Unexpected Diversion Accidents

Accidents in which a switch was irrationally moved when a train was immediately approaching

Accident Reports

The Interstate Commerce Commission (ICC) was created by Congress in 1889 to regulate railway rates. It assumed safety concerns in the agitation that began around 1900 to compel the adoption of the block system by railway companies. This effort was unsuccessful, so it turned to compelling the adoption of automatic train control. Sometime in this period, the commission's membership ceased to contain representatives from the railway companies, reflecting rather the reforming attitude of the times, and a limited knowledge of railway engineering and operation. The Safety Division was created, and under the Accidents Report Act of 6 May 1910 received mandatory reports of accidents causing significant property damage or personal casualties. The Safety Division considered the act a mandate to investigate accidents and make recommendations, and did so up to its dissolution, when its functions passed to other governmental agencies.

The accident investigation reports of the ICC, 1911-1966, are available on the internet at DOT Website. The included search capacity makes this collection of special convenience to researchers. These documents were prepared with OCR programs, and contain many typographical errors. The sense, however, is usually easy to determine, and the misprints lend some humor to the dry accounts. Prior to 1911, there were no official reports, and one must rely on accounts from newspapers, the Railroad Gazette, and state regulatory agencies. These reports are often inaccurate, incomplete and biased. The best analysis of early accidents is found in Charles Francis Adams's Notes on Railroad Accidents (1879). Adams, among other things, was a Massachusetts railway commissioner. The book by Robert B. Shaw, Down Brakes (London, P. R. Macmillan, 1961) brings Adams up to 1960. There is no index in this work, except for a very useful list of accidents with page references, and few footnote references.

In 1920, when the ICC received extended powers under the Transportation Act of that year, the power to compel installation of Automatic Train Control (ATC) on certain divisions of railways sealed the confrontational relationship between the ICC and the railway companies, which became a permanent feature. Because of this, the efforts of the Safety Division proved negligible, to the injury of everyone involved--companies, regulators and the public. Many senseless regulations were issued, many reasonable ones rejected, as a result of the antagonism. In the ATC fiasco, the ICC began by inviting inventors to contribute, and they did in a rush, but all their inventions were impractical and useless. The only ones to survive into practice were those taken up or developed by individual companies, notably the Pennsylvania Railroad, where coded track circuits and cab signals were developed in cooperation with Union Switch and Signal, and the General Railway Signal Co., which developed the intermittent inductive system. The ICC's compulsion was planless and random, sometime absurd. Since those times, the knowledge of the commissioners or their descendants has probably improved, and the technical competence of railway company officers has probably declined, but the relationship between them remains the same. The American distrust of "company" men among regulators is at the root of this problem. If men were honest, a mixture would be best for everyone, but, of course, men are not honest.

Our purpose in studying these accidents is to discover their causes, and ways to prevent them. The ICC, C. F. Adams, R. B. Shaw and railway companies exhibit the same interest. The recommendations made by the ICC at the ends of their reports are notable for their lack of utility, often missing the point entirely and failing to give suggestions for the future. Although the causes of accidents are sometimes clear, the proximate cause, and the relative importance of supporting causes, is not recognized. It would seem that the investigators have had little practical railway experience. It is common, especially in the earlier reports, to chastise railway management for lack of supervision and other grave faults. If a line has no block system, one is recommended. If a line has a block system, ATC is recommended. It seems clear that no such action was expected to be taken, and apparently none ever was.

The accidents that will be studied in this paper are those in which trains were unexpectedly diverted at a switch, called an "open switch" in railway jargon, and especially when the switch is opened immediately in front of a train moving at speed. There seems to be some psychological phenomenon operating here that is not generally recognized, and is displayed by these accidents. The opening of the switch is an irrational action that would never be taken in normal circumstances, and cannot be properly explained by the perpetrator. Both experienced and inexperienced persons can be seized by this irrationality, but inexperienced persons can exhibit other reasons for opening the switch that are not irrational, merely incorrect. A selection of accidents that do not show the irrational action will be discussed for contrast.

The following table lists such accidents, discussed below, in which trains were unexpectedly diverted at a switch by an irrational action, and also some of similar type with other causes for contrast. The list is the result of computer searches designed to find the accidents of interest. There is no good way to be sure that all accidents of this type have been seen, so the list may be incomplete. To find the ICC report, search for the location (omit state abbreviation) with the search facility, and select the report desired from the list obtained. The digitization includes many misidentifications, some of which are humorous. Be cautious especially with numbers.

Unexpected Diversion Accidents, 1911-1966
Year Road Location Type Time Signals Trains Fatalities Notes
1923 MP Amboy, AR meet 9.50 pm MBS 2 psgr 2+0
1927 MC Carpenter, MI meet 9.18 am MBS psgr & frt 1+0 inexperience
1928 SOU Danville, KY yard 5.15 pm ATC frt 1+0
1929 A&WP East Point, GA meet 9.49 am ABS psgr & frt 0+0
1929 SL-SF Henryetta, OK derail 5.59 am ABS psgr 2+12 sabotage
1931 GT Battle Creek, MI yard 12.45 pm ABS psgr & yard 0+0 switchtender
1931 MC Town Line, MI derail 5.19 pm ABS pass 0+0 operator-switchtender
1934 SP Manteca, CA derail 10.30 am ABS psgr & frt 3+0 smoke obscured signal
1936 L&A Moore, LA derail 11.50 pm None psgr 2+0 sabotage
1938 SP Tortuga, CA meet 1.36 am ABS 2 psgr 3+8 Shaw
1938 CB&Q Napier, MO junction 4.25 pm MBS psgr & frt 2+0 9900, Pioneer Zephyr
1943 CGW Melbourne, IA meet 12.37 am ABS psgr & frt 0+0
1945 NP Prosser, WA meet 4.14 am ABS 2 psgr 1+0 Shaw
1946 B&O Lincoln Park, NY yard 5.35 am ABS frt & caboose 2+0 Shaw
1947 T&P Ranch, TX meet 7.05 pm APB 2 frt 3+0 Shaw
1949 B&M Canaan, NH meet 2.34 pm ABS 2 psgr 0+0 Shaw
1950 LIRR Huntington, NY meet 4.11 pm MBS psgr & frt 0+0 Shaw; inexperience
1953 UP Red Desert, WY derail 9.20 am CTC frt 3+0 Shaw*; sheep and sectionhand
1956 ATSF Robinson, NM meet 3.01 am ABS 2 psgr 20+0 Shaw*
1957 SOU Atlanta, GA yard 8.04 am None 2 cuts 4+0 Shaw; power switch
Type: meet--collision on siding; yard--between yard movements; junction--at a junction switch; derail--derailment.
Signals: ABS--automatic block system; APB--absolute permissive (automatic) block; ATC--automatic train control; MBS--manual block system; None--timetable and train orders only.
Trains: types of trains involved--psgr, frt, yard.
Fatalities: first figure is employees and others, second figure passengers.
Notes: Shaw--appears in table in Shaw, p. 143. * means discussed there.

Several features of this list are worthy of notice. First, most of the accidents occurred in Automatic Block Signal territory; there are none for meets under timetable and train order operation alone. Some accidents even occurred in Automatic Train Stop territory. The ICC investigators seem somewhat disconcerted by this fact. Secondly, the fatalities are quite low, usually confined to employees. In fact, only 8 passengers were killed in all, in a single accident, when a lightweight coach was slightly telescoped by a baggage car. Thirdly, the accidents begin abruptly in 1923, and end in 1957 just as abruptly. This may be an artifact of the search procedure, however. The peculiar nature of these accidents may not even yet be fully appreciated.


A train must follow the route set for it. This is one of the great advantages of railways, though it brings with it the negative factor that evasive action cannot be taken in an emergency. The route is normally set by a person not on the train, but one that has alighted from the train (a member of the crew) or one assigned to that duty at that point. The route is set by a track arrangement technically called a turnout in American parlance, arising from the days when one wagon turned out for another to pass on a single-lane road. The more colloquial term is a switch, which actually refers to the movable points that select the direction to be taken. A switch is said to be facing with respect to an oncoming train if the train can take one of the possible routes, and trailing if it comes from the opposite direction, when no choice is possible.

