, and a boiler with a working pressure of 200 lb. per sq. in. This engine is better known than No.292, perhaps because it underwent several transformations and acquired an exceedingly bad, though deserved, reputation. It appeared that it was impossible to ensure the steam tightness of the various joints of the smoke box and around it, a condition arising partly from the unsatisfactory design of the pipework itself but also from the fact that this engine had a French style cylinder layout but lacked the very important bracing of the frames employed in French practice. As a result the front end was too flexible. This problem was never cured. Early in Gresley's regime the engine was superheated and some alterations were made at this time in the other front end arrangements but these did not suffice and in 1917 Gresley rebuilt it as a two-cylinder simple. This involved a fairly substantial redesign: even the wheelbase was altered, and the result was a less ramshackle looking machine. However, the engine was still not satisfactory and was scrapped late in 1924. The Vulcan Foundry had at least copied fairly closely the French cylinder dimensions and the volume ratio between high and low pressure sides was 1: 2.7. All four cylinders had a common stroke of 26 in. The inside, low pressure cylinders had a diameter of 23 in. and the outside, high pressure cylinders had a diameter of 14 in. There were four sets of Walschaert's valve gear.

Fig. 16, GNR compound Atlantic No. 1421 (author's collection)

The third GNR compound Atlantic was No.1421 which appeared in 1907 (Fig. 16). This was again an Ivatt engine and closely followed the Atlantic's design. The outside, high pressure cylinders were almost identical to those of No.292, but the inside, low pressure cylinders were enlarged to a diameter of 18 in., this being made possible by their valves being placed on top of them instead of between them, thus increasing the volume ratio to roughly 1:2.5. The boiler pressure was again 200 lb. per sq. in. as in the other two compound Atlantics. Not surprisingly, this engine was an improvement on No.292 and the rearrangement of the low pressure cylinders resulted in a layout of pipe-work which made it an easier matter to fit a superheater and this was done in 1914. However, it remained less satisfactory than the standard Atlantic and gave a fair amount of mechanical trouble so in 1920 it was rebuilt with two cylinders and piston valves to match the other Atlantics which were being converted from slide to piston valves.

There are very few records of tests involving these three locomotives. However, O. S. Nock in 1953, when browsing through some tests made in pre-grouping days with the NER Dynamometer car, came across a test made in 1909, beginning at Doncaster, with the new East Coast Royal Train, prior to a visit to Edinburgh by King Edward VII. The test was primarily to record the rolling resistance of the new train, and so did not include any details of the engine working. This was unfortunate, for the GNR locomotive was No 1421 and the performance was spectacular to a degree. The load was 302 tons behind the tender, and with this a speed of 60 mph. was attained in 4.75 miles from the start; and then, gradually working up to a sustained speed of 73 mph. on the level. The train passed Selby, 18.4 miles, in the unprecedented time, for that period of 19.5 minutes. Checks spoiled the continuation of the run to York, where the NER took over the haulage.

Fig. 17, NER No. 731 as built (author's collection)

In 1901 W. M. Smith had begun design work on a further NER compound locomotive class, and authority was given 1st June, 1905 to build two of what were to be the last NER compounds. The two locomotives, Nos. 730 & 731 were both built at Gateshead Works and in addition to being the only four cylinder engines on the NER they were unusual in having Belpaire fireboxes (Fig. 17 shows No. 731 as built). No. 730 was fitted with Stephenson link motion, whilst No. 731 had Walschaert valve gear, and the two engines were ready for work on 15th April and 29th May, 1906 respectively. Under the Smith system of compounding the two high pressure cylinders, 14.25 in. diameter and 26 in. stroke, were placed outside the frames and the two low pressure, 22 in. diameter and 26 in. stroke, were inside; all four cylinders driving onto the leading pair of coupled wheels. The high pressure and low pressure cranks on one side of the engine were 180 degrees apart and each pair were at 90 degrees to each other. Consequently only two sets of valve gear were necessary as 7.5 in. diameter piston valves with inside admission for the H.P. cylinders and 10 in. diameter valves with outside admission for the L.P. cylinders and, therefore, the valves on one side of the engine moved in unison (Fig. 18 shows the arrangement of cranks on 730 & 731). As built both engines used saturated steam but they were both fitted with 18 element Schmidt superheaters in 1915, when boiler pressure was reduced from 225 lb. per sq. in. to 200 lb.

Fig. 18, Crank arrangements on NER Nos. 730 and 731 (M.Peirson)

The locomotives could be worked as simples, semi-compounds, or full compounds. On starting, steam direct from the boiler to the L.P. cylinders could be regulated by a reducing valve on the side of the smokebox; this was controlled by a spring-loaded auxiliary valve which could be varied by the driver to produce a required pressure (normally between 30 and 150 lb. per sq. in.) in the L.P. steam chests. When the permitted L.P. steam chest pressure was reached the reducing valve automatically cut off the supply of live steam from the boiler and the engine commenced to work as a compound. By means of the hand controlled auxiliary valve the driver could allow live steam into the L.P. steam chests when climbing a bank, thus working the engine as a semi-compound.

