Ivatt left the locomotive fleet in good order, with his ninety-two large Atlantics bearing the brunt of main line express work.  To these should be added the experimental Nos. 292 and 1421, and the twenty-two Klondykes.  Gresley succeeded Ivatt on 1st October 1911.  There was clearly no need for Gresley to rush into designing anything with more power.  Improvements in superheating seemed to be the way forward, and he concentrated his efforts on the Atlantics.  Another avenue to be exploited was the fitting in 1909 by Ivatt of a Klondyke boiler to Class L1 (LNER R1) No. 133.  There were still four more of these boilers lying around since 1903, but it required instructions from the Locomotive Committee before they could be used on further 0-8-2Ts.  This came at last and the four boilers were fitted in May 1913 to Nos. 135/38/45/51.  The next order for Klondyke boilers (with superheaters) was issued in November 1915.  Two were fitted to 0-8-2Ts Nos. 129/49 with the new Robinson type and two to Nos. 135/38 with second-hand Schmidt type (almost certainly the sets removed form Atlantics Nos. 1451/60 respectively).


Gresley continued the Ivatt tradition, with straight frames for engines with narrow fireboxes, and extensions at the rear for those with wide fireboxes.  In general the main frames were 11/8in. thick, spaced 4ft.-1½in. apart (the standard width for firebox casings being 4ft.-0½in.).  Prior to Grouping Gresley only introduced one new class with a wide firebox, which was his Class A1 Pacific type 4-6-2.  The main frames were joggled inwards towards the rear followed by a parallel section under the cab 3ft.-5in. apart.  Separate outside frames, 1in. thick and 6ft.-0½in. apart were bolted to the main frames.  Strengthening plates, 1in. thick and almost 6ft. long, were bolted and welded to the inside faces of the main frames at this joint.  This rear end provided a substantial support for the wide firebox.  These outside frames also supported the axle boxes for the trailing carrying wheels.  The centring device comprised Cortozzi slides, or Cartazzis as they were usually referred to at Doncaster, above the axle boxes.  The slides were inclined at an angle of 1 in 7.1 and used the weight of the engine at the rear for centring.


The GNR numbered their boiler diagrams, possibly introduced during the late-Gresley period. Diagrams were initially arranged in engine class order, which could get out of hand with extinct and new classes. For example Diagram 1 had been allocated to the Ivatt Single-wheelers and after their extinction was available for the Pacifics.  This system was retained until 1928, when an entirely new LNER network system was devised.  The system (ignoring the former LCDR. engines that were maintained at Doncaster for a short time and tagged on at the end of the list) went as follows, using LNER engine classes.

Table 15 - Boiler Diagram Numbers



LNER Classes









C2, Q1, Q2, R1






C12, D4, E1, G1, J4, J7, J50, J55, J57



(GNR G3, J15, J17)



D3, J3, J7



D1, D2, J1, J2, J5, J6, N1, N2






K2, Q3







10A - (GNR J14)









O1, O2




Gresley introduced a number of new boiler types and for the reader’s convenience the 1928 Boiler Diagrams numbers are quoted rather than the earlier GNR system.

DIAGRAM 8.  Gresley’s first new boiler design (ordered January 1912) was used to rebuild 4-4-0s from Class D2 (LNER D4) to Class D3 (LNER D3) and 0-6-0s J5 (LNER J4) to J4 (LNER J3). Also used for 0-6-0s J9 (LNER J7).   In effect the engines were rebuilt with 4ft.-8in. diam. boilers in place of 4ft.-5in.  No new engines were built with this type of boiler, and none was fitted with a superheater.  A new feature was the construction of the barrel with one ring instead of two. 

DIAGRAM 6.  The second design (ordered March 1912) was for his first ten Class H2 (LNER K1) 2-6-0s Nos. 1630-39, with 4ft.-8in. diam. barrels.  This Diagram became extinct in 1936 when the last K1 was rebuilt to K2.  