All turnouts used on railways are now split switches. The outer rails are continuous and unmodified, one perhaps going straight on, and the other curving one way or the other. These are called the stock rails. The closure rails are the other rails of each route, crossing at a frog that permits the flange to cross from one side to another at that point. A frog has wing rails at the toe end, that bend and run on either side of the point of the frog, which extends to the heel end. The heels of the points are hinged at the end of the closure rails about 6-1/2" away from the stock rails. They are planed to a 1/4" width 2" below the rail head at the point or toe, and connected by several rods of adjustable length, called head rods and back rods, usually spaced 2' apart. Sometimes manganese steel wearing points are provided, about 2' 6" long. In use, switch rails rest against one or the other of the stock rails and are supported by them. The stock rails often have rail braces on the outside to resist lateral forces; 9 is a typical number of braces. The standard throw of the points from one side to the other is 4.75", enough to clear the backs of the wheels. The clearance must be larger than the difference between the gauge of the track and the distance between the wheel backs, plus the width of a flange (the check gauge, about 4' 6.75"). The points slide on greased switch plates. Guard rails are necessary to hold the wheel flanges away from the point of the frog. They are placed beside the stock rails opposite the frog. The frogs of yard switches may be self-guarding, using raised surface to press against the faces of the wheels, but such frogs are not used on main lines.

A split switch may be run through safely in the trailing direction however it is set. If it is not set for the direction in which it is used, the connections to the points will be broken or distorted, and action must be taken to render the switch safe for facing movements. In an emergency, this is done by "spiking" the switch points in one position or the other, until permanent repairs can be made. Some switches, however, can be routinely run through in a trailing direction. In such a spring switch, the points are held closed by a spring, which permits trailing movements on either route, no matter how the points are set. Usually, a damper, an oil dashpot, is provided so that the points remain set as pushed for a short time, to prevent constant oscillation. Normally about 30 seconds are required for a point to close again after being pushed aside. This time is adjustable by selecting ports in the mechanism. It is very important to verify that a spring switch is properly closed after it has been used, and that nothing is keeping the point away from the stock rail. This is often indicated by a signal associated with the switch, which shows danger when the points are not properly against the stock rails. A so-called safety switch will snap to either position depending on how it was last run through. Such switches are often used in yards.

The angle of a frog is specified in American railway engineering by its number, defined as shown in the diagram at the right. The larger the number, the smaller the frog angle. The curve joining the frog and the heel of the point rail is sharper, the smaller the frog number n. A number 10 frog was typically used for siding switches, which gave a 6° (955 ft radius) curve between points and frog. This curve was not superelevated, of course, which meant that it could not be negotiated safely at high speeds. In fact, a No. 10 switch cannot be negotiated comfortably above 20 mph, and the usual speed restriction was 15 mph. Centralized Traffic Control (CTC) permitted trains to enter and leave sidings without stopping, and a higher speed was desired. A No. 15 frog permits 30 mph, and a No. 20 allows 40 mph. Such turnouts became used for main-line crossovers and junctions on high-speed lines, giving rise to the definition of "medium speed" as either 30 or 40 mph, depending on the nature of the turnouts used. There are now turnouts that permit 90 mph on the diverging route, but they are not used in the United States.

Where a turnout is seldom used, as for switches connecting spur tracks with a main line, a movable-wing frog is often used, which can be left-hand, right-hand or double. Normally, a spring holds the wing rail against the tongue of the frog so that there is no gap. When the turnout is used in the diverging route, the wheel pushes the wing rail aside. This reduces maintenance and gives smoother running. There were two accidents in 1966 where passenger trains moving at speed were derailed at movable-wing frogs in which the hold-down housings for the wing rails had failed due to poor maintenance and inspection. These were the derailment of No. 93, with 14 cars, at Hoopeston, IL on the C&EI at 60 mph on 11 April, and the derailment of No. 22, with 12 cars, at Peotone, IL on the IC at 80 mph on 31 May. There were no fatalities in either accident. The late appearance of such accidents may reflect the demoralization of the time.

The length of the switch point also controls the safe speed through a turnout. The familiar No. 10 was used with a 16' 6" straight point. With 39' curved points, 50 mph is possible with a No. 20 frog. The switch rail makes an angle with the stock rail, called the switch angle, and the wheels are suddenly diverted through this angle. For the No. 10 turnout with 16'6" points, the switch angle is 1°44'11". At the prescribed speeds, the shock at this point is small. At higher speeds, the pressure against the stock rail rises, and the rail may be overturned or the flange may climb the point, either of which is unpleasant. The Red Desert accident demonstrates that 50 mph is too fast for diverting a Big Boy 4-8-8-4 at a No. 10 turnout. However, such turnouts have been negotiated successfully at 30 or 40 mph.

A turnout may be operated by hand or power. In most accidents studied here, the turnouts are operated by hand by means of a switch stand, generally located about 9' from the center line of the track on the side towards which the turnout diverges. For main line switches, a "high" switch stand consists of a rotating vertical axis moving in a casting bolted to the ties. At the bottom of the axis, a crank is connected to the switch points through the head rod. At the level of the top of the casting, from 2' to 3' above the base, is an operating handle that is moved in a horizontal plane, usually through 90°. It can be moved vertically by means of a joint at the ends of its travel to fit into notches in the casting, which hold the handle fixed. There is some natural "spring" in the connections that press the points against the stock rails at either end of the throw. The handle is secured by a padlock whose hasp passes through a shackle that engages the handle. The padlock is attached to the casting by a chain, so it will not become lost. Employees authorized to handle switches are issued a switch key that fits in all the padlocks used by a company. At the top of the vertical shaft, at about 8 ft high, is a socket for a lamp, either oil or electric, with four lenses, two of one color opposite each other, and two of a second color opposite each other. A target is attached to the shaft and rotates with it, showing the position of the switch. Often, the target shows, for example, a red disc when the switch is set for the turnout, and nothing when the switch is set for straight-through. These two positions are customarily called "open" and "closed." It is better practice to have targets of distinctive shapes for open and closed, such as a white disc and a red arrow. When a switch is required to be "locked," it means that the padlock must be closed, so the switch key is necessary to unlock it. If it is required to be "fastened," this may mean that the hasp is to be passed through the shackle but not necessarily locked.

In Automatic Block Signal (ABS) territory, track circuits control the aspects of the lineside signals. Every main-track switch in ABS territory is equipped with a circuit controller directly connected to the switch points (not to the switch stand) that opens or shunts the track circuit when the switch point is moved 1/4" or more away from the straight stock rail. This operates the signals the same way that a train in the block would, and the signals governing the block that contains the turnout display a stop aspect. This is an extremely valuable safety feature of ABS. Where a switch stand is located in the vicinity of the block signals, it is not equipped with a lamp so that its light will not be confused with a block signal. The block signals serve to indicate the position of the switch automatically. If a block signal became extinguished, the green light on a switch stand might be taken erroneously as its aspect, especially at a distance. Switch stands away from block signals may have switch lamps, and, in fact, this is good practice. The colors of the switch lamps are generally red and green for mainline switches.

Where there is not a block signal in the immediate vicinity, the switch may be equipped with a block indicator, often a miniature signal in a case that shows Stop or Red when the block is occupied, and Clear or White when the block is unoccupied. Before unlocking and operating a hand switch, the employee must look at the indicator, and not unlock the switch if the indicator shows Block Occupied. When the switch is operated, the signals governing the block go their most restrictive aspects, protecting the turnout. If there is no block indicator, the employee is instructed to open the switch if no train is evident in the block, and wait a certain length of time to ensure that no train was entering the block as he opened the switch (see below). If a train is seen approaching, the switch should be quickly closed and locked again.

