Le Crocodile

An introduction to French cab signals and speed recorders


  1. Introduction
  2. Cab Signals
  3. Speed Recorders
  4. The Crocodile
  5. References


This article gives an introduction to cab signals in the days of mechanical signalling, with specific reference to developments in Belgium and France that created the nearly ubiquitous crocodile, so named by the cheminots for its resemblance to a reptile crouching between the rails. I will give my best interpretation of the nature and use of this device, based on my references, but of course I am not an expert on it. The crocodile is rather different from any device in British or American practice, and I hope to make it more comprehensible to the English-speaking world. The speed recorder, which became closely associated with the crocodile, will also be discussed.

Much of this article refers to French practice, in the era of private railways, and the action transpires largely in the first two decades of the twentieth century. Railways in France were State property, leased from the ballast up as concessions to private companies to maintain and operate. The large companies, which will appear often in this paper, were the Nord, the Est, the Paris-Lyons-Meditérranée (PLM), the Paris-Orleans (P-O), the Midi, the Ouest, and the Etat. (The Etat, which later absorbed the Ouest, was originally not considered a major company, so there were often said to be six major companies.) All operated radial trunk lines from Paris, except for the Midi, which later was associated, but not merged, with the P-O. The Midi served the territory southwest of a line from Bordeaux to Sète, including all the lines to Spain. The Societé National des Chemins de Fer (SNCF) absorbed all these companies when it was formed on 1 January 1938, creating a national railway system. By some wonderful stroke of luck, it yet survives in these days of privatization (2004).

French railway signalling was diverse and nonstandard until the ministerial order of 15 November 1885, which prescribed standard practices for hand, flag, lamp, whistle and fixed signals. Red was the colour for stop, green for caution (go slow) and white for clear, as in Britain and America. Yellow was used for stop on shunting lines, and violet for direction indicators. These general prescriptions were in force until the reform of the signalisation Verlant of 1934, which introduced green for clear, and yellow for caution, somewhat later than in Britain and America. French signalling is characterized by round, square or diamond-shaped targets rotating about a vertical axis and distinctively painted. Semaphores were also used, especially for block signals and route indicators. Whenever two lights were displayed, one was direct from the lamp, the other by means of a reflector. The carré showed a blue backlight when off, a white backlight when on. Its target carried a reflector as well as a blue lens. Both the disque rouge and the carré were invented by Robert of the Chemins de fer du Nord in the 1860's.

In the days of "time interval" working in France, trains ran from station to station, started at appropriate times by the stationmaster with his pocket whistle or guidon. There were few if any intermediate signals, which mainly protected stations and junctions. Approaching a station, a driver would first see the disque rouge in a familiar location. These could be operated by wire at a distance of 1500 to 1800 metre, but were usually located 800 to 1000 metre from the station. When "on" they were repeated by a trembling bell or a visual indicator near the employee operating them. A driver finding one "on" was immediately to make himself "master of his speed" and proceed prepared to stop short of a train ahead. A péal Aubine might be used to set the signal "on" automatically after his train. This was the point at which the crocodile was also placed, to give an audible warning of a restrictive signal. A sign by the track marked the point at which the signal gave sufficient protection, which was at braking distance from the disc (about 800 m). By night, the disque rouge would show a red light when on, and a white light when off.

The next signal encountered might be a signal annunciateur, a green and white checquerboard showing the aspect of the stop signal ahead. This signal would show two green lights by night (horizontal with full braking distance, vertical with restricted distance), and a white light by night. Finally, the stop signal would appear, a carré which must not be passed when on. This signal, showing two red lights by night when on, and one white light when off, would sometimes be doubled with detonators. A block semaphore would be treated in the same way. This signal would show one red light and one green light, horizontally, when on, and a white light when off. All the double lights were produced by mirrors, with one lantern. Both the semaphore and the carré might be present, the carré protecting switches and crossings, the semaphore authorizing entrance to the block. This was generally the way signals were used in France until the Verlant Code of 1934.

One might expect that Belgian signals would closely resemble French signals, but that is far from the fact. Belgium followed its own path, with some influence from France, it is true, but generally with distinctive practices. Its early signals were rotating targets, but the stop signal was a red disc (about 1 m in diameter) and the distant signal a red rectangle (1.4 x 0.7 m), located 800 to 1200 m from the stop signal. The lamp could be behind a red glass in the target, or separate with a moving spectacle. Both signals showed red when on and white when off, with a blue backlight when off, produced by a blue glass behind the target. Both were stop signals, since the distant signal protected a train standing at the red disc. The driver had to approach the distant signal prepared to stop if it were on. Sometimes a distant alone protected a small station. These signals were replaced by warning semaphores after 1907. The last disappeared only in the 1950's.

