Dorset Joint Railway
This page deals with some aspects of early signalling arrangements on the Somerset & Dorset Joint Railway (S&DJR) which are not specific to any particular location on the railway, but relate to the general situation pre-1900.
The S&DJR was formed originally in 1862 as the Somerset and Dorset Railway (S&DR) from the union of the Somerset Central Railway (SCR) and the Dorset Central Railway (DCR). It became a 'Joint' line in 1875 when it was leased jointly by the London & South Western Railway and the Midland Railway. The opening in 1874 by the S&DR of the "Bath Extension" from Evercreech Junction to Bath has been taken as a convenient starting point for these notes, as that event marks not only the expansion of the future S&DJR almost to its fullest extent, but also the first introduction of 'proper' signalling in any quantity. Before that date S&DR signalling had been relatively primitive and records are very scant.
Prior to the late 1870s or early 1880s the general attitude towards, and implementation of, railway signalling within the UK was very different to the situation that is familiar to those acquainted with the post-1900 era. The Board of Trade (BoT) came to play an important part eventually in the control and regulation of railways, but for many years the BoT functioned only as an advisory body and their regulatory powers were granted piecemeal by various Acts of Parliament. However by 1874 the BoT had gained powers over new works on passenger lines and consequently the Bath Extension was signalled from the outset to a higher standard than would exist elsewhere on the S&DJR for several more years.
When the Bath Extension was opened in 1874 it was the first portion of the S&DR to be equipped from the outset with proper interlocking of signals and points and with the passage of trains regulated by Absolute Block working. At that time the Station Agent (the S&DR name for the Station Master) had control of everything at his station; the signalling block instruments were placed in his office and worked by a telegraph clerk under his direction, while the signalman functioned merely as a 'pointsman' and operated the lever-frame as instructed. The separation of the signalman from the control of the block working contributed in part to the disaster at Foxcote (near Radstock) in 1876 and after that awful event the S&DJR began to relocate the block instruments into the signal-boxes.
Very little is known about the original signal-boxes of the Bath Extension. It has been suggested that in fact there were no 'boxes' as such at all, but simply lever-frames housed in the station buildings (which might account for their bay windows?). However that theory seems to be based purely on the fact that 'signal-boxes' are not marked specifically on the Ordnance Survey maps of the periods. A report to the BoT from a preliminary inspection in June 1874 includes the comment that "...the locking frames will require to be protected from the weather...", which would seem to suggest that certainly they were outside of any building at that time.
In contrast, there are references to 'signal-boxes' in various S&DJR records (eg the Officers Minutes) and a report in the Bath Chronicle newspaper of 23-July-1874 says "...the levers are in covered boxes on the platform...". It would seem fairly clear therefore that some form of 'signal box' did exist at various locations on the Bath Extension by the time that the inaugural train rain on 20-July-1874. However it is probable that most of these signal-boxes were little more than simple wooden huts covering the lever-frames - certainly they were very small, as it became necessary to extend them subsequently just in order to accommodate the block instruments.
It is likely that the original signal-boxes on the Bath Extension were placed on the platform close to the main station building, so that the signalman did not have to walk far between the lever-frame and the Station Agent's office. Certainly this was the case at Wellow, as is clear from evidence in the BoT Report on the 1876 Foxcote disaster, and some plans for 1886 show it then to have been at the Radstock end of the Up platform (see signal diagram) in approximately in the same position as the later ground-frame. The lever-frame at Chilcompton was also in a similar location relative to the station building in 1878 (see signal diagram). One known exception was at Radstock, where the signal-box was sited adjacent to the level-crossing and photographic evidence shows that it was a large brick-built structure.