A switch may be equipped with an electric lock that prevents it from being operated when the lock is on. The lock may be controlled by an operator at a nearby office or by a signalman in an interlocking tower. The train crew member may have to telephone the office controlling the lock in order to have it released. This always occurs under CTC, at either hand or dual-control switches, and is a very good practice. However, electric locks were not generally used, even in situations that would seem to call for them, as at junction switches. Where the lock is not controlled by an operator, the trainman operates a handle inside a box. This sets the protecting signals at Stop, and after a certain delay, say 1 minute 40 seconds, lights a light showing that the lock has been released. At West Belt Jct, PA on the Pittsburgh & West Virginia, an accident occurred on 19 November 1936 at such a switch lock when a signal maintainer had altered the circuit so that the signals went to Stop simultaneously with unlocking the switch, not immediately.

In yards, the switch stands are generally of the low type with the weighted operating lever moving in a vertical plane parallel to the track, geared to the operating rod. A rotating lamp with colored targets surrounding the lenses displays the position of the points. The colors used for yard switch lamps may be red and green as for main-line switches, or yellow and white (Pennsylvania Railroad) or yellow and green. Yard switches may or may not be locked, since they are in constant use and speeds are low. On secondary tracks outside of yards, switch lights were usually not provided. Reflectorized targets were very effective, easily seen in the headlight's beam at night, and should have been used more often.

Hand-operated switches were very common on American railways. On main lines, they were operated by the crew members of the trains using them, who were responsible for seeing that they were left locked in the proper position. One type of turnout frequently used were the switches at the ends of sidings. A siding is a track, usually parallel to the main track, designated for meeting and passing trains. It is a special track, that must be kept clear of obstructions so it is always available for its purpose. Trains using it must move at restricted speed, prepared to stop short of any train or obstruction (another train may also be taking refuge there). Normally, the front brakeman and the rear brakeman (flagman) operated switches for their train. If the front brakemen was otherwise occupied (by flagging ahead, for example) the fireman handled the switches, while the conductor acted similarly at the rear end. On a passenger train, some other employee might have this duty, especially at the head end, since the front brakeman on a passenger train generally accompanied the conductor as he collected tickets. On a motor train with a three-man crew, the baggageman usually handled the switches, as well as flagging when necessary.

Spring switches were often applied to sidings, permitting a train to leave the siding without having to open and close the switch manually. These switches might have a white target with "SS" on it to identify them as spring switches that can be trailed through. This has now become a great convenience when trains no longer have anyone on the rear end to handle this job. Though spring switches may have been regarded with some suspicion, they have been remarkably trouble-free. The ICC reports only one accident due to a faulty spring switch, on the Virginian at Kumis, VA on 13 March 1941. In this case, dirty oil in the damper caused the points to fail to return fully after the switch had been trailed through. Although a distant switch signal 4000' in rear of the signal showed Caution, indicating that trains should stop and examine the switch before using it, the indication was ignored. The fireman was riding in the rear cab with the head brakeman enjoying the night (he had nothing to do, of course, on an electric locomotive, but add to the adhesive weight), and did not even see the signal when the brakeman noticed it but took no effective action. This stretch did not have automatic block signals, so a special 2-position upper quadrant semaphore was used to protect the switch, showing caution when the points were 3/16" or more open. In automatic block signal territory, a spring switch is protected by a block signal immediately in its rear (in the direction of trains approaching it on the main line). A sign, such as a letter "S" on a yellow disc, may show that it concerns a spring switch. These days, a high signal is provided in the same place outside of automatic block territory, and there may or may not be a distant signal as well. These signals apparently do a very good job, from the lack of reported accidents.

One sort of accident at spring switches is much more common than the failure of the switch itself, and occurs only in automatic block signal territory. When a hand-operated switch is opened, the circuit controller shunts the track circuit and the signals controlling access to the switch display Stop. It is normal practice to open the switch and wait 2 or 3 minutes (the Consolidated Code says 5 minutes, which is a bit excessive) to make sure that no train is approaching at speed on the main track. When spring switches are used, this 2 or 3-minute clearance does not exist, since the track circuit is shunted only when the train has already fouled the main line. Note carefully that the spring switch is not the cause of the accident; this can only happen when there has been some oversight and the train improperly enters the main track. It is simply that the backup system does not work. This can be cured by providing a leaving signal on the siding, or an overlap track circuit which the train could pull up into and wait the desired clearance time, or, of course, a switch indicator.

The Consolidated Code of Operating Rules (a detailed version of the Standard Code adopted by 16 companies) adds a paragraph to Rule 104 to cover spring switches: "104(H). Spring switches are identified by special targets, lights or both. A train or engine, stopping on a spring switch while trailing through and activating the switch points, must not make a reverse movement until the switch has been operated by hand and it is known that the switch points are in proper position for safe movement. When a train or engine moving in either direction is stopped by a signal governing movement over a spring switch, the switch must be tested by opening and closing by hand, and switch examined to make certain it is properly lined, locked or secured and that points fit before movement is made...." Automatic block system rule 513 also applies: "513. Before a train or engine enters on or fouls a main track, or crosses from one main track to another, at any switch operated by hand, it must wait five minutes after any switch connected with the movement has been operated to establish block signal protection." A list of exceptions follows, most of which are quite obvious, when there can be no objection to proceeding at once. This includes when a train has just been met, when a switch indicator or signals check clear, when there is a train or cars standing on the main track, or when authorized by a control operator under CTC, which will cover most cases when a spring switch is used at the end of a siding.

At junctions and terminals, switches might be handled by switchtenders designated for the purpose. At a small place, the operator usually acted as switchtender if there were junction switches. The time table usually indicated how junction switches should be lined when they were hand-operated. The siding switches would be handled by train crews as above. Sometimes a station helper handled the switches under the supervision of the operator. The good side of switchtenders was that if there was only one, there was no confusion as who should handle switches, and it was obvious from the physical location of the switch what it did. The bad side was the constant running about and danger of confusing hand signals, and the chance of unwanted "help" from other employees.

The correct way to handle such switches was, of course, by concentrating the controls in one place and interlocking them so that they had to be operated correctly. The interlocking was absolutely necessary since all levers look alike, and don't explicity show what they do. In fact, detection was introduced to prevent throwing a switch under a train. This was originally done mechanically, but the addition of track circuits made possible elaborate safety methods that not only prevented throwing a switch under a train, but also moving the switch immediately in front of it. For that reason, there are no accidents of the type we are considering around interlocking plants. In order to throw a switch, the signalman first had to restore the signal to stop, and by this time rationality would take hold. Curiously, the ICC never recommended interlocking, though it was an obvious solution to many problems, and usually improved operations so significantly that companies did it themselves without being told. Generally, however, interlocking was rather thin on the ground in the United States, where light traffic did not really justify it.

Places from which signals and switches are operated remotely are now called Control Stations, and the employee who operates them the Control Operator.


Since most of the accidents we are considering happened in ABS territory, it is necessary to understand how signals affect train operation. When there is only a single track, some provision has to be made for moving trains in both directions. Opposing trains must meet at some point, using a siding, as we have mentioned above. It must be understood that ABS signals do not control trains like traffic signals control cars. Of course, trains stop at a Stop signal, but they may or may not proceed at a Clear signal. The movement of trains was controlled by the timetable and train orders, and trains would move safely whether or not there were any block signals at all. The block signals are only an additional safeguard against lapses, especially improper flagging. Now, of course, trains usually move as they are directed by signals, under CTC, so it may be useful to explain how it used to be done. We will only consider the part necessary for understanding the accidents we are considering, as the whole is much too extensive a subject.

First, under timetable and train orders, the rear of a train was protected only (in theory) by flagging. Rule 99 stated that when a train was stopped or moving so that it might be overtaken, proper protection must be provided by means of fusees, torpedoes and a red flag or red lamp. Of course, efforts were made to prevent sending a fast train after a slow one, and keeping trains far enough apart that a flagman could get back when required. With light traffic, the system works fairly well, and only breaks down when traffic becomes dense and fast. That is, flagging works best where it is not really required.