DIAGRAM 2.  His third design (ordered February 1913), with 5ft.-6in. diam. barrel, was for his 2-8-0s.  Their fireboxes were 9ft.-6in. long which must have made firing strenuous.  These boilers were still being built down to 1947, when ultimately superseded by Thompson’s Diagram 100A boiler.

DIAGRAM 3.  His fifth design was described in the LDO as based on the preceding design just described, “but with barrel 4ft shorter and firebox 1ft. shorter”.  This was intended for new Class H3 (LNER K2) 2-6-0s, but the first boiler was instead used for Ivatt Class K1 (LNER Q1) 0-8-0 No. 420 rebuilt (February 1914) to Class K2 (LNER Q3).  A design changed appeared after Grouping (N676, April 1931) with future barrels constructed of one ring instead of two, tapering slightly so that the existing smokeboxes could be accommodated.

DIAGRAM 12.  Following the general return to 4ft.-8in. diam. boilers for the Ivatt 0-8-2Ts from 1912, Gresley found a use for thirty of the displaced 4ft.-2in. boilers for his new Class J23 (LNER J51) 0-6-0Ts (introduced December 1913), though their barrels needed to be shortened first.

PROPOSAL ONLY.  In 1915 he considered a 4-cylinder Pacific, and an outline engine diagram was prepared.  This shows that the boiler was likely to have been a stretched-out Large Atlantic boiler, with the barrel 20ft. long (requiring a lengthy smokebox).  The overall length of firebox casing would have been 7ft.-4in. (instead of 6ft.-6in.) and the length at the foundation ring 6ft.-9in. (instead of 5ft.-11in.), providing an estimated 36 sq. ft. of grate area.  The decision depended upon satisfactory results from his 4-cylinder Atlantic No. 279.  He then became impressed with the proportions of the Pennsylvania Railroad Class K4s 4-6-2 though some of the features were not practical e.g. 27in. diam. outside cylinders and he would never accept a Belpaire firebox on one of his engines.  However, the K4s seems to have had a lengthy combustion chamber and this may have caught his eye.  Here was a method of shortening the boiler barrel, while still keeping its weight centred over the coupled wheels.  He then lost interest in 4-cylinder propulsion and concentrated his efforts on designing a 3-cylinder Pacific.

DIAGRAM 96.  His sixth design (ordered June 1919) was 6ft. diam, for his three-cylinder 2-6-0s in Class H4 (LNER K3). 

DIAGRAM 94.   His seventh and final design for the GNR (ordered January 1921), with a 6ft.-5in. maximum diameter barrel, for his new Pacifics Nos. 1470/71.  The firebox casing length was 9ft.-5½in., including a lengthy combustion chamber, and the barrel length was down to a more reasonable 19ft.  (As will be seen later, the combustion chamber could be lengthened by a further 1ft. and the barrel shortened proportionately.)  The order for boilers for the first two Pacifics was issued 10th January1921, though the order for steel plates had already made (21st October 1920).  The original intention was 157 small tubes x 2¼in. diam., but this was increased to 168 (28th November 1921). Ten further boilers were ordered (10th July 1922) for Nos. 1472-81.  Two further boilers were ordered as spares (21st September 1922), but they were used instead for the next two new engines Nos. 2543/44 ordered after Grouping.

Safety Valves

Gresley fitted small Atlantic No. 986 (October 1916) with Ross “pop” safety valves. Its boiler was second-hand and so needed a raised brass seating block (drawing W150, 21 July 1916) on which to mount them.  Unusually, at its next two boiler changes, the valves and special seating were transferred to the next boiler fitted.

Gresley fitted Ross “pop” safety valves to Classes H3 (LNER K2) Nos. 1680 onwards and O2 (LNER O2) Nos. 477 onwards.  Classes A1 (LNER A1) and H4 (LNER K3) had these valves from their first appearance.

Steam Collection

The collection of dry steam was an important issue. Stirling preferred to collect steam at its highest point in a pipe running the length of the boiler, and control its output to the cylinders with a regulator valve inside the smokebox where it was easy to maintain.  Ivatt collected the steam higher still in a tall dome above the boiler which also housed the regulator valve which was accessed through the removable dome lid. 