Signal-boxes on the Bath Extension were fitted with interlocked lever-frames giving centralised control of the points and signals. It is believed that these lever-frames were supplied by Saxby & Farmer. It is possible that a few interlocking frames existed elsewhere on the S&DR before 1874, but the overall picture would have been of points and signals operated from individual local levers. More remote signals might have been operated by wire from levers placed on or near the platform - some photographs of Spetisbury show two such levers at the north end of the platform. After 1874 the S&DJR began to install many more signal-boxes in conjunction with various new passing-loops and the work accelerated with the programme of doubling the line, so that by the late 1880s the situation looked much like that of the post-1900 period (although not necessarily with the same signal-box structures). Although the Bath Extension's signalling was a vast improvement on the earlier situation, the arrangements were still fairly basic, as can be gleaned from the relevant pages that were added into the 1864 S&DR Rule Book. Only Distant, Home and Starting signals for each direction were provided at passing loops, and the liberal provision of ground signals was a thing of the future.
Disc-and-Crossbar Signal at Spetisbury
The late Robin Atthill stated in his book "The Somerset and Dorset Railway" that the SCR used 'disc-and-crossbar' signals in line with Bristol & Exeter Railway practice, while the DCR used the 'revolving disc' type favoured by the London & South Western Railway. This is a reasonable theory, but the source of his information is unknown and it has not been possible to establish any further evidence. In contrast it should be noted that the 1864 S&DR Rule Book illustrates only the 'disc-and-crossbar' type of signal. On the ex-DCR line some 'disc-and-crossbar' signals were recorded at Wincanton in 1876 and there are several published photographs showing them in use at Spetisbury. The S&DJR Officers Minutes for 21-Oct-1875 state "...all disc-and-crossbar signals to be superseded by semaphores as renewal becomes necessary...". Whatever the earlier situation, the Bath Extension was equipped throughout with semaphore signals, and this may well have been their first large-scale use on the S&DJR. Click here for more information about later S&DJR signals.
The provision of single-line passing loops, in the form of two platforms with both loop roads signalled for passenger working, appears to have been fairly scant throughout the S&DR until the mid-1870s. On the ex-DCR line nothing is known about the early arrangements at Cole and the loop at Wincanton was not installed until 1876. The situation in the Templecombe area is very confused prior to the doubling of the line to Wincanton in 1884, although there may have been a passing-loop at the DCR station prior to the opening of the Templecombe Junction Railway in 1870. Passing loops were installed in 1878 at Stalbridge and Shillingstone. A new signal-box was inspected at Sturminster Newton in 1876, but it is not clear if the loop there had existed previously. Information is lacking on Bailey Gate, but that loop may have coincided with the new signal-box which was authorised there in 1876. Only at Blandford, where some new work was inspected in 1878, is it known that a loop had been in use for some time previously.
Less is known about the ex-SCR part of the line. There was a loop and two platforms at Glastonbury in 1878 when the extension of the Wells branch line was opened, but it is likely that the loop had been there for some time. Loops and second platforms were provided in 1891 at Pylle and West Pennard, although in both cases it is believed that the loops had existed already as sidings. Edington Road was merely a platform on the single-line until it became Edington Junction with the opening of the Bridgwater Railway in 1890, at which time it became a passing-loop with two platforms (with an additional bay platform for the Bridgwater trains).
When the Bath Extension was opened in 1874 the single line was divided into the block sections BATH JUNCTION - WELLOW - RADSTOCK - MIDSOMER NORTON - BINEGAR - SHEPTON MALLET - EVERCREECH VILLAGE - EVERCREECH JUNCTION. It might seem fair to assume that each of the intermediate stations mentioned above had a passing loop with two platforms, but this was not necessarily the case and contemporary descriptions of the line give cause for doubt. The Bath Chronicle account of the opening of the Bath Extension reported as follows:-
As the line is a single one, there is no positive necessity for crossing the rails, but as there are generally two lines at the stations, it would be desirable at some time to carry over at least the platforms of the more important stations, and to start the trains from opposite sides."
DH Gale in "The Somerset and Dorset Railway - Tourists' Descriptive Guide" (published in 1874) says merely:-
"At each station there is a siding for the purpose, we suppose, of allowing trains to pass each other."