A block system divides the line into intervals called blocks, and sees that opposing trains are not allowed into a block simultaneously. This is done by communication between operators at the ends of the block, and is called a manual block system (MBS). When a passenger train enters a block, no train is allowed to follow it until it has left the block at the other end. This is called absolute block, and is excellent protection. It eliminates the need for flagging if properly implemented, but in the United States the lack of sufficient clearance lengths and signals meant that flagging was still a very good idea, and MBS was also considered as only a backup to the timetable and train orders.

Freight trains were permitted to follow one another into a block, a practice called permissive block. The following train was warned, and proceeded prepared to stop short of a train or obstruction (in theory). It is strange that permissive block was necessary in view of the frequently light traffic, but long blocks made its use convenient. Block protection is very greatly diluted by the permissive block, and wise crews attended to their flagging. There are interesting cases of the simultaneous failure of timetable and train orders and the manual block system that indicates that when one fails, the other does at the same time, so that two half-measures do not make a full measure in safety.

Whenever there are two opposing trains on the same track, one is always superior, the other inferior, established by the timetable as modified by train orders. By timetable, first class (passenger) trains are superior to second class (freight) trains, and so on. All scheduled trains are superior to extra trains. Between trains of the same class, trains in a specified direction are superior to those in the opposite direction. Train orders may alter these relationships, and provide for the movement of extra trains, which cannot move at all on single track under the timetable alone. They most important type of train order specifies a meet between opposing trains: "No 5 Eng 4173 meet Extra 3030 East at Niland." No. 5 is, we suppose, a first-class train and is, therefore, superior to Extra 3030 East. It can move as far as the west siding switch at Niland without worrying about Extra 3030 East. The extra must approach the west siding switch at Niland under control, as it can go no farther on the main line. It stops, and its head brakeman climbs down and goes ahead to unlock and open the west siding switch, and signals the engineman to proceed. As the engine goes by, he climbs on the pilot and rides to the other end of the siding, where the train stops before reaching the clearance point, about 200 ft from the switch. At the rear end, the flagman swings down as the caboose passes the west switch, which he closes and locks. He then walks forward to the caboose and rotates the markers to show green to the rear. The front end was notified that the rear end was in the clear by hand signals or a fusee before the days of radio, often a difficult job on a long train. The engineman then extinguished the headlight, showing that the train was clear of the main track.

Let's suppose this is ABS territory, and consider what the signals were doing while Extra 3030 East was arriving at Niland. At each end of the siding is a pair of high signals, one for each direction, located about 70-100 feet from the siding switches. Let's say they are two-position, lower-quadrant home signals. As Extra 3030 approaches, the signal on the right is at Proceed, while the one at the left shows Stop to westbound trains. When the brakeman opens the switch, the signal on the right goes to Stop. At the far end of the siding, the eastbound signal shows Proceed, while the westbound signal shows Stop. a few thousand feet farther east, the distant signal shows Approach. When the rear of Extra 3030 is past the signals, the westbound signal now goes to Proceed, since the block is clear. When the switch is closed, the eastbound signal clears from Stop to Proceed. At the far end, the westbound signal now goes to Proceed (it had to wait until the switch was closed). Now all the signals show Proceed. Actually, the signals may be approach lighted, which means that they show Stop and the lamps are off until a train shunts the track circuit in their rear. In this case, all these signals will be at Stop until No. 5 is a few miles away, and then will clear, showing that No. 5 is approaching.

No. 5 will see the distant signal at Proceed, which will indicate that Extra 3030 either has not yet arrived, or is in the clear on the siding. Probably a light brake application will be made preparatory to stopping at the west end of the siding, but as soon as Extra 3030 is seen on the siding with its headlight off, No. 5 will carry on at full speed. If there were no ABS, a definite reduction in speed would have been made approaching Niland, since it would be necessary to stop if Extra 3030 were not seen in the clear. If this were a timetable meet (that is, if Extra 3030 were moving with respect to No. 5's schedule), Extra 3030 would have to be in the clear at least 5 minutes (or whatever the rules of the company required) before No. 5 was due, or be protected by flag.

With No. 5 safely by, the head brakeman goes to the east siding switch, unlocks it, and opens it. This sets the westbound signal to Stop if it has already cleared, and may approach light the eastbound, or leaving, signal, which is at Proceed. He rejoins the engine as it passes him. At the rear, the markers are restored to show red to the rear. The flagman swings down as the east switch is passed, which he closes and locks. The westbound signal clears, while the eastbound is now at Stop behind the train. When he is again on the caboose, the head end is signaled, and the train proceeds. This process has been described in some detail to show the duties of each person involved.

Although the conductor and engineman are jointly responsible for the safety of their train, it is a sad but true fact that, especially on passenger trains, the conductor concerns himself with tickets, and leaves running the train up to the engineman. The safety of timetable and train order operation depends on the cooperation of conductor and engineman, and when this cooperation is absent, safety suffers. Often, the conductor delegates his responsibility to the rear brakeman or flagman, usually a reliable man, which has prevented many an accident.


The earliest example of irrationally throwing a switch in front of an oncoming train so far found in the ICC reports is the accident at Amboy, MO on the Van Buren District, Central Division, Missouri Pacific Railway, on 15 Dec 1923. This was a single track line operated under MBS rules. Northbound Train No 106, with engine 5508 and 6 cars, had entered the south end of the siding at Amboy for a train-order meet with southbound No 103, with engine 5524 and 7 cars, including a business car in which General Superintendent Brooks was riding. No 106 stopped a few carlengths south of the clearance point at 9.45 pm. No 103 was 3 hours 13 minutes late, and was running rapidly, passing Marche, 6.16 miles north, at 9.43 pm and arriving at Amboy less than 7 minutes later. The meeting point and station whistle signals were sounded, and speed was reduced to 30 mph or so approaching the meeting point.

When he was 8 or 10 poles away, the engineman of No 106 noticed that the switch light was green. A man with a uniform cap and a hand lamp on the eastern side of the track gave him a highball, which he answered with 2 short sounds, and then opened the throttle. When he was 3 or 4 car lengths from the switch, the switch light suddenly changed to red, and he applied the brakes in emergency, reducing speed to about 25 mph at the instant of collision, at 9.50 pm. The enginemen of No 106 were killed, as well as one trespasser of whom we hear no other details.

The man who was responsible for the switch was Train Porter Talley, who had been employed for two years and had a satisfactory record. He was riding on the engine of No 106 when it stopped, and went to the switch. The switch lock was found lying on the headblock tie, so he had unlocked the switch preparatory to lining the switch for his train to leave the siding after No 103 was by. This, of course, is strictly against the rules. Rule 104 provides that the switch shall remain locked until the train to be met is passed. The evidence of the other crew members show that Talley was not supervised or instructed, and all but ignored in the critical minutes.

Talley denied that he was ever near the switch, never gave a higball, and was well up on the bank to the east of the track, on the opposite side to the switch stand. The ICC examiner totally discounts his claims, which conflict with all other testimony. There simply was no one else in the vicinity to unlock and line the switch. He was seen on the bank with his cap and hand lamp, gave the highball, then ran down the bank, across the track, and threw the switch. He would have had about 27 seconds to do so, and would have been out of sight of the engineman of No 103 (behind the boiler). Malice can certainly be ruled out, and it was probably an irrational action of the kind of which we have many other examples. From the comments in the report, it seems that the irrational nature of the accident was not quite appreciated at this date.

A curiosity in the report is that the engineman of No 103 said that the headlight of No 106 was dimmed, which normally indicates that the train is still fouling the main track. The headlight should have been extinguished in this case, but no comment is made on the subject.