Gresley continued the latter practice for a while though as boilers became larger so the domes became shorter, reducing the height above water level where the steam was collected.  At one time he was tempted to try the Swindon arrangement for superheated boilers in which the regulator valve was placed in the smokebox on the output end of the superheater header.  Superheated steam circulated continuously through the elements, thereby avoiding the risk of burning the element ends through becoming too dry.  Doncaster produced a drawing (N298, August 1914) for trial on an Ivatt 0-8-0, but the scheme appears to have been abandoned because of the 1914-19 War.


When Gresley succeeded Ivatt he was not slow to act.  A batch of replacement boilers for Class K1 (LNER Q1) was ordered (23 October 1911), nine to have the usual Schmidt superheaters but one to have the Robinson type, in which the element ends were expanded directly into the header making a better steam-tight joint. (This was the only difference between the two types of superheater.)  The Robinson type was fitted to No. 402 (to traffic April 1912), and for a while new and existing engines were fitted with either type.  In 1912-13 five Class C1 (LNER C1) Nos. 1453/54/55/57/58 which already had Schmidt 18-element superheaters were give Schmidt 24-element superheaters (in 3 rows of 8 flue tubes).  It is possible ten sets had been ordered but only five were used for trial purposes.

The last reference to ordering Schmidt superheaters for new engines was in June 1913, when 9 sets of the 18-element type were for ordered for new Class J22 (LNER J6).  These were fitted to Nos. 567-75.  The 10th set was for the Doncaster “twin-tube” and this was fitted to No. 563.  The last references to existing engines receiving Schmidt superheaters were Class C1 (LNER C1) Nos. 1404/05/06/16/47 in 1914-15. This was the 24-element type, but it is not known if these were ordered specially or if the material was already in stock from a pervious order.

The next development (January 1913) was the Doncaster superheater, or rather Doncaster straight-tube (to differentiate it from the later “twin-tube”).  This had separate saturated and superheater headers arranged vertically, to avoid transference of heat between the two sections.  The three superheaters Schmidt, Robinson and Doncaster straight-tube, differed solely at the header end.  The arrangement of elements and heating surfaces was unaffected.  The last reference to Doncaster “straight-tube” superheater was 10 sets ordered (12 August 1913, drawing N266) for new Class H3 (LNER K2) Nos. 1640-49. The order was amended (12 November 1913) to 10 Robinson sets (drawing N273).   The reason for the change is not recorded, but was probably because the “twin-tube” had just appeared and seemed a better option.

The Doncaster “twin-tube” was an improvement on the “straight-tube”, its two headers were arranged horizontally instead of vertically.  The top one was for saturated steam and the bottom one for superheated steam. Whilst the steam still made two passes through the boiler, this was now performed through two adjacent flue tubes, one above the other.  There were 17 elements and 34 flue tubes, the latter now 4in. diam.  The elements extended further into the flue tubes picking up more heat. One advantage with this arrangement was that an element could be changed without disturbing the others.  Class J22 (LNER J6) No. 563 received the first set (August 1913).  A 21-element version was introduced (September 1915) for the Large Atlantics (C1).

Class J23 (LNER J51) 0-6-0T No. 167 was fitted with a Robinson superheater (April 1914).  The 18-element type was intended, which was a novel fitting for a 4ft.-2in. diam. boiler.  At the last minute it was decided to make it 16 elements so that more small tubes could be fitted.  Gresley afterwards admitted it had been fitted as an experiment and that the small saving of coal, only 3.7 per-cent compared with the other engines of the class, had showed it was not worthwhile fitting any more.  It remained superheated until it was rebuilt to Class J50 (October 1930).