The Nord-Belge railway was formed in 1854 to take over the lines in Belgium previously leased by the Nord, and always was friendly to the practices of the mother company. This brought the disque rouge, the carré, the crocodile and the Lartigue electro-semaphore into Belgium. Later Belgian signalling practice was, however, based on the semaphore, both two- and three-position, and used British route signalling practices with semaphore signals on chandeliers. The distant signal was distinguished by a broad arrowhead instead of by a fishtail arm.

It is customary in American and British signalling practice to say that a train coming up to a signal is in rear of the signal, but after it passes will then be in advance of the signal; that is, to take the point of view of the train, not the signal. In common language, as well as in some of the references, the use of "in advance" may be exactly the opposite. I often use the convenient British term "off" to describe a signal showing clear or safety, and "on" when showing its most restrictive aspect. The corresponding French terms are ouvert and fermé, "open" and "closed." The aspect of a signal is what it displays; the indication is the the information conveyed.

Cab Signals

From the time of the introduction of the first fixed signals in the 1850's, the desirability of a warning of an adverse signal in the driver's cab was recognized. This was especially true when fog obscured the signals, but was also useful in general to reinforce the vigilance of the driver. A cab signal may operate in two ways: it may warn the driver or cause him to take some further action demonstrating his vigilance, or it may take control of the train by initiating a brake application that will bring the train to a halt. An audible signal, because it actively attracts the attention, is always a part of a cab signal, though a visual reminder is often present as well. It has also been suggested that the warning should be loud enough to be heard by other members of the train crew as well, when present at their usual positions on the train. The driver may demonstrate his alertness and prevent a brake application by operating a vigilance device, which is usually a lever to pull or a button to press at a certain moment. We shall consider a "cab signal" (French, signal d'abri) as a general term for any such system, including those usually called "automatic train control" (ATC) or "automatic train stop" (ATS), unless specifically noted.

There was a general objection by railway authorities to all safety devices that their use would result in a dimunition of the vigilance of employees, as reliance would be placed upon them instead of on a good lookout. It was supposed that a cab signal repeating wayside signals would be depended upon to give notice of an adverse signal, instead of keeping a careful lookout ahead for the actual signal. The small chance of a driver's overlooking a signal would be replaced by the possibly greater chance of failure of the automatic apparatus. There was also an objection to any system that produced an automatic stop, beyond the control of the driver, since such a stop could be dangerous in itself. French authorities strongly disapproved of automatic stops. They also rejected recognition of signals at clear, a feature that was so much liked on the GWR warning system (discussed below). Indeed, the GWR warning system was envisaged as replacing wayside signals, but this was never done. Unless the warning system includes a brake application, the case of a totally incapacitated (or deeply asleep) driver is not properly handled. In American and British practice, the driver must keep his foot on a pedal or his hand holding down the brake handle to avoid a brake application.

A cab signal may operate by mechanical contact, by electrical contact, by a combination of the two, or by electromagnetic contact without physical contact, either by induction or by electromagnetic waves. All of these methods have been used, and others more exotic still. Here we consider only mechanical or electrical physical contact, which were the only available alternatives until the 20th century. Mechanical contact, using hanging levers, treadles and similar means characterized most of the first attempts at cab signalling. Efforts persevered on such systems, in spite of the mechanical difficulties encountered, which included shock, foreign bodies, ice and snow, misalignment, and so forth. However, electrical contact eventually won out on all systems brought into practice. The GWR and Miller systems used combined mechanical and electrical contact, however. Most current cab signals work by induction, or by the pickup of signals transmitted through the rails, and do not involve physical contact.

The detonator (UK) or torpedo (US) or pétard (France) was invented by E. A. Cowper in 1842. It is a very effective audible signal that is attached to the railhead and explodes when run over. It is usually manually placed, though it may be placed on the rail by a signalman operating a lever remotely, or automatically when a signal displays danger. We do not include the detonator as a cab signal proper. Since detonators command an immediate stop, they are generally used at signals where a stop is required. They give definite evidence that a signal has been passed at danger, and this is probably their most important function. In France, they were typically used to supplement the carré.