Tortuga, CA is about halfway between Indio and Yuma, in far southeastern California, on the Salton Subdivision of the Los Angeles Division of the Southern Pacific. At 1.36 am on 20 September 1938, westbound Train No. 5, the Argonaut, unexpectedly entered the siding and collided with stationary Train No. 44, the Californian. Except for the engineman of No. 44, who jumped just before the collision, the engine crews of both trains were killed, as well as eight passengers in the forward end of chair car SP 2418, the third car of No. 44, where the standardweight baggage car RI 4120 had overridden the end sills and telescoped the end of SP 2418 by 18 feet. SP 2418-1419 were a lightweight articulated set, built about a year earlier. The major concern of the ICC report was the strength of the new lightweight Cor-Ten steel cars, not the cause of the accident.

The line was equipped with ABS, about as described in the example of a meet given above. The speed limit was 65 mph, the grade was slightly falling westward, and the track was tangent (straight). No. 5 was superior by direction to No. 44, and they usually met nightly in this area, but normally No. 5 was made inferior to No. 44 by train order, and it would take siding. On this night, as usual, No. 44 received Order No. 473 at Indio: "No 44 has right over No 5 Indio to Araz Jct." The Southern Pacific did not use engine numbers for identification of trains, since all engines carried a train indentification board with the train number displayed. A later order, Order No. 405, had the effect of reversing the precedence: "No 5 meet No 44 at Tortuga and No 4 and No 6 at Iris No 44 take siding." No 44 received this order at Niland, 11 miles west of Tortuga. No 5 received the orders at Yuma and Glamis, respectively. The orders were understood and executed properly by all concerned, and have nothing directly to do with the accident. The reason for these changes of precedence is not stated in the report, but presumably there was some good reason.

Head Brakeman Jacobson of No 44 exhibited no unusual behavior that was noticed by his colleagues. He had had 18 year's experience and was reliable. He was also well-rested, since the report says that he slept for 94 hours [sic] the night before, and had had a nap of 1-1/2 hours in the afternoon. He rode the pilot of the engine from the west siding switch until it stopped 232 ft from the east siding switch, about a car length from the clearance point. The engineman alighted to oil around, and Jacobson apparently walked forward to the switch, as was his habit. Normally, No 5 took siding for No 44, so he would unlock the switch and set it for the siding. This was normal and proper. The action would set the westbound signal at Stop, and the distant signal 2069 ft away at Proceed with Caution (arm horizontal, yellow light). It is an SP peculiarity to name this aspect Proceed with Caution instead of Approach, but it meant the same thing: reduce speed immediately and prepare to stop at the next signal.

However, this night there was no need to open the switch, and, testimony establishes that Jacobson did not do so. He did, however, unlock the switch, which was in direct violation of Rule 104(C), which prescribed that the switch was to remain locked, and he was not to go within 20 feet of it. The signals were observed to both show clear when No 44 stopped, and the eastbound signal to display Stop as soon as No 5 entered the block over two miles away. When No 5 was nearing the distant signal, the engineman gave the whistle signal - - o for the meeting point, after the conductor had prompted with a single sound on the communicating signal, as was proper, according to Rule S-90. The speed of No 5 was then reduced from 60 to 35 mph, in preparation for identifying No 44. The home signal at the siding was observed to display Proceed.

As No 5 approached the home signal, the conductor of No 44, watching from the rear, observed the signal to go to Stop (the report says eastbound signal for westbound signal here, an obvious mistake) and thought No 5 had passed the signal. What had actually happened is that Jacobson had suddenly thrown the unlocked switch as No 5 was only a hundred feet or so away. The engineman of No 5 gave some whistle sounds and put the brakes in emergency, but the brakes could hardly have taken effect when No 5 had entered the siding through the No. 10 turnout and collided with No 44, driving it back 88 feet. The engines and front cars of both trains were derailed, but the rear cars were undamaged and remained on the tracks.

From the evidence in the report, it is clear that Jacobson had no idea why he had done what he had done. He said that he had gone to the switch and opened it as usual several minutes before No 5 arrived, out of habit, which he could not have done, since the signals testified to the contrary. The switch was found lined and locked for the main track after the accident, and only Jacobson could have done this in his terror. We mention these things not to accuse Jacobson, but only to show that he had no idea why he had opened the switch, and tried to find some logical explanation. Actions such as Jacobson's will be described here as "irrational" to distinguish them from carelessness and confusion, which also cause open-switch accidents.

In the 1942 revision of the SP Rule Book, Rule 104(C) was changed to require trainmen and others not to go closer than the clearance point (about 190 ft) until the train to be met has passed, while the switch was to remain locked. When this rule is respected, accidents such as this one cannot occur. The ICC makes no such recommendation in its report, which concerns itself with Cor-Ten steel instead of the causes of the accident and their remedy.


An accident four years earlier, on 3 August 1934, at Manteca, CA on the Merced Subdivision of the Western Division of the SP attracted less attention. This line in the Valley of California was single track, with ABS. Manteca is near the northern end of this stretch, and has a 4000' siding with No 10 turnouts at each end. There are signals for both directions at each end of the siding, with home and distant arms. Engine 2384 was working extra near Lathrop with six cars and a caboose when it took refuge at Manteca. It had to protect against two westbound extras, and clear first-class No 55 while switching there. Brakeman Garza was detailed to this task, and went to the east switch. Although he had had 14 years' experience, testimony showed that he was totally unprepared for his task, especially as his conductor was elsewhere and had only discussed the extras with him, not No 55. He seemed totally unaware of the existence of No 55 as he idled in the vicinity of the unlocked siding switch, clearly violating Rule 104 and not giving effective protection against the expected extras.

Section hands were burning weeds along the track to the east, and the smoke may have obscured the westward signal. Whatever the case, No 55 had received nothing but clear signals and was moving near the maximum speed for passenger trains, 65 mph, as it had every right to do. Its timetable rights had not been restricted, so it had no need to concern itself with such inferior objects as a work extra. As it approached Manteca, the signal abruptly went to stop and the brakes were applied in emergency, reducing the speed to 50-55 mph before the engine derailed and overturned as it attempted to enter the siding. It was 10.30 in the morning, and the weather was clear. Three employees, among them the engine crew of engine 2401, were killed, along with one tramp.

Garza made several attempts to give a reason for his actions, among them that he had opened the switch so that the signals would slow down the expected extras, and that he had no idea that No 55 was due. Nevertheless, it was clear that the switch had been thrown at the last moment, not earlier or with any calculated purpose, but irrationally. The ICC report dwells on Garza's incompetence.


Ranch, TX was on the Western Division of the Texas and Pacific Railway, on the sandy plains west of Odessa. There was no town there, just a siding. At 7.05 pm on 30 December 1947, Extra 657 East, standing on the siding, was run into by westbound No 67, Eng 622, running at 55 mph. Three of the four men on the engines were killed. The weather was clear and calm, though darkness had just descended. The line is tangent for about 12 miles to the east and 4 miles to the west at Ranch, descending westward at about 26 feet per mile. The line was equipped with 3-position color-light, APB automatic signals, and there were block indicators at the switches.

A block indicator is a miniature signal or lights placed near a switch, at the clearance point of a siding, or a motor-car turnoff in ABS territory to show when a train is closely approaching. An illuminated light would mean that no train was approaching, so that a nearby switch could be safely opened. The open switch would then be protected by the block signals on either side of it. If the light was extinguished, it meant that a train was approaching, probably beyond the protecting signals, and the switch should not be unlocked. If there are signals nearby, a block indicator is usually not necessary, but may be provided if necessary. The block indicators on the T&P at Ranch were miniature color-light signals, showing green (clear) and red (occupied).