Gresley decreed (30 July 1914) no further Schmidt superheaters were to be obtained, though the existing sets were allowed to last out their normal lives, which was well into the ‘twenties.  This left the Robinson type and Doncaster “twin-type” as the only ones to be ordered in future.   However there was a later variation of the twin-tube, which was the “triple-tube” (drawing N294, July 1914) in which the element branched into two smaller tubes just prior to entering the flue tube and at the firebox end, reverted back to one tube at the return bend.  There were thus three tubes within a flue, two of 1in. diam. plus one of 1¼in. diam.  A 17-element version was fitted to Class J22 (LNER J6) Nos. 563 (6/1915 - 6/1927); 3627 (8/1928 - 7/1932), and a 21-element version was fitted to Class C1 (LNER C1) No. 1417 (5/1917 - 6/1924).

The next development was the 32-element Robinson superheater, within 4 rows of 8 flue tubes.  This was fitted to the early Gresley Pacifics and the two Class P1 2-8-2s, apart from the trial fitting on both classes of the 62-element ”E” type superheater. 

Conjugated Valve Gear

A feature of Gresley’s three-cylinder engines was their conjugated valve gear to operate the valves of the middle cylinder.  Considering a circle with three radii marked at 120 degrees and with a knowledge of trigonometry, it is possible to describe a system which will locate a third radius using levers.  Harold Holcroft (1882-1973) realised this fact and took out a Patent (No. 7859, 1909) for a conjugated valve gear.  Unfortunately for him he had allowed it lapse in 1913.  Gresley had also realised this mathematical fact and in November 1915 applied for Patent rights for a conjugated valve gear for three-cylinder engines, employing either rocking shafts or motion levers. 

The motion arrangement drawing was completed in April 1916 and a month later Gresley deposited the Complete Specification at the Patent Office. In August that year Gresley’s patent agents informed him about the Holcroft Patent, now lapsed, but Gresley decided to proceed as his three-cylinder engine was under construction.  Just what was “under construction” is not known, as this was four months before the Engine Order was issued (EO283, 8 December 1916) and the frames were ordered (23 February 1917).  Meanwhile the Complete Specification was accepted (Patent No. 15769, October 1916).

No boiler was ordered for EO283 (engine No. 461).  There was a spare boiler in stock as when the previous order had been issued (13 May 1913) for the boilers for five 2-cylinder 2-8-0s Nos. 456-60, one more was added (17 July 1913) to provide a spare.  This was most unusual, almost without precedent, and it does raise the question of how long he had been planning this 3-cylinder engine.

Gresley wrote to the Institution of Locomotive Engineers (December 1918) “it was not until some twelve months after, that I discovered through the Patent Office that Mr. Holcroft had a gear for the same purpose”.  “twelve months after” what is not clear, but probably after his first application, but this is what you pay the Patent agent for.  The original correspondence with his Patent agent was filed away in the LDO, where I rescued it in 1965.

The reciprocating motion of the three valve spindles follows the rotary motion of the connecting rods.  Two of the valve spindles are connected by links to the outer pins of the equal motion lever, providing a combined motion at its fulcrum equal to half the necessary value of the equivalent rotary motion of the third valve spindle, but 180 degrees out of phase. The fulcrum is at the short end of the 2 to 1 lever, which at the long end double the motion and corrects its phase.  At the long end of this lever is a short link connected to the third valve spindle.  The equal motion is 2ft between pins with tits fulcrum midway, while the 2 to 1 lever is 4ft between pins.  As the spindle links do not move in a straight line, because of the slight angular movement of the levers, these nominal lengths are adjusted by a small fraction to compensate.

In its application on No. 461 the three cylinders were inclined at 1 in 8 so that the inside connecting rod would clear the leading coupled axle.  The simple form of the valve gear could not be fitted as it would have meant arranging the 2 to 1 levers at 1 in 8 too.  The alternative arrangement was used in which the valve gear on either side of the engine drover the outer arms of a rocking shaft which operated an outside valve.  The inner arms of these two shafts operated the middle valve and the 2 to 1 ratio was achieved by making the inner arm on the right-hand shaft twice the length of the inner arm on the left–hand shaft. The middle steam chest was located low down, to the left of the middle cylinder.  