A similar signal is a wayside gong, operated by a treadle. When the signal is restrictive, a bar is raised enough that it is depressed by the flanges of each wheel at one end of the axle. One sound is made for each axle passing over the treadle. This was first suggested at an early date, and numerous examples can be found in the British and American patent literature. None was ever regularly used in practice. The important consideration is the mechanical arrangements, of course, not just the idea of ringing a bell. Such a device (maybe the last) was invented by M. A. Louis, and was tried near Rouen on the Paris-Havre line in 1915.

Automatic train stops on underground railways typically operate by raising an operating lever at a stop signal that catches and operates a cock that produces an emergency application of the air brake. This has only been found practical in such circumstances, where the track is inaccessible, and the trains are uniform.

Inventions for giving warnings of adverse signals date from as early as 1850 in Britain. An early example of a mechanical cab signal was Anderson's Audible Signal, introduced in 1865 on the North British Railway between Carlisle and Edinburgh. A gong on an engine or brake van was rung when a hinged lever beneath the vehicle struck a block between the rails. There was a roller on the bottom of this lever, and the wooden block was inclined in the direction of approach. This wooden block was moved sideways by a mechanical linkage from the distant signal. When the distant was "off" the block was to the side and did not engage the lever. When the signal was "on" the block was in the way of the lever, and the gong was rung. It is easy to imagine how the system could be modified by using two levers to give different and distinctive signals for "clear" and "caution". This arrangement seems simple and effective, but like all such devices was not used permanently. A similar system on the London, Chatham and Dover Railway in 1870 used a leaf spring beside the track that was bowed upwards when the distant was "on" to operate a similar lever mechanically, and ring a gong. At about the same time, Axel Vogt, Master Mechanic of the Pennsylvania Railroad, placed a glass tube above the cab of a locomotive, connected to the new automatic air brake pipe. A distant at "on" caused a lever to project so that the glass tube would be broken, and the brakes applied. A similar device was invented by P. Ribard in France in 1891 and was tested. Yet another of about the same time used a knife to cut a leather strip that released a spring.

These examples (and others) were at least tried experimentally, though none was used widely nor for any length of time. A search of the patent literature will uncover many schemes similar to these examples. Nearly all the patented schemes were totally useless, and most were completely impractical. Railway authorities grew very annoyed by such proposals. Inventors had a very poor grasp of railway operations, and of the requirements of service. The existence of a patent is no guarantee that the invention is useful, practical, or new. All government inquiries into railway safety with requests for apparatus, as in America, Britain or France in the early years of the 20th century, brought forth a flood of inventions, most completely ill-conceived and worthless. Nearly all useful suggestions came either from railway companies, or from established signal manufacturers.

The most important requirement of an acceptable signalling device is that it must "fail safe," which means that the failure of any part results in a more restrictive indication. This means, in part, that any indication of safety must be actively maintained, and that the restoration to danger is the result of gravity, a permanent magnet, or other unfailing force (not, for example, by a spring). Less obvious is the requirement that the system should fail so rarely that its restrictive indications will not be distrusted and ignored. Of course, the usual requirements of low maintenance, economy and reasonable lifetime must be met. Not all the devices ever used in railway signalling satisfy these requirements exactly, but most do, especially in British and American practice.

The two rails are electrical conductors more or less insulated from each other. That the rails can form parts of an electrical circuit for detecting the presence of a train, or for communication by telegraph with the train, was realized by Steinheil in 1838 when he was developing the electromagnetic telegraph. It is very simple to arrange a circuit so that the wheels and axles of the train close a circuit by bridging between the rails. The consequent current can then be used to operate an electromagnet for signalling purposes. Such an "open track circuit" appeared very early and was used in many patented signalling schemes. However, it was nearly always rejected because it does not "fail safe." A break in a wire, or the failure of the battery, causes a "false clear."

Contact can also be made with a short conductor near the track by means of a brush or shoe on the vehicle, with the rails used only as a common return. The presence of a train is detected by the momentary closure of the circuit. This, too, is an open-circuit system that does not fail safe, but has been the basis of many inventions, among them the Lartigue crocodile. With respect to the crocodile, this is justified since a failure is not necessarily a safety hazard; its real purpose is only to foster discipline. The train is always considered to be in the hands of its crew. In practice, the crocodile fails very rarely.

It was not until 1872 that William Robinson, in the U.S., devised the fail-safe "closed track circuit." This is discussed in detail elsewhere on this website, but its essence is the establishment of a closed circuit between a battery and a relay with connections made to the two rails. The current normally holds the track relay closed when no train is present. When the wheels and axles of a train bridge the two rails, an added current flows, causing the voltage applied to the relay to decrease, and the relay to drop out. Any failure of the current also will cause the track relay to drop out, so the arrangement is fail-safe. Note that the track relay drops out by the action of gravity, not a spring. The closed track circuit has been fundamental to American railway signalling.

