The Webb "Tumbler" Interlocking

An ingenious catch-handle driven interlocking

F. W. Webb, the celebrated Chief Mechanical Engineer of the London and North-Western Railway, is best known for his locomotives, but he also made important contributions to railway signalling, notably his interlocking frame, and the Webb-Thompson electric train staff. The interlocking frame, manufactured at the LNWR signal works at Crewe from iron and steel made in LNWR furnaces and formed in LNWR rolling mills, only found application within the LNWR family, but the train staff was used worldwide.

Anticipating the comprehensive signalling of their lines, in 1873 the LNWR directed Webb to create a signal works, formed a Signal Department, and appointed Mr G. Edwards as Signal Engineer to assist Webb. Many railway companies designed and built their own locomotives, in spite of the existence of excellent independent locomotive manufacturers, who had to survive on foreign orders. Some companies followed the same route with signalling apparatus. The Midland never used outside contractors, in fact, and both the LNWR and the GWR later became self-sufficient. There was even less reason for this than in the case of locomotives, since independent signal contractors, among them Saxby and Farmer, Stevens and Sons, McKenzie, Clunes and Holland, and the Railway Signal Company all offered first-rate equipment at reasonable prices, and developed worldwide reputations for quality. In fact, the excuses for inside supply that were offered at the time seem lame and insufficient, and there is evidence that inside supply was more expensive than using contractors.

The LNWR had earlier relied mainly on Saxby and Farmer for mechanical signalling equipment, but after 1873 an alternative had to be found. Webb attacked the problem with his usual gusto, and by 1876 had designed a substitute for the Saxby and Farmer apparatus. He recognized quite clearly that Saxby's principle of early locking and late release was essential. This meant that when the signalman even thought of pulling a lever, all conflicting levers should be locked before any motion took place, and any lever that was released when the lever was pulled should not be unlocked before the lever had made a full, effective stroke. Saxby did this by operating the locking by the movement of the catch handle that released the lever in its quadrant enough to lock all conflicting levers, and then to continue the motion to release levers when the catch handle returned as the dog fit into its notch in the quadrant for the reverse position of the lever.

Both operation by the catch handle, and locking the lever by preventing the catch handle from releasing the lever in the quadrant, were firmly patented, and, in fact, had been the subject of a lengthy legal dispute between Saxby and his employee Easterbrook, that was most unsatisfactorily resolved by giving half the invention to each, thus hindering its employment. Webb decided to lock the lever instead of the catch, by a piece that would engage a stud in the frame, high enough on the lever for a strong lock. The other half of the problem was solved by making the movement of the catch handle, which was now the familiar LNWR loop, lift the lever, which then operated the locking. The locking itself was quite different from that used by Saxby and Farmer. The result was a machine that had no resemblance whatever with the Saxby frame, but accomplished the same ends. However, the judges were not fooled by the mere change of fulcrum, and ruled that the method of operating the locking infringed Saxby's patents.

Back at the drawing board, Webb realised that there was a way to make the lever motion operate the locking and still achieve early lock and late release. Stevens and Company had been doing this for years in their simple, effective locking. The first motion of a lever only took up lost motion and slack, and effective operation of points or signal did not occur until the middle of the stroke. Moreover, the function was complete before the end of the stroke, and the last part only strained the connections to press parts firmly against their stops. He already had such a frame ready in 1876, with the benefit that the lever was again in one solid part, instead of in two parts with relative motion (one part connected to the leadout, the other moving to operate the locking). By 1885, the design had been considerably simplified, and it took the form that became classic.

The principle of the Webb locking is shown in the figure. The lever is not shown to aid in clarity. The locking piece D, acting like a catch, is pivoted on the lever at E, and engages the projection B at its ends. If the lever is free, it rises at once when the lever is pulled, in fact in the first 1/32 inch of motion, in the Webb official line. Link F connects piece D with the sword piece A; the inclination of F is exaggerated in the figure. The sword piece slides on the lever fulcrum C, and in guides below. Locking bars G reciprocate longitudinally in the frame. Locking studs or dogs, riveted to the locking bars where necessary, prevent a sword piece from moving in either its normal or reverse position by engaging in the notches. When a lever is reversed, the sword pieces move a distance equal to the pitch between notches, so that they again register with the locking bars. The method of driving the locking bars is not shown in the drawings, and is not mentioned in verbal descriptions. They could be driven either by linkage from the sword pieces, or by inclined cams. I also do not understand the force causing the locking piece D to rotate further when the lever is reversed, completing the motion. However this is done, the first motion causes the locking studs to only move halfway, which maintains the existing locking and engages the new, while the last motion completes the unlocking.