This was a most complicated arrangement and afterwards Holcroft advised Gresley that it was not essential for the three cylinders to be in the same inclined plane.  All that was needed was to displace the middle crank from the theoretical 120 degrees setting by the angular difference in the inclinations of the middle and outside cylinders.  By arranging the outside cylinders nearer to the horizontal it was then possible to use the simple form of the conjugated valve gear, with 2 to 1 motion levers in front of the steam chests. Gresley embodied this principle in his H4 2-6-0s.

This was still rather complicated and for the later 2-8-0s design work was handed over to N.B. Loco. Co.  This firm recommended setting all three cylinders and steam chests at 1 in 30 which Gresley agreed to.


Gresley’s first bogie engine was Ivatt Atlantic No. 279 (Works order dated April 1914) which he rebuilt in 1915 with four cylinders.  To clear the inside cylinder valve he employed 3ft.-2in. diam. wheels, as in the Vulcan engine No. 1300, and so the main frames remained parallel at the front.  The Atlantic bogies continued to give trouble, and early in 1922 the bogies of Nos. 298 and 1424 were rebuilt with high-tensile steel frames.  Then from April 1922 new frames were specified 11/8in. thick plate instead of 1in., thereby dispensing with the strengthening plates which Ivatt had been forced to apply.  The bogies for new Pacifics in 1922 conformed to the new standard.

The bogies for his A1s (introduced 1922) were constructed of 1½in thick steel plate, spaced 4ft.-0¾in. apart.  The helical bearing springs were stiffer to support the greater weight on the bogie.  The axle journals were 9in. long as in all previous Ivatt bogies, and 6½in. diam.  The permitted side play of 3½in. was found to be inadequate and there were signs of the rear bogie wheels having been in contact with the cylinders, and the maximum side plate with safety was deemed to have been 3¼in.  Part of the bogie frame casting was cut away (from August 1927) on existing and new Pacifics to permit 4in. side play.

Pony Truck

Class H2 2-6-0 (LNER K1) was Gresley’s first important design.  No. 1630 was developed in quick time during 1912.  Engine ordered 23rd March, boiler ordered 26th March 1912, frame drawing completed 22nd April, pony truck drawing completed 15th May, and the engine into traffic in the August.  A significant feature of the new engine was its pony truck with double swing links for centring and suspension.  To be developed so quickly suggests Gresley had working on the idea for quite some time.  A further nine engines (Nos. 1631-39) were ordered shortly afterwards (8th August) and these appeared in the first half on 1913.

The frame plates were 7/8in. thick, set 2ft.-11½in. apart. The wheels were 3ft.-8in. diam. (3ft.-2in. on Nos. 1631-39) and the radius of the bar arm was 6ft.-4½in.  The bearing springs on the axle boxes were the laminated type (changed to helical in 1914).

The next design was Class O1 (LNER O1) introduced December 1913.  The wheels were 3ft.-2in. diam.  The bearing springs were the laminated type, altered to helical probably during their 1916 repairs in Works.  The length of radius of the bar arm was 6ft.-0½in.

A similar pony truck was used for Class H3 (LNER K2) introduced August 1914. The wheels were 3ft.-2in. diam., radius of the bar arm was lengthened slightly to 6ft.-7in., and laminated bearing springs on Nos. 1640-47 only. Commencing with No. 1648 (June 1914) helical bearing springs were fitted instead.  The earlier engines were afterwards brought into line.

The next design was Class O2 (LNER O2) with three cylinders introduced May 1918.  No. 461 (only) had the usual 3ft.-2in. wheels, but later engines had smaller wheels 2ft.-8in. placed 10in. further forward on account of the proximity of the outside cylinders. The bearing springs were helical.  The length of radius of the bar arm for these later engines was 6ft.-6in.

Next followed Class H4 (LNER K3).  The radius of the bar arm was 5ft.-11½in. long, the bearing springs were helical.

Tender Orders

Gresley carried on building Class B tenders, summarised as follows:

Table 16 - Gresley Class B Tenders

















































Further Class B orders were placed after Grouping.