Two very practical cab signal systems combining mechanical and electrical action were the Great Western Railway automatic train control (ATC) in Britain, and the Miller automatic train stop system in America. Both used a ramp engaged by a shoe on the locomotive. In the GWR system, this ramp was between the rails and the shoe was beneath the cab; in the Miller system, the ramp was to the right of the track, and the shoe was on a tender journal (so its height would be constant and not affected by springing). When a ramp was encountered in either system, the shoe was raised and this mechanical motion initiated a delayed brake application. When the ramp was electrically energized, the current operated a contactor that cancelled the brake application. In the GWR system, this cancellation caused the ringing of a trembling bell, an audible indication that the signal was clear, and in the Miller system a green light appeared. If the ramp was not energized, then a whistle was blown in the cab in either system. The GWR system displayed a "sunburst" pattern in front of the driver, and the Miller system a red light. If the driver operated the forestalling lever within a prescribed interval (a few seconds), the whistle ceased to sound and the visual display was cancelled. Otherwise, the Miller system operated the automatic brake valve causing an emergency application, while the GWR system destroyed the brake vacuum, also bringing the train to a halt.

On the GWR system, ramps were located at distant signals. As a train progressed, each clear distant signal gave a brief and reassuring ring of the bell. A distant signal at caution made a startling whoop of the whistle. It was originally thought that this worked so well that the wayside distant signal could be removed, and reliance placed entirely on the ATC. Though this would have been quite practical, distant signals were not removed, since they gave important locational tools. The Miller system was used on the Chicago and Eastern Illinois Railroad from 1911 to 1950 on double track between Chicago Heights and Clinton, Illinois. There were one or two other installations, but they were neither extensive nor long-lasting. Ramps were placed at the three-aspect block signals, and energized only on a clear aspect. At engine terminals, test ramps were provided. They were, of course, not energized. When an engine passed over them, the driver could forestall (or allow the penalty application to take place) and check that the system was working. This system was superseded only when spare parts could no longer be obtained. The GWR ATC lasted many years into the BR era, until it was replaced by the inductive AWS. These systems combined repetition of signals in the cab, vigilance control (an inert ramp could be located wherever a warning was desired) and automatic train control, stopping the train if the driver did not act. They were fail-safe, because any electrical fault would cause a brake application.

Speed Recorders

It happened that the question of cab signals in France around the turn of the century coincided with the question of installing speed indicators and recorders in locomotives. Switzerland already required speed recorders in all passenger locomotives, so the necessary equipment was, to some degree, already available. Speed indicators were not used to any extent in America or Britain until the time of the diesel locomotive. Enginemen were expected to measure their speed by timing mileposts, and the rhythm of their engines gave them a good feeling for speed. With electric and diesel locomotives, however, these clues are absent, and some sort of speed indicator was necessary.

All such devices were speed recorders as well as speed indicators. The record was kept on a paper tape that could be unrolled and examined at the end of a run to provide evidence of the speeds attained on the journey. Accidents had occurred on lines with steep gradients and sharp curves, and it was expected that speed recording would lead to better control of train speed. Some French railways provided bonuses for keeping time and for economy in fuel use, and it was suspected that such incentives might lead to excessive speeds at times. Knowing that one's performance could be examined was a strong motive for keeping within the speed limit.

There was very little opposition or objection to installing speed recorders, unlike the question of cab signals. The only drawback was the expense, which was not great. Speed recorders became closely associated with crocodile cab signals, because they were applied to record the actions of the cab signals and the driver. The installation of cab signals and speed recorders proceeded hand-in-hand.

Two examples of the first speed recorders, were the German Hausshaelter and the similar Swiss Hasler, used on the Midi and the P-O and in trials on several other lines. They were operated by clockwork, and recorded the speed as a function of time. A cylinder that could move up and down was rotated by clockwork. It was a cylindrical rack driven by a gear whose rotation was taken from the locomotive wheels. Every 12 seconds the cylinder rotated so that the gear encountered a groove that allowed the cylinder to fall, still rotating, and pick up the gear again. The amount that the cylinder rose in each cycle was proportional to the speed. The height of the cylinder drove the recording pen as well as the indicator needle. If the speed was decreasing, the follower was pushed down on the cylinder at the end of each cycle. Unfortunately, I do not have a drawing, and so cannot be more specific about the principle of operation. The Hasler was similar, but recorded every three seconds. These are examples of integrating speedometers.