Order 98 prescribed: "No 67 eng 622 wait at Ranch until seven nought five 705 pm for Extra 657 East," and was delivered to No 67 at Odessa, and to Extra 657 at Monahans. Extra 657, 22 cars and caboose, entered the siding at the west end and stopped short of the clearance point at the east end at 7.02 pm. This may have been cutting the time a little close, but Extra 657 was in the clear five minutes before the time on No 67 expired. The headlight of No 67 was seen in the distance, rapidly approaching. The front brakeman of Extra 657 checked the block indicator at the clearance point, and noted that it was red, as expected. No 67 had clear signals at signal 597.9 about a mile and a half to the east, and at signal 599.5 at the headblock. The eastbound leaving signal was at stop when Extra 657 stopped, since No 67 was long past the next siding. The engineman of No 67 sounded the station whistle (one long) and the meeting whistle (two long and one short), and received a highball from his rear, since Extra 657 was seen in the clear and the waiting time had just expired. The head brakeman of No. 67 came out of his cabin on the tender and saw both signals at green. All of this was completely regular, and showed snappy operation.

The brakeman of Extra 657 walked to the siding switch, unlocked it, and threw it in the face of No 67, before there was time for the speeding train to make any response. The big 2-10-4 with 55 cars and caboose swerved through the turnout and collided with Extra 657 225 feet from the switch. Of the engine crews, only the fireman of Extra 657 survived, jumping in the nick of time. There was no explanation for the brakeman's actions. The rules of the T&P provided that the siding switch should not be unlocked until the train to be met had passed, and that trainmen approach it no closer than 30 feet, and stand on the opposite side of the track from the switch stand when practicable. This rule was clearly violated.

We see that ABS gives no protection against this kind of error, and even ATC would not have helped. The ICC seems to have realized this, and recommended the installation of electric switch locks. This recommendation was not carried out, neither here nor elsewhere, but it was indeed valid.

An electric switch lock is a simple device, and had been available for many years when these accidents occurred. A lock rod attached to the points passes through a box where a notch in the rod mates with a dog that prevents the points from moving. If one presses on a treadle (for example) an electrical circuit is completed that energizes a solenoid that raises the dog out of the notch, and the switch can be thrown. The circuit is completed through normally-closed contacts on a relay energized by the track circuit approaching the switch. When this circuit is occupied by a train, the treadle cannot complete the circuit, and the switch cannot be thrown. This simple circuit requires several modifications to make it practical. There should be a time release that will reclose the circuit to the lock relay after the track circuit has been occupied for a certain length of time, such as two minutes for a 1-mile circuit. This will guarantee that the switch cannot be operated in front of a train at speed, but will free the lock when a train stopping at the switch arrives at it. Such a time release would now be implemented electronically, but an electromechanical release has long been available. In case the mechanism fails, a seal could be broken to gain access to the lock case, and the lock lifted directly. It can easily be arranged that a screw would have to be turned for a minute to open the box. An electric lock such as this is much cheaper than CTC (which has equivalent protection), and not excessively inconvenient, but it was never generally used, as far as I can determine.


Melbourne, IA was on the Chicago Great Western 5th District, Iowa Division, between Marshalltown and Des Moines, where it crossed the Milwaukee Road. There were automatic block signals, with 2-position upper quadrant home and distant signals, approach lighted and overlapped. The maximum speed for passenger trains was 60 mph. At 12.37 am on 17 February 1943, eastbound freight train No 62, standing on the siding expecting to be passed by eastbound passenger train No 12, as well as to meet westbound passenger train No. 16, was run into by No 15. No 15 had managed to reduce its speed to 20 mph at the time of collison, and there were no fatalities. The engineman of No. 16 saw the signal go to stop when the switch was opened, but thought it was No. 12 at the other end of the siding, to which he had rights. When the position of the switch was seen, however, an emergency application of the brakes was initiated.

No 62 stopped on the siding 349 ft from the clearance point at 12.05 am, about opposite the station, which was near the east switch, to clear No 12. It also held order No 5, reading: "No 62 Eng 876 take siding and meet No 15 Eng 930 at Melbourne instead of west passing track Marshalltown." It seems that No 62 had previously been given right over No 15, since under normal conditions it would have taken the siding by rule, as the inferior (second class) train. The head brakeman of No 62 said he opened the switch expecting No 15 to enter the siding after his own train had backed to the west end of the siding to make room, so that No 12 could get by both trains. Of course, it would have been much better for No 12 to take siding at the west switch and let No 15 by, since No 12 had no rights beyond this switch, as the engineman of No 15 well knew. Any such procedure as the brakeman proposed would have required a conference with the crew of No 15 as well as some unadvisable backing, so it appears to be only an excuse concocted after the fact. This accident was probably also due to irrational action, though confusion cannot be ruled out.

The ICC recommended electric switch locks in this investigation, possibly for the first time.


Prosser, WA is between Pasco and Yakima on the Northern Pacific main line. The line was single-track, with APB. Early in the morning of 23 September 1945, eastbound 1st No 2 collided with westbound 2nd No 1 at the west end of the siding. 1st No 2 had been moving at about 60 mph with clear signals as it approached Prosser siding, but the signal changed suddenly to Stop when the oncoming train was 150' away. An emergency application reduced speed to about 30 mph at the point of collision. The fireman of 1st No 2 was the only fatality.

The train order establishing the meet said: "Second No 1 take siding and meet First No 2 at Prosser. Second 1 gets this order at Prosser." This was quite proper, since the siding at Prosser is west of the station, and Second 1 would receive the order during its station stop, before reaching the siding. The conductor himself lined the east siding switch and let his train into the siding, while the head brakeman walked forward through the 12-car train. He alighted at the front of the train and walked to the west siding switch, violating the rule that he stay 20' away and on the opposite side of the track from the switch stand. He unlocked the switch and threw it in the face of 2nd No 1.

The brakeman claimed that he thought his train was on the main line, and was opening the switch for 1st No 2 to enter the siding. However, since he only opened the switch when the rapidly-moving 1st No 2 was very close, according to the testimony of its engine crew, and the aspects of the signals, this is only an attempt to make sense out of his unexplainable act. The ICC recommended switch locks in this case.


Canaan, NH was on the Boston and Maine line from White River Junction to Concord. This line was equipped with ABS. At 2.34 pm, in clear weather on 19 August 1949, southbound No. 332 was stopped on the passenger siding at Canaan station when it was run into by northbound No. 307. This was a train order meet, established by orders No 23, "No 332 meet No 307 at Canaan" and No 24, "No 332 take siding at Canaan to meet No 307." No 332 was the superior train, and would have held the main line, but the superiority was reversed by train order for some reason, possibly because No 307 did not stop at Canaan. However, the train orders were properly executed and in no way contributed to the accident. The south switch was to be handled by the baggageman of No 332, as is often the case on passenger trains when the head brakeman accompanies the conductor. As No 307 approached at speed, the baggageman proceeded to the switch, unlocked it, and threw it in the face of No 307, which was running under clear signals. He could not explain his actions. Although he normally would have opened the switch for No 307 to take the siding, he certainly should not have done this with No 307 closely approaching, but rather when his train had first arrived. The speed of No 307 was low enough that there were no fatalities, but there were numerous injuries. This accident did not occur at night, which indicates that mere darkness is not an explanation.

The ICC again recommended electric switch locks, as at Ranch, Melbourne and Prosser.


The accident at Robinson, NM on the Santa Fe, 5 September 1956 is the most famous accident of this type, perhaps because of its high death toll. Most of those killed were dining-car employees riding in a lightweight dormitory car, ATSF 2318, built in 1937 and strengthened in 1950, that was destroyed. The accident will only be sketched here, since it has been described elsewhere in detail.

This part of the New Mexico Division was a single-track line with overlap ABS at the time of the accident. Double stop aspects were used to ensure that opposing trains were kept apart. Eastbound No 8, the Fast Mail, was stopped on the siding for a train-order meet with westbound No 19, the Chief. The fireman of No 8, an experienced man who had been promoted to engineman, had walked forward and unlocked the switch following the regrettable practice of so many trainmen, to be ready to let his train out when No 19 was by. No 19 had received only Proceed signals approaching Robinson and had only briefly checked its speed while identifying No 8. It was accelerating when the fireman threw the switch in its face. At 3.01 am it collided with No 8 on the siding at a speed of about 63 mph. The fatal dormitory car was the second car of No 19, which was derailed and crushed between its heavier neighbors.

