Train Order Accidents

A review of 68 accidents under the American system of time table and train orders from ICC reports, 1911-1966

Introduction

This paper critically reviews a number of railway accidents that represent failures of the distinctively American time table and train order method of operation, taken from ICC accident reports with occasional added background. It would be more pleasant to review operations that did not involve accidents, but the stories are not available, so we must do with the pictures we have, and be satisfied with the hard lessons. The table of accidents below makes it easy to jump to any particular account. It is much easier to do this than to browse the unclassified ICC reports directly. This treatment is now deeper and more extensive and technical than that in Shaw.

The availability of historic ICC accident reports for the years 1911-1966 on the internet makes it easy to research operation under time table and train orders to find out how failures occurred. The computer search is much more powerful than poring over piles of documents, however limited its capabilities. Shaw's book discusses accidents in which there were public casualties, ignoring the more frequent events when there were few casualties, and those largely confined to employees. However, these less tragic happenings teach lessons as important as those taught by the more spectacular ones, and they are more pleasant to consider. They are also a view of operations in the classic period of American train orders, perhaps a major resource, and so are of intrinsic interest. The discussion here will carry forward Shaw's treatment, and extend it at several points. I differ with Shaw, largely on emphasis, on a few points. The reader can form his or her own opinions, since the information is freely available on the Internet, at the link given in the References.

Most train operation today, worldwide, is by signal indication. The driver of a train responds to immediate, simple stimuli and must obey them precisely. Operation by time table and train orders requires completely different skills, including memory, analysis and judgment, reference to time and to written instructions, and calculations of time and distance. It is, therefore, infinitely more interesting and stimulating. American railroad men were once proud of their skills, and rightly so. Today, many people who are employed by railroads seem to be either uninterested in developing the necessary mental skills, or are incapable of doing so, while the demands have become less challenging as the traffic is less varied, many lines have vanished, and ownership become more concentrated and bureaucratic and less personal, so that changed operating practices are essential.

Shaw says that the first ICC investigation by the Bureau of Safety under the Accidents Report Act of 6 May 1910 was of a Soo Line collision at Superior, WI on 5 June 1911, but this was actually the fifth investigation. Investigation No. 1 was of a derailment on the Pennsylvania's Belvidere Division at Martin's Creek, NJ in the same year on 29 April due to rough track, in which 12 passengers were killed. The feature of this derailment was gas. The wooden cars were lighted with acetylene or Pintsch gas, and they were completely consumed. The dining car gas tank exploded in the fire several hours after the accident, providing additional excitement. The highest-numbered investigation that I could find was No. 4099, concerning a derailment and collision on the Milwaukee on 20 July 1966.

Operation by telegraphic messages began around 1855, as soon as problems with the Morse patents were overcome and partnerships with telegraph companies established. The normal commercial telegraph lines of the period were far too unreliable for railway purposes, so that telegraph dispatching became practical only when the railway companies could control their own lines with exclusive access to them. Telegraph dispatching was overlaid on the existing well-developed time table system, making it much simpler and safer, and this relation remained until operation by signal indication superseded it. Trains can run purely on time table authority, but this means no extra trains, and no adjustment for delays of superior trains, without confusing and hazardous rules. Train orders provide for the operation of extra trains, and for the elimination of delays due to delayed trains, as well as for much else besides. Train orders are a special kind of telegraphic message, written on distinctive forms and subject to strict regulation of the methods of transmission and delivery. They are quite different from ordinary messages, and contain solely information dealing with the movement and safety of trains, which ordinary messages must not contain.

The dispatcher is the key person in the train order system, and the only one whose work is not checked or monitored. There is only one dispatcher for a given length of track, called a division, subdivision or district, so that there is no division of command that could lead to confusion. He communicated by telegraph or telephone with the operators along the line, each having a train order signal to stop trains for the delivery of orders, or to let them pass. Finally, the conductor and engineman received orders in writing, in the same words to all concerned in the movement. With radio, the dispatcher can communicate directly with the train crews, eliminating the need for operators (as on electric interurban lines). All but a very few of the accidents discussed here were before the time of train radio, so that only communication by wire with operators was available. On steam roads, labor agreements ruled out copying of orders by train crews, using the telephone, except in emergency. The use of radio to communicate between the front and rear of a train was an extremely valuable facility, nevertheless. Now, of course, there is no longer anyone on the rear, so this is no longer important. The changes of the present day are a completely different subject, and will not be considered here.