The most popular speed recorder turned out to be the French-designed Flaman recorder, in which the tape was driven by the locomotive wheels at the scale of 5 mm per km. The speed was determined by integration, as in the Hausshaelter recorder, but with intervals of 3.6 seconds. The integration is performed by a ratchet wheel driven by a square cam, so that the wheel moves forward four spaces for each rotation of the cam. The action is the same regardless of the direction of travel. Another shaft, driven by the clock, controls the resetting of the ratchets. The scale of the speed curve is 0.4 mm per kmph, and the range of the recorder is 0 to 130 kmph. This is the lower curve on the chart, showing speed as a function of distance, v = v(s).

The upper, or time, curve showed the time as a function of distance, or t = t(s). The pen moved vertically, driven by an Archimedean spiral at a uniform rate of 2 mm per minute. The width of the time graph is 2 cm, representing 10 minutes elapsed time. When the pen reached the top, it returned quickly to the bottom, producing a sawtooth graph. The appearance of the tape is sketched at the right, including the cab signal line between the two sections. The ticks on this line are made by the driver and by the signal, and so are double. On an actual tape, speed was labelled every 10 km/h, and time every 2 min. Full information about the movement of the train could be determined from the tape. Note that when the train was stationary, the time trace would move vertically. A full tape would record for 4350 to 4970 miles. Different sizes of driving wheels were allowed for by changing the gears that drove the square cam. The Flaman recorder was used on the Nord, Est, PLM and Etat. By 1914, about 80% of French locomotives were equipped with a speed recorder.

With the Alsace-Lorraine Railways, acquired during the War, came a new type of speed recorder, the Deuta. This recorder operated on the now-familiar principle that the torque exerted on a short-circuited armature by a rotating magnetic field is proportional to the speed of rotation. The armature need only be restrained by a spring to give a deflection proportional to speed. This was probably not as accurate as an integrating speedometer, but was much simpler mechanically and gave a continuous indication of the speed. Most modern speedometers are of this type, using a drag proportional to speed.

The Crocodile

The crocodile is an electrical contact placed between the rails (in the four-foot way) to provide warnings in the locomotive cab. It is distinctively French, originating on the Chemins de Fer du Nord around 1872, spreading throughout France and penetrating even into Belgium after 1900. Unlike similar devices in Britain and the United States, it does not have a mechanical component, but is purely electric. It was intended principally to provide evidence of the alertness of the driver, not to act to control a train automatically. Contact was made with a brush of silicon copper wires mounted beneath a locomotive passsing over it.

The crocodile is an invention of the ingenious Lartigue and Forest of the Nord. It was placed 100-200 metres in rear of a distant signal, usually a red disc of "deferred stop." When the distant signal was "on" a contactor placed +12 volts on the crocodile (8 Leclanché cells). The negative pole of the battery was connected to earth (the rails). The locomotive circuit went from the brush, through a Hughes electromagnet (a polarized relay, explained below), and finally to the wheels and to the rails. The short pulse of current that flowed when the brush contacted the crocodile released the electromagnet, causing a steam whistle to sound in the locomotive cab. The driver depressed a lever to again engage the magnet and reset the whistle.

After speed recorders were introduced around 1915, the current pulse was recorded as a short line on the speed recorder tape, together with a short line made by the driver by pressing a button when he perceived the signal. If the distant signal were "off," the crocodile was not energized, and nothing happened. Later, a negative voltage from a pole reverser was applied to the crocodile of an "off" signal, which did not affect the Hughes, but made a tick on the recorder tape in the opposite direction to that made by an "on" distant signal. This is illustrated below in the section on speed recorders.

By 1880, all the double lines, and all the locomotives, of the Nord were equipped with the crocodile. In 1914, the Nord operated 2386 miles, of which 1392 miles were double track, all fitted with the crocodile, together with 2350 locomotives. There were 41 crocodiles on single tracks, at specially important points. This is a very early date for effective cab signalling, for which the Nord deserves much credit. Other French railways did not follow the Nord for many years, not until after 1915 under government prodding.