These two engines should have been, chronologically, Class "W", but they were actually Class 4CC 'four cylinder compound' (LNER C8) and were true masterpieces of locomotive design, and, extremely powerful. Unfortunately, Mr. Smith died in harness on 25th October, 1906, only a few months after his greatest engines had been put into service. Because of their success it was decided to perpetuate the design on the NER and on 12th December, 1907, W. Worsdell was authorized to construct another ten engines similar to No. 731. However, the mercenary attitude of Smith's executors in demanding excessive royalties forced the scheme to be abandoned and an era of NER locomotive history came to an end.

For practically the whole of their existence Nos. 730 & 731 were stationed at Gateshead, although for a short period about 1931 they were on loan to Hull. They were classified as Class C8 by the LNER and in November 1933, instructions were given that they were "to be transferred from Gateshead to Heaton; put in good order and placed in stock and utilised at busy holiday periods". However, when No. 731 entered Darlington works to be overhauled it was decided not to proceed with the idea and it was withdrawn from service in December 1933, after running 669,328 miles. No. 730 followed to the scrapheap in January 1935, after spending time in railway exhibitions and running 752,939 miles.

Fig. 19, GCR No. 365 Sir William Pollitt (author's collection)

On the GCR, John G. Robinson was impressed by the performance of W. M. Smith's compound locomotives and ordered four three cylinder compound Atlantics to be built with the same system. They carried Nos. 258 & 259 (December 1905) of Class "8D"; and Nos. 364 & 365 (February 1906) of Class "8E" (Fig. 19 shows No. 365, Sir William Pollitt). The only difference being the mainframes as all four had the central high pressure cylinder driving the leading axle resulting in it being positioned forward of the two outside low pressure cylinders. All the engines used Stephenson valve gear driven from the rear coupled axle. The inside high pressure cylinder was 19 in. by 26 in. with a 10 in. diameter piston valve. The outside low pressure cylinders were 21 in. by 26 in. and had unbalanced slide valves. In 1911 No. 365 was fitted with a superheater, followed by No. 364 in 1913. Nos. 259 & 258 were not fitted with superheaters until 1921 and 1927 respectively. These engines worked the same services side-by-side with the other "simple" Atlantics and often worked the so-called "Sheffield Special" which at one time was the fastest train in Britain. They received the classification of C5 under the LNER and were all taken out of service just before Nationalisation (1948), thus becoming the longest surviving compound locomotives on the LNER.

The last 'compound' locomotive built by the LNER was the W1 Hush-Hush No. 10000, completed in November 1929. This was originally built as a four cylinder compound and a high pressure water tube boiler, but was rebuilt into a simple expansion engine with a conventional steam boiler in 1937, and so ending the long line of compound locomotives to run over East Coast metals.

There was, and always will be, some surprise that compounding never featured largely in the British Isles, especially since it was adopted generally on the continent of Europe and in America with some 1,000 Worsdell-von Borries compounds in service abroad. The reasons for this are probably many and varied, from the poor design or high maintenance costs of many experimental compounds to the harsh railway operating environment and limited space afforded by the loading gauge in Britain. Added to this were the concerns of engineers such as Wilson Worsdell that the successful operation of such complicated machines meant that more time had to be taken to train engine crews to utilize them to their full effect, i.e. they couldn't be placed on a 'common user' roster. Many will also point out that when superheating was developed the advantage of compounding was decreased when it came to thermal efficiency, and that superheaters were easier to build and maintain especially during the harsh conditions of two world wars. Whatever the thinking, as this article has shown, the constituent companies of the LNER built 279 compound locomotives spread over 15 classes, most of which could be considered as successful and capable engines. This can only add fuel to the debate surrounding the abandonment of compounding and the fact that the full potential of the British steam locomotive was not exploited. In France, the work of Chapelon showed what could be done with a modern compound and it is surprising that no one but Mr. Smith took the same path on this side of the Channel.


Thank you to Euan Cameron and Allan Rogers of the North British Railway Study Group for kind permission to use Euan's drawing of No. 224 and background information.

Also, kind thanks to George Moffat for permission to use the image of No. 727, and to '65447' of the Great Eastern Society for providing background information on the G16, and images that enabled me to produce the line drawing.

And finally, thank you to the playwright John Moorhouse for proofreading, editing, advice, and constant encouragement and support.


"The North Eastern Atlantics", by K. Hoole; Publ. Roundhouse Books.

"Locos of the North Eastern Railway", by O. S. Nock; Publ. Ian Allan Ltd.

"Compound Locomotives of the British Isles", by Tom Pearce-Carr; Publ. Finial Publishing.

"The North Eastern Railway. Its Rise and Development" by W. W. Tomlinson; Publ. Longmans, Green and Company.