The Santa Fe rule book provided that a trainman go no closer than the clearance point, and not unlock the switch, until the train to be met had passed. This rule was clearly and deliberately violated. The engineman of No 8 seems to have made some motions to warn the fireman away from the switch, but as he died in the collision, we do not know for sure. The fireman had no explanation for his actions, though the usual attempt was made to lay the blame elsewhere. The clear violation of rule 104 made him fully culpable for the results and the deaths of 20 men.

In this accident, and the other similar ones, the most important contributing cause is the lack of discipline among trainmen and officials. Every senior trainman should carefully impress on new brakemen that this rule must not be violated. However, hanging around the unlocked siding switch is a common habit of front brakemen, and the Santa Fe trainmasters had obviously made no point of correcting this behavior.

Carpenter and Huntington

We now consider two accidents, one from 1927 and the other from 1950, that have the common feature that an inexperienced person throws the switch, though in both cases a regular trainman, after a senior trainman gestures to him to get away from the switch, which the inexperienced person interprets as an instruction to open the switch. These accidents can be ascribed more to confusion than to irrationality.

Carpenter, MI was on the Michigan Central's Detroit-Bay City line, which operated under manual block rules. Southbound freight Extra 7826 South, with 32 cars and caboose, had stopped on the siding 300 ft from the south switch in the clear autumn morning of 18 September 1927. Northbound passenger train No 203, with 8 cars, was approaching at 60 mph on a clear block. Brakeman Huebler, on only his second trip, unlocked the switch, apparently unaware of the MC rule that said the switch should remain locked, he should be no closer than 20' to it, and on the opposite side of the track. He leaned on the switch stand, preparing to let his train out of the siding when No 203 had passed, and acting like very many brakeman unfortunately do. When No 203 was no more than 100' away, he saw his engineman making repeated side-to-side gestures at him. He opened the switch at once, and at 9.18 am No 203 buried itself in Extra 7826 South. By some miracle, no one was killed except the baggageman of No 203, who perished in the wreckage. Brakeman Hueber had been issued a timetable and a book of rules, but had not read them and was unacquainted with the rules. It is the conductor's responsibility to know that his brakemen understand what to do, so he was equally culpable in the event. The ICC made no recommendations in this investigation.

Huntington, NY is near the middle of the Long Island Rail Road's single-track line from Hicksville to Port Jefferson on northern Long Island, which was operated under manual block rules. On the fine, hot afternoon of 5 August 1950, another fresh brakeman was acting as Brakeman Huebler had 33 years earlier, and had unlocked the siding switch to let his freight train out when an opposing passenger train had gone by. His train, Extra 101 West, with 20 cars and caboose, had been switching for some time at Huntington, and intended to go out on the main after No 642, Eng 29 and 8 cars, had passed. His conductor, seeing him at the switch, motioned for him to get away by sideways gestures. Earlier, he had told the man not to line switches until he instructed him to do so. The brakeman opened the switch in the face of the approaching passenger train, which collided with his own engine at 4.11 pm. There were no fatalities, but a goodly number of injuries. The ICC made no recommendations about switch locks, apparently having decided not to waste its breath.

In both of these accidents, signals would not show when a switch was open, because the ordinary manual block system does not include track circuits. In this case, it would be even more important to provide switch locks, perhaps operated by the signalmen, to safeguard trains from open switches. However, a primary feature of these accidents is the misinterpretation of gestures by an inexperienced trainman.

The Red Desert, WY accident on the Union Pacific in 1953 was similar, except that the inexperienced employee was a Puerto Rican sectionhand who did not understand English. He was somehow assigned to a mainline switch during the passage of a flock of sheep across the line for a reason that is not clear, and threw it in the face of a Big Boy while the section foreman was yelling at him. Shaw, or the website, will furnish further detail, but it does not contribute much to our present purpose.


Napier, MO was on the Chicago Burlington and Quincy line from St. Joseph to Pacific Junction. From St. Joseph to Napier the line was double track with ABS, and from Napier north the line was single track with MBS. Napier is the junction for the Wymore Division, which led northwestward across the Missouri to Wymore, NE, where one line went northward to Lincoln and one westward eventually to Denver, both "back" routes. In clear weather at 4.25 pm on 2 October 1939, northbound train No 21, whose equipment was the Pioneer Zephyr, No 9900, collided at high speed with southbound freight train No 92, Eng 4973, just north of the junction switch on the Wymore Division. The engineman of No 21, and a road foreman of engines riding in the cab, were the only fatalities.

The double-track line makes a 1° curve to the right just south of the station at Napier. The No. 11 junction switch to the Wymore division is located on the southbound track, so it is a trailing switch for trains moving with the current of traffic. The superelevation of the curve is opposite to what is required for the divergence. There are separate train-order signals at the station for the main line and the Wymore Division. North of the station, a crossover leads from the northbound main to single track, while the northbound main continues as a siding. The junction switch is normally operated by the station helper, as directed by the operator on duty, and no other person is authorized to do so. There is a yard-limit sign about a half-mile south of the junction. The junction switch was not protected by any kind of signal for a facing movement. This arrangment is quite safe under normal conditions, and had not been the site of accidents except for this one notable exception.

On this day, No 21 was running against the current of traffic because the northbound track was out of service for maintenance as far as Forest City, 5.9 miles south of Napier. The movement was authorized by order No 34: "No 21 motor car 9900 has right over opposing trains on southward track Forest City to Napier." First-class trains can generally ignore yard limits, even when there is no ABS (Rule 93), but yard limits cannot be ignored when moving against the current of traffic, since yard movements would not expect the trains on the wrong track. The Burlington rules did not expressly state this, although Rule 93 now contains this provision explicity. Also, Rule 98 provides that trains must approach junctions prepared to stop unless the switches are properly lined and the track is clear.

No 92 arrived about 4 pm, and while the engine was taking coal and water, its head brakeman took it upon himself to line the junction switch for the Wymore Division, in preparation for his train's departure for St. Joseph. When No 21 was observed approaching at speed against the current of traffic, there was some running towards the junction switch, but No 21 got there first. No 21 had crossed over at Forest City at 4.20 pm. Its speed had been about 75 mph approaching Napier, and was only slightly checked at the yard limit sign. The track appeared clear ahead, so the throttle was opened again. Then, finally, the red target on the junction switch was seen, and the brakes applied in emergency. The speed was reduced to 45 or 50 mph in the 859 feet available before 4973 was encountered.

The principal cause of this accident was failure to observe Rule 98. There was no signal protection, so No 21 should have prepared to stop at the junction until the switch was seen. Since it was in an improper state, No 21 should have been controlled so that it could stop short of the switch. Since the junction was within yard limits, Rule 93 could also have prevented the accident. When No 21 was occupying the southbound track, safeguards that normally existed were nullified. The trailing junction switch became a facing junction switch, without signal protection. This should have rung bells somewhere, but it did not. A contributing factor was the unauthorized operation of the switch by the brakeman. There is a difference between junction switches and siding switches that was not obvious to him. Since this switch found itself a facing switch so often (movements against the current of traffic were frequent), it should have been electrically locked under the control of the operator. The ICC examiners made no useful comment, and buried the prime reason for the accident in secondary matters.

Much of the report dealt with the damage to 9900. Lightweight equipment always received extensive discussion in these reports, though it had nothing to do with the causes or remedies. The Pioneeer Zephyr came through the analysis rather well, and its integrity was one reason there were no passenger fatalities and only 25 injuries.

This accident did not involve irrational actions nor confusion, but negligence in conforming to simple, basic rules that would have provided safety if observed.

East Point

The accident at East Point, GA on 23 October 1929 also illustrates negligence, not irrational action. East Point was on double track, about 6 (or 10 miles, according to the report) from Atlanta, near where the Central of Georgia line from Macon joins the Atlanta and West Point for the run into Atlanta. This line had ABS as far as East Point, and MBS beyond. Near the point of the accident, on the eastbound track, there was signal A-74, normally at caution for the beginning of MBS ahead. Further on, there was a crossover to a siding to the south of the eastward main track, with industry spurs leading off of it.