In the 1885 signalling code, the distant signal was called an "advanced signal of deferred stop." It was a red disc with a white border. It was fermé, closed or "on", when the disc was transverse to the track, and ouvert, open or "off", when parallel to the track, so only the edge was visible. By night, it showed a single red light or a single white light. When an "on" disc was sighted, the driver was immediately to make himself "master of his speed" by all means at his disposal, and, after bringing his train under control, to proceed with caution at not to exceed 30 km/h (15 km/h for freight trains), stopping short of any obstruction ahead, switches, crossings, or absolute stop signals. The red disc was normally used as a distant signal for a "square signal of absolute stop," the 2x2 red and white checkerboard that could not be passed when "on". A green and white checkerboard square was used by some companies as a distant signal instead, often indicating a divergence in advance that required limited speed. Crocodiles were used with this signal, as well as with the red disc, on the PLM and the Est.

A diagram of the original Lartigue and Forest crocodile as used on the Nord after 1872 is shown at the right. The commutator on the signal closed the circuit when the signal was "on," which meant that a stop was required at the next signal. The box labelled "Hughes" was mounted in reach of the driver on the locomotive. The crocodile was a wooden baulk covered with sheet brass on its upper surface.

As shown in the diagram at the left, the Hughes electromagnet consisted of a strong permanent U-magnet with coils wound around soft-iron pole pieces on each leg. Normally, the magnetic force was amply sufficient to hold the lower lever against the poles. The upper lever then held the steam valve closed. If a sufficient current of the proper direction passed through the coils, the field of the coils overcame the permanent magnet, and the lower lever dropped. The downward motion was aided by a spring (not shown) surrounding the vertical rod. This also allowed the upper lever to fall, opening the steam valve and sounding the whistle. On locomotives with the Westinghouse brake, an air whistle was used. The apparatus was reset by pressing the reset lever, which raised the lower lever so that it was again attracted, and the steam was shut off. The lever could also be held down as a forestalling lever. On locomotives with vacuum brake, the system was also arranged to apply the brakes automatically, but the penalty brake application was eliminated with air brakes, since their emergency operation was rapid and violent.

The crocodile could also be arranged to give notice of the passage of a train past a fixed point to stations and block posts. To do this, a separate crocodile was installed connected to a battery and an electrically operated bell at the fixed point. When the brush contacted the crocodile, the circuit was completed, but the current passed in the opposite sense, and did not sound the whistle. However, the Hughes electromagnet at the fixed point was connected to operate with this reversed current polarity. Usually, the two crocodiles, one after the other, were located at the distant signal approaching the station. The Nord had 125 stations so equipped. Test crocodiles (crocodiles d'essai), with (+) voltage applied, were used at engine terminals and other locations where the system was to be tested. These could be quite short, not metres long as actual crocodiles were.

The arrangement described is used when the direction of traffic is constant, as on normal double track. If trains operate in both directions on one track, measures must be taken so that the apparatus works only for trains in the direction approaching the signal. This can be done by displacing the crocodile and brushes to one side, as was done on the Midi. Vehicles that travel in both directions would have to have a switch to select the brush to be used. An easier solution, and the one actually used on the Nord, was to retain the central location, but to place an additional crocodile at each side of the normal crocodiles. Contact with one or the other of the end crocodiles would establish the direction, and automatically operate a reversing switch on the locomotive so warning would be given in the proper direction only. A "tumbler" switch was also devised that was placed at the end of the crocodile and operated mechanically by the pressure of the brush, and this placed the voltage on the crocodile when the train was moving in the corresponding direction. When encountered by a train in the opposite direction, the crocodile remained de-energized.

Similar devices had been suggested in Britain much earlier, and like the crocodile, were "open circuit" devices depending on the completion of a circuit. Such arrangements had the fatal flaw that failure of the battery or the breaking of a wire (not unknown contingencies) would imply "safety" when "danger" should be indicated. For this reason, such devices were strictly and completely rejected in British and American practice. When the Belgians first considered the crocodile around 1914, they, too, rejected it for the same reason. Nevertheless, the French persisted with the open-circuit crocodile, and the reason seems to be that the crocodile was a supplementary measure, and was valuable in spite of this serious drawback. In the French literature, there is very little if any mention of the open-circuit dangers of the crocodile.