The final version was the widely-used LNW Tumbler locking, manufactured from 1876 to 1906. This was a development from his catch-handle driven frames of 1874-76 that might have resulted in litigation with Saxby and Farmer over their catch-handle patents, and completely sidestepped the problem by using an indirect lever drive that proved very effective. The essential component in its functioning was the 'tumbler,' a poorly-chosen term that can be confused with the Midland tumbler, which was something totally different. A better term might have been 'rocker,' but that was already appropriated by the Saxby and Farmer link. From the drawings I had, I could not see how the tumbler was driven, although how it worked was fairly clear. R. D. Foster simply says the tumbler 'flips over' when reaching the final position. This reference contains excellent drawings and photographs of the frame.

When I later consulted the first-rate Guide to Mechanical Locking Frames of the Signal Study Group, I was hoping for some clue as to how the Webb tumbler worked. It was a small detail on the drawing that made everything clear. In all my drawings, the lever obscured the centre of the tumbler, so I did not know that there was a small lug there that engaged the backs of the lugs on the quadrant casting, and it is this that makes the tumbler move as is necessary. I have made a schematic drawing to show how the tumbler worked, and the reader should find a drawing of the complete Webb frame to understand how this element was used. I could have seen this in the drawings in Foster, but I did not.

In the figure, the tumbler is shown in mid-stroke, when the lever on which it is pivoted can freely move back and forth. Surfaces that are labelled by the same capital and small letter are those that come into contact. Surfaces A and B lock the lever in normal or reverse position, and cam the tumbler up sharply when the lever is moved a very small distance--Webb's 1/16 inch. The tumbler is linked to the hook rack, which in this frame corresponds to the tappet iron in a tappet interlocking. If the rack is locked, the tumbler will not rotate and the lever is prevented from moving.

When the lever is normal, to the left, the tumbler rotates anticlockwise, either from the weight of the hook rack or the action of the lug on the tumbler. The hook rack is all the way down. When the lever is pulled, the tooth rides up on the lug, rotating the tumbler clockwise into the position shown. This is sufficient to lock all conflicting hook racks. As the lever is pulled over, no further rotation of the tumbler occurs until the lug c contacts surface C, and surface B is clear of surface b. Then the tumbler rotates further clockwise, raising the hook rack the rest of the way, releasing conflicting racks. If the rack is locked in this position (by a locking stud) then the tumbler cannot rotate anticlockwise, and the lug and tooth will prevent the lever from moving. The teeth rest on the quadrant all the way between surfaces A and B, but can drop off the ends.

The tumbler was later replaced by a cam or Z drive, using a stationary plate with a Z slot and a follower connected to the hook rack. This may have been simpler, and perhaps easier than replacing worn lugs, but it would not seem to have been as effective as the original design. Wear of the tumbler and quadrant was the principal defect of this frame. In any case, the bar and stud locking was arranged vertically. As with tappet locking, special locking could be accommodated. The locking bars were operated by cranks from the hook racks. The frame was so heavy that it was supported on the ground, not on the signal cabin structure, like other frames.

In the later LNW Tappet frame, made from 1903 to 1930, the tappet locking was driven from one end of a rocker segment, as in the Saxby and Farmer frame, except that rollers connected with the catch handle box ran on the top and bottom of the rocker instead of in a slot. This locking, arranged in horizontal locking boxes, was a complete departure from the Webb frame, and was catch-handle locked instead of lever locked. This method of driving the locking does not suffer from wear.

The Webb locking is simple and robust, and gave good service. It is ingenious and original as well. It is, however, very heavy and massive. The crossbar locking takes up a great deal of room. The locking for the 208-lever frame at Rugby South Junction was 18 feet deep. The sword pieces were divided into two sections, connected by a compensating lever so that as one rose, the other fell, removing much of the static load In published drawings, each locking section accommodates 10 locking bars. For comparison, the Saxby and Farmer 240-lever frame at Brighton required only 4-1/2 feet depth. The Saxby and Farmer patents were not the only ones that had to be avoided. Both cam plate actuation and tappet locking were protected by patent at the time.

Signalling on the London and North-Western in 1885 is described by A. M. Thompson, then its signal engineer, in the Minutes of Proceedings, Institution of Civil Engineers, for the session of 5 May 1885. In the discussion, F. W. Webb, C. E. Spagnoletti, J. S. Farmer, I. A. Timmis, T. Blackall, J. W. Fletcher (LNWR telegraph superintendent), G. Edwards, Henry Johnson, W. Langdon, A. J. Hamilton-Smythe, and R. Price-Williams are heard from.

R. D. Foster, A Pictorial Record of LNWR Signalling (Oxford Publishing Co., 1982).

Composed by J. B. Calvert
Created 12 July 2004
Last revised