A&WP engine 231 was working extra on the eastward track, and had been switching on the siding. The engine was kicking a cut of 4 cars westward on the siding at about 6 mph when Central of Georgia Train No 18, the morning local from Macon to Atlanta, with engine 424, a baggage car, and two coaches ran through the crossover and struck it at 9.49 am, in clear weather, moving at about 15 mph. There was no derailment, and only minor injuries to trainmen and passengers.

Brakeman White of Work Extra 231 was an experienced trainman. He conceived of saving a little time and effort when he was in the vicinity of the crossover while the switching movement was prepared. He said he knew that No 18 had to be cleared, and looked at his watch, and thought it read 10.46, meaning that No 18 had passed. Even if it were 10.46, he had no reason to make such an assumption. He could only have concluded that No 18 was by if he had actually seen and identified it, which he had not. He opened the crossover switch on the main track, since he knew his train would be going out soon. Again he did wrong: both switches of a crossover should be opened or closed at the same time, not left one open and one closed. We must presume that the running orders of Work Extra 231 specified "not protecting against extra trains," since otherwise the open switch would have to be protected by a flagman. The ICC report does not say.

No 18 was passing signal A-74 at caution at this very moment, and was in sight, if brakeman White had looked, which he did not, committing a further error. Had he thrown the switch a minute earlier, the signal would have been at Stop, and No 18 would have stopped. In this case, the switch was opened inadvertently and negligently in front of the train, not irrationally in panic. The trainman was not inexperienced, but possessed that peculiar type of experienced ignorance so often found, as with Brakeman Garza at Manteca.

No 18 was the train number on A&WP tracks, suitable for an eastbound train; the Central of Georgia number would have been an odd number (probably No 11), appropriate for a northbound train on this road.

Moore and Henryetta

Another type of open switch accident is represented by the derailment of Louisiana and Arkansas Train No. 3, with engine 392 and 7 cars, at Moore, LA on 28 September 1936. The engine crew was killed when the engine overturned. This single-track line did not have ABS, so the position of a switch ahead was known only from the color of the switch light, and the continuity of the reflection of the headlight in the rails. As No. 3 approached the switch at 11.50 pm, running at a reduced speed of 40 mph, the green light was first seen, but as the engine drew closer, the reflection from the rails showed that all was not right. The switch was "cocked," that is, the points were slightly moved from the stock rail so that the wheel would drop off the rail, and a derailment was guaranteed. The switch lock had been smashed, and the lamp forced so that green would still be shown.

The malefactor was seen running for the woods in the darkness, shielding his face, and was never apprehended. No motive was apparent for the attack. This type of accident is sabotage, and does not involve throwing the switch immediately in front of a train. The saboteurs are normally either those with a grudge against the company (not against their victims on the train), or children and youths of a bad nature. As we have seen, although there may be considerable property damage, there is only a small chance of personal damage, and that mainly to employees and not to passengers.

A more costly accident due to sabotage happened at Henryetta, OK on the Creek Subdivision of the Southwestern Division of the St. Louis-San Francisco Railroad on 18 August 1929. Northbound Train No. 118, running at about 60 mph, derailed at a switch to a spur track within yard limits nearly a mile south of the station. The switch had been "cocked" as at Moore, and the wick of the lamp had been turned down. The engineman and fireman were killed when the cab was ripped from the overturned engine, and also 12 passengers in a steel-underframe (wooden body) coach.

The switch lock and chain were found by some boys where the saboteur had dropped it. The lock was closed and undamaged, which suggests that the saboteur had a switch key. No hint as to the saboteur or his motive was discovered. ABS was installed on the Creek Subdivision the next year.


The fatalities in these accidents were overwhelmingly the engine crews, especially those of the diverted trains. There were generally no passenger fatalities at all, except at Tortuga and Henryetta, where the derailed cars got out of line and penetrated each other. The low toll is evidence of the inherent safety of railway travel in strong cars that remain in line in an accident.

Accidents at a misplaced switch may have very different causes. Accidents in which the switch has been carelessly lined and forgotten are very few, since operating rules and ABS protect against these lapses very well. We have discussed one example above (Napier) and explained the reasons for its occurrence. Another clear-cut case is that of sabotage, making use of the weakness of a split switch that results when the points do not fit correctly, and the switch is said to be "cocked" (Moore and Henryetta). ABS largely prevents this sort of crime. It also happens that an inexperienced man throws a switch when he misinterprets a signal as an instruction to do so (Carpenter and Huntington, and Red Desert, as well). This merges into the case of greatest interest, when an experienced man inexplicably opens the switch against all reason when a rapidly approaching train is close. The number of examples shows that this is a real psychological phenomenon. It is not favored by any particular time of day or inclement weather or by state of rest. The examples are from all times of day, usually in clear weather, and the trainman involved is sufficiently rested. All of the examples found also occurred in ABS territory, where the signals protect a misaligned switch, unless it is misaligned immediately in front of a train.

In such irrational cases, there is always an attempt to make some rational explanation, which falls short and is contradicted by evidence. These excuses may instance force of habit or misapprehension of location or type of train approaching. The fact is that no one rationally throws a switch in front of a moving train in the manner required; this is no flying switch or other such maneuver. The most reasonable attempt is to say that the switch was thrown to prevent an accident, under the misapprehension that the switch was other than it was and is being restored to its proper position. This may seem plausible, but it is probably only a fiction. One can normally tell how a switch is lined in an instant, and never mistake its position. There is probably some mental reason that leads to this sort of action, and it is called up strongly by the conditions of a meet, where the train observed is moving rapidly. It may also be of importance in other human endeavors, but unrecognized.

Additional apparatus, such as electric switch locks, may prevent such accidents, and such installations may be justified. It is really disturbing to realize that a train may be moving under clear signals, and yet not guaranteed a safe route. Under CTC, misplaced switches cannot occur since trainmen do not handle switches except under stringent precautions. However, they are also prevented by conforming with the paragraphs of Rule 104 that prescribe that a siding switch remain locked until the train expected has passed over it, and that trainmen remain a certain distance away and on the opposite side of the track if practicable. Having to jump across the track in the path of an approaching train discourages such actions, of course. This rule is easy to understand and respect, but we see that it was constantly flouted, and perhaps the reason for it is not clearly understood.

All of the analysis here is on the basis of conditions existing when the accidents discussed occurred, not under conditions currently existing, where methods of operation and rules are different, passenger trains have largely disappeared, and in many places trains run more slowly and less often (or not at all). Radio communication and the lack of crew on the rear ends of trains also make significant differences. The lessons taught, however, remain valid.

A Recent Accident

Another kind of open-switch accident is not a case of unexpected diversion, but simply a violation of Rule 104, where a switch is left in an incorrect position. Serious consequences are avoided in most of these accidents because of the protection of turnouts by block signals, or even by seeing the switch target at a sufficient distance. Both lighted and reflectivized targets should be considered equally effective at night.

A more serious open-switch accident occurred at Graniteville, SC (13 miles N of Auguta, GA on the line to Columbia) on 6 January 2005 on the Norfolk Southern, in non-ABS territory. A local freight entered the siding around 7 pm and remained there with the switch open. A following through freight train went through the switch that night and collided with the stationary train. The engineman of the through train died as a result of the carelessness of the conductor of the local train, who, considering the available information, was completely responsible. This, perhaps, illustrates the difficulty of employing responsible men at the present time.

The accident was made more serious by the rupture of a tank car containing chlorine, which killed a number of people in the local community, and which attracted most of the interest of the media. News comment showed how little the public knows of railway operation. It was mentioned that "electronic controls" were not in service, which meant a circuit controller at the switch to shunt a track circuit. No mention was made of the primary responsibility of the conductor for the position of switches used by his train, something which used to be carefully observed.

Return to Railway Index

Composed by J. B. Calvert
Created 20 February 2002
Last revised 11 January 2005