The cab signals of the crocodile originally, and for many years, had no connection with the brakes, and its actions were not recorded (since speed recorders had not been introduced). The purpose was not to apply the brakes if a distant signal was passed at caution, as in the GWR system in Britain, but to ensure the vigilance of the driver. The Nord appears to have arranged the warning system to apply the Smith vacuum brake before 1885, at least experimentally. This was not the automatic vacuum brake, but the straight vacuum brake, on the locomotive. The vacuum ejector was caused to operate by the action of the Hughes electromagnet, applying the brakes. Around 1897, Marin suggested that the operation of the cab signals should be recorded on a tape, and this was tried in 1897 on the Etat system. On the Est, it was arranged that a mark was made on the tape when the driver operated a lever at the moment of sighting a signal. Since every distant at "on" was recorded on the speed recorder tape, together with the driver's response, an examination of the tape would show if the driver had been vigilant, or was surprised by the whistle. Every driver would do his best to ensure that his tape showed him to be alert, so recording would overcome the objection that cab signals would decrease the driver's vigilance. It was soon realized that a recording speedometer would be a valuable aid to the driver in itself, beyond its connection with cab signals. By the 1920's speed recorders were in general use in France, recording the operations of the crocodile as well as the speed of trains.

Government agitation for a warning system began around 1900. The Est and the PLM began trials with the Nord's crocodile around 1902, while the Midi and the P-O investigated the mechanical Poublan system, the Ouest the Ribard system (with a hanging lever and treadle), and the Etat the Van Braam and Cousin systems, also with treadles, under government prodding. A Ministerial circular of 21 March 1911 mandated the installation of cab signals, but compliance was very slow. Many mechanical systems were tried, such as the Marin system on the Est, but with little success, and gradually the electrical contact of the crocodile was adopted generally. The P-O, curiously, did not use distant signals but only the square absolute stop signals, protected by detonator placers. Therefore, they could not see the utility of cab signals. After the World War, they adopted distant signals (but still as stop signals) and the crocodile, dropping the Poublan system. By 1922, the crocodile was nearly universal in France. The Midi, however, continued to argue that detonator-placers would serve instead of cab signals, though it had used crocodiles since 1911.

The hanging arm of the Augereau system on the Etat was replaced by a radio-frequency link with a Ruhmkorff coil spark transmitter at lineside, and a coherer and relay on the locomotive. A 15 m copper wire 2 m above the ground served as a transmitting antenna. A copper wire on the buffer beam and along the sides of the locomotive was the receiving antenna. When the wheels operated a treadle current was sent to the transmitter if the signal was at danger, and a whistle was sounded on the locomotive. This is another open-circuit arrangement, and any problem would result in no warning. If it had been arranged so that the radiofrequency pulse would have cancelled the warning in case the signal were clear, then it would have been fail-safe. This system was used between Paris and Chartres on 40 signals, but it was soon abandoned in favour of the crocodile.

As has been mentioned, Belgian railways had rejected the crocodile as not fail-safe. On 17 April 1929, a Paris-Amsterdam express entering Hal in Belgium collided with a postal train killing ten postal workers in a wooden sorting van. The driver of the express claimed that the distant had been "off" so that he did not receive timely warning of the stop signal. This, of course, is always the classic defense of any driver passing a signal at danger. Two other accidents around this time were due to excessive speed on hilly curved lines. Accordingly, the Etat Belge adopted the crocodile in 1929. This shows rather clearly that the crocodile was intended to enforce discipline rather than to stop trains passing signals at stop. The first Belgian line equipped with the crocodile was the newly-electrified Bruxelles Nord--Anvers Central line, inaugurated 5 May 1935. In a few years, all main lines had been so equipped.

Lartigue and Forest, and the Etat Belge Railway in 1914, rejected mechanical action as unsatisfactory. It is to be noted that in the GWR system, and the similar Miller system in America, contact shoes were lifed by the lineside ramps with good reliability. This mechanical motion started the action, and if the ramp were energized, then the current would forestall a penalty brake application. Unlike the crocodile, this was a "fail-safe" system that gave many years of good service, on the GWR and British Railways until the 1980's, and on the Chicago and Eastern Illinois from 1911 until 1950.

The insidious and persistent enemy of the crocodile was ice, and in particular the rime (French, givre) or hoar-frost, deposited by a freezing fog, that could insulate the horizontal contact surface and prevent the brush from making contact, perhaps giving a false safety. The original crocodile of the Nord was satisfactory because of the temperate climate on this line near the Channel. In 1910 the Est, where winter conditions were more rigorous, developed a grid crocodile and a brush of steel leaves that was the basis of the later French standard crocodile. Other similar crocodiles, such as the Cousin, with sharp edges were also introduced by various companies. The Nord, Etat Belge and later the SNCB adopted the anti-icing crocodile invented by J. Colas, which had a reservoir holding paraffin oil and a series of wicks that kept the contact surface thinly oiled. A special steel wire contact brush was used with them. Any rime that formed glided easily off this surface under the action of the brush. The paraffin had to be replaced only once a year, consumption being only 4 litres annually. These crocodiles were 12 cm broad and were placed 95 mm below the rail surface. On a gradient, the Colas had to be level, because of their liquid content, and the distance below the rail varied from a maximum of 90 to 105 mm at the ends. These were replaced in the 1950's with the standard French crocodile, consisting of three wavy steel bars 3 m long, with approach ramps of 2 m length at each end. These crocodiles were 10 cm wide, placed 92 mm below the rail surface. Incidentally, the GWR and Miller ramps had no trouble with ice, since the contact shoe would force it off quite effectively, and there was no horizontal surface, only a sharp vertical edge. incidentally, the Colas crocodile was originally part of a system that used a magneto on the locomotive rather than batteries at the signal.