Gresley’s only new tender design for the GNR (R83) was his Class G 8-wheeler, carrying 5000 gallons of water and 8 tons of coal, introduced in 1922.  The wheel spacing was 5ft.-3in. + 5ft.-6in. + 5ft.-3in. Their wheels were 4ft.-2in. diam.  The frames were 24ft.-1½in. long, 1in. thick and set up 5ft.-6½in. apart.  The width over the side sheets was 8ft.-0½in. and that over the running plate was 8ft.-9in.  The main tank was 21ft.-9in. long.  All the following dimensions are approximate. The main tank was 4ft.-4in. high and 8ft. wide.  The front section of the tank contained the coal space, overall length 10ft.-8in.  This was self-trimming like the 3500 gallon type but the sides also tapered inwards and not just the back.  Here the water space tapered inwards from 8ft across to 2ft.-8in. wide at the floor of the coal space at a point which was 5ft.-8in. from the front of the tank.  The sides tapered inwards from 8ft. across to 2ft.-8in. wide at the floor of the coal space.  Below the coal space was the lower part of the main tank 8ft. wide and 1ft.-3in. deep.  The floor of the coal space was 5ft.-8in. long and 2ft.-8in. wide and level.  On this rested the shovelling plate, slightly longer than the available floor space, so that its inner end rested on the taper section, having the effect of raising it slightly at its inner end.

The first order was TO39 (17/2/1921), two tenders T.5212/13 for engine Nos. 1470/71  which appeared in 1922, followed by the second order TO40 (9/8/1922), ten tenders T.5223-32 for engine Nos. 1472-81 which appeared in 1923.  Further Class G orders were placed after Grouping.  

Tender Shortage 1918-22

Seventy tender engines were supplied by contractors in 1918-21 and all were delivered without tenders, as follows:

45 Class H3 (LNER K2) Nos. 1660-79 (1918), 1680-1704 (1921),

15 Class O1 (LNER O1) Nos. 462-76 (1919),

10 Class O2 (LNER O2) Nos. 3477-86 (1921).

In the same period Doncaster built 23 0-6-0s, 10 2-6-0s and 1 2-8-0, all of which stretched the resources of the Tender Shop, building new tenders and refurbishing those taken from withdrawn engines.  The process of cutting up withdrawn engines could not keep up with withdrawals, and storage in the scrap yard was limited.  Around a hundred engines (tank and tender) were taken out of stock in 1920-21 and stored wherever space could be found. In the case of the tender engines this released their tenders.  As examples, the following were taken out of stock in 1920.

0-6-0 143A, 3/1920, broken up 24/3/1921

0-6-0 173, 6/1920, broken up 1/1921

0-4-4T 242, 12/1920, condemned 8/1/1921

0-6-0 322, 4/1920, no disposal shown

0-6-0 327, 12/1920, sent to Romania

0-6-0 341, 21/12/1920, condemned 8/1/1921

0-4-4T 507, 3/1920, condemned 11/1921

0-4-4T 515, 3/1920, condemned 11/1921

0-4-4T 516, 12/1920, condemned 1/1921

0-4-4T 654, 12/1920, broken up 1/1921

0-6-0 732, 3/1920, broken up 2/1921

0-6-0 733, 12/1920, condemned 8/1/1921

0-6-0 736, 12/1920, condemned 1/1921        

0-6-0 798, 12/1920, broken up 1/1921

0-4-2 960, 12/1920, broken up 1/1921

0-6-0 1022, 6/1920, broken up -/1921

0-6-0 1038, 12/1920, sent to Romania

0-6-0 1089, 12/1920, sent to Romania 12/1920

Large numbers were taken out of stock during 1921, none at all in 1922 and several in 1923, 1924, 1925 and 1926.

As at 31st December 1922 there was one engine in store, Class E1 2-4-0 No. 1066, taken out of stock July 1921 and broken up July 1923, so it never appeared in LNER stock.

Tender Brakes

The single 21in. brake cylinder was placed at an angle towards the front of then tender thus both clearing the axle and increasing the distance between the sloping piston and axle.  The 8-wheelers had twin 18in. cylinders which also sloped.