Certainly there were stations elsewhere on the S&DJR with only a single platform where trains were crossed by the simple expedient of shunting one into a siding. However the S&DJR Officers Minutes for 21-Oct-1875 record a decision to "extend to thirty wagons and engine and tender" the loops at Wellow, Radstock, Midsomer Norton and Binegar. There is reference to a loop "south of the station" at Shepton Mallet and it was decided to provide a loop at Chilcompton (which was installed in 1876). In contrast the Officers' meeting of 8-Nov-1876 decided to take out of use the loop at Evercreech Village (later known as Evercreech New).
The signal-box at Foxcote had been a 'block post' which did not have a passing-loop and it was not situated at a station. This situation did not provide any danger by itself, provided that the block telegraph working was operated correctly. However the BoT made great complaint about this arrangement in their Report after the 1876 accident, mainly (so it would seem) because the S&DJR had opened the box without BoT authority in contravention of their original undertaking on block working, rather than because of any foreseen safety hazard. Certainly the BoT Report ignored the other S&DJR locations where it is known that block-posts existed without passing-loops (eg Midford), although most (but not all) were located at stations. At that time it was a BoT requirement that passenger trains stopped at stations should be protected by signals, so even the smallest one-platform station would have had at least a couple of signals worked by levers on the platform. It was relatively simple to provide a set of block instruments, with the result that the S&DJR had far more block posts without passing-loops at that time than it would have been the practice to provide in later years.
It is important to note some features about the design and operation of the early passing-loops. The provision of signals tended to be limited to just Distant, Home and Starting signals in each direction - shunting signals were (virtually) unknown. (For examples see signal diagrams for Chilcompton 1878 and Wellow 1886.) One contemporary peculiarity which existed on the Bath Extension was that the starting signals were not interlocked with the distant signals, so that the latter merely acted as repeaters for the home signals, whereas in later practice a distant signal could not be cleared until all the stop signals in advance controlled from the same signal-box were 'off'. It is uncertain when the interlocking was updated, but certainly at Wellow the old arrangement still existed in 1886.
The BoT set a limit of 100 yards for the mechanical operation of points, which was increased at various times (sometimes with differing limits for facing and trailing points) until it reached the present limit of 350 yards. This limit had the effect of restricting the length of a passing-loop that could be controlled from a single signal-box, particularly if for some reason (eg supervision of a level-crossing) the signal-box could not be placed at the mid-point of the loop. Consequently there were several loops on the S&DJR at which it was necessary to provide a ground-frame (GF) near one end, released by the signal-box, in order to work those points which were too far away. (As far as is known, in all such cases the release was mechanical by rodding between the signal-box and GF.) Over the years these GFs were superseded as the extensions of the BoT operating limits allowed the points to worked directly from the signal-box, but some lasted quite a while; the GF at the north end of Stalbridge was not abolished until 1955, when its points were converted to motor operation.
A few words of caution are necessary about the interpretation of some of the signal diagrams which may be found in the course of research. Early diagrams tend to provide only basic information and may give rise to some confusion, as it is not unknown for signals to be omitted from diagrams or for there to be contradictions in the locking-tables. (Even in the 1960s at least one diagram hung in an S&DJR signal-box contained an error.) In later years it was the practice to state on a diagram the names of the block-posts to which the signal-box worked on either side, but in earlier years it was common simply to give an indication of the direction (eg "to Bath", "from Evercreech" etc). Some drawings for new works circa-1876 show distant signals with square-ended arms; it is not known whether this indicates that the S&DJR had not yet adopted the 'V' notch in the end of distant arms, or merely that the draughtsman was unaware of the revised form of the signals that he had to draw. Some diagrams printed in recent publications may have been 'reconstructed' from unconfirmed evidence, in order that a reasonable indication could be given of the signalling at a particular location. Similarly some authors have made inaccurate comments as a result of attempts to reconcile diagrams, plans and photographs from different periods.
© CJL Osment 2002