At a speed of 120 kmph, the contact time was 0.09 s, quite sufficient to operate the electromagnet. Indeed, only 0.005 s was necessary, as was found by test. The original 12V became 16V and finally 20V, though temporary crocodiles remained at 16V. The Est introduced a pole changer that applied a (+) voltage when the signal was on, and a (-) voltage when it was off. This doubled the information provided by the crocodile, so that the positions of all distant signals could be recorded on the speed tape, together with the driver's response. The whistle was still activated with a (+) voltage on the crocodile, but a (-) voltage sounded a gong. This gave two distinctive signals, like the GWR system of so many years earlier. The final improvement was the interpretation of a de-energized crocodile, detected by its electrical resistance when the brush passed over it, as "stop." Unless the driver pressed a button within 4 seconds, a penalty brake application was initiated. It seems, however, that brake applications were not generally a part of crocodile cab signals.

When recording of cab signals was introduced, the crocodile was moved closer to the signal, often directly opposite it, to reduce the chance of a change of the signal between the time the locomotive passed over the crocodile and when the locomotive actually passed the signal. If a signal changed suddenly to stop in the face of the driver, it would appear that he had not noticed it and had been surprised, when that was not the case.

An important element in many electrical signals was the Hughes electromagnet, essentially a polar relay, but one that can provide mechanical force. It was invented by the English physicist Hughes (1831-1900), not the American inventor of a good but complex printing telegraph, David Hughes of Kentucky (or so my references imply). A permanent horseshoe magnet fitted with soft iron pole pieces, around which coils are wound, attracts its armature when no current passes through the coils. The effect of a current through the coils depends on its direction. If it acts to strengthen the magnetic field of the permanent magnet, there is no change, and the armature is attracted even more strongly. If the current acts to create a field opposing the permanent magnet's, at some point the field is neutralized and the armature is released, perhaps under the action of gravity or a spring. The Hughes electromagnet responds only to currents of one sign, ignoring the other sign. Polar relays are similar, and can be made to open or close contacts when currents of different signs are applied, or even neutral contacts that open or close independently of sign. The Hughes allows the crocodile to respond only to the (+) voltage on the crocodile applied by a distant signal at caution, and not to the (-) voltage applied when the signal is clear. A polar relay in the speed recorder may respond to both polarities and distinguish them on the speed tape.


J. B. Snell, Continental Railway Handbooks: France (Shepperton, Surrey: Ian Allan, 1971). A useful introduction to French railways in English.

B. Dieu, Histoire de la Signalisation Ferroviaire en Belgique, Tome II (Bruxelles: P.F.T.T.S.P., 2003). pp. 123-138.

M. Cossmann, Revue Générale des Chemins de Fer, 23, No. 2 (February 1900), pp. 131-137. "Note sur le crocodile Lartigue & Forest."

L. Galine, Exploitation Technique des Chemins de Fer (Paris: 1901) pp. 206-208.

G. Dumont, Traité Pratique d'Électricité Appliqueé à l'Exploitation des Chemins de Fer (Paris: 1885) pp. 244-247.

____________, Le Génie Civil, 59 pp. 166-168 (1911).

____________, Le Génie Civil, p. 235 (10 Avril 1915). The wayside gong acoustical signal.

F. Maison, Bulletin of the International Railway Association Vol. III, no. 11, November 1921, Report No. 1, pp. 1709-1800. A report written in 1914, delayed on account of the war. A thorough discussion of French cab signals and speed recorders.

J. Verdeyen, Bulletin of the International Railway Association Vol. IV, no. 6, 1922, Report No. 2, pp. 537-552. The above report brought up to date in 1922.

Return to Railways Index

Composed by J. B. Calvert
Created 22 May 2004
Last revised 28 May 2004