The telephone was adopted for dispatching beginning around the time of the First World War, prompted mainly by the difficulty of obtaining qualified operators and their high cost. Anyone physically capable can operate a telephone without special training, and the lower level of skill goes with lower wages. Sometimes it is viewed as an advance in technology, but most people involved would say that it is no improvement on telegraphy. Telegraphy concentrated on the individual characters of a message, and was not subject to the uncertainties of aural perception, so it was better adapted to the transmission of train orders. This explains its late survival in many places, in spite of the lack of trained Morse operators.

Failures of the train order system can be classified in three natural categories: (1) improper creation of orders; (2) failure of delivery of orders; and (3) misinterpretation or overlooking of orders, corresponding to the dispatcher, operators and train crews. The mistakes they made are discussed below. The Standard Code of train rules was adopted by the General Time Convention (Later ARA and AAR) in July 1889. It was a compilation of the best practice at the time, and was advisory, not compulsory. How well the work was done is shown by the fact that there have been no substantial changes or additions since that time, except perhaps for the introduction of the Clearance Card as a list of orders delivered, and these rules formed the basis of a uniquely North American method of operation. The Standard Code required conscientious, intelligent and disciplined employees, with the capacity to understand written instructions (such men and women may now be hard to find). In my opinion, it was a safe, efficient and economical means of controlling operations for low or moderate traffic densities, especially when supplemented by an automatic block system.

A very important element in time table and train order operation was flagging. Time table and train order give full protection against opposing trains on single track, but only limited protection against following trains, for which a man with a flag or lamp on the ballast is the main resource. This is a separate consideration, requiring rather elaborate discussion to explain it fully, so it will only be mentioned here when it has some direct bearing on the story. Flagging failures result in rear-end collisions, train-order failures in head-end collisions. We are mainly concerned with the latter. The chief role of an automatic block system is to provide protection against following trains and open switches, and to ease the job of the flagman. It is a complementary and desirable adjunct of train-order operation, permitting high speeds with safety.

The most important common factor in train order accidents revealed by this study is the abdication of the conductor's responsibility for the operation of the train jointly with the engineman, a mutual check that is the foundation of the Standard Code. In nearly every case, this is a contributing if not the principal cause of accident. Conductors are found repeatedly failing to consult with their enginemen, not showing orders to brakemen or other interested crew members, not being aware of the location of their train, not making train identification at meets, failing to observe Rule S-90 on exchanges of signals approaching meeting or waiting points, failing to supervise flagmen, and occupying their minds with waybills and tickets. When trains are operated by signal indication, conductors become as useless as firemen on diesel locomotives, and are no prettier.

The initial search strings for the reports of interest were "overlap* near authorit*" which found 16 documents, and "fail* near deliver*" which found a surprising 63. Note the use of the wild-card marker * that allows for variations in phrasing. The phrase "fail near delivery" only finds 4 documents, not 63. It does not appear possible to search for complete phrases, only individual words combined by the operator "near".

One could easily form the opinion that railway accidents are frequent and most employees are careless or incompetent. Nothing could be farther from the truth. The ICC reports represent a small number of incidents where certain factors have combined to cause disaster. If we discussed all the cases in which disaster did not occur, these unfortunate incidents would be utterly buried. I recently heard that some 800 truck drivers die every year in accidents (in addition they take a lot of the rest of us with them), which is a higher toll than on the railways around the turn of the century when employee casualties were a disgrace. Our data base here is all reportable accidents for the years concerned, not just an arbitrary selection.

It must be taken into account that train rules not only changed with time but were also modified by individual companies. There were also special instructions in operating time tables that varied greatly. Dispatchers usually operated according to company policy, which was spelled out in a manual. Sometimes the peculiarities are known to me, and sometimes not. Fortunately, the principles did not vary significantly in spite of all this variety of detail, so the accidents will be discussed on this general basis. The most important thing about them is that they reflect actual, not theoretical, conditions.

Form 31 and Form 19 Orders

The reader will understand the reports better knowing what a train order is, and what the difference is between Form 31 and Form 19 orders. This section has been prepared with that in mind. To "copy" an order means to write it in manifold using double-sided carbons, as it comes in on the sounder (or telephone). To "meet" means for two opposing trains to pass by one another at a siding on single track. When a train is in the clear on a siding for a meet, its headlight is extinguished; otherwise, the headlight shows that the train is not in the clear. To "pass" means for one train to overtake another in the same direction at a siding on single track. When a train is in the clear on a siding to be passed, its markers are turned to show green (or yellow) to the rear when it is clear of the main track; otherwise, red markers show the train is not clear of the main track. A train has "right" if a train order gives it precedence over another train. Otherwise, the time table may specify the precedence of trains by "class" or "direction." Precedence is the right to move on the main track, perhaps restricted by a schedule or other provisions, with the other train, in the same or opposite direction, keeping out of the way, taking siding when necessary to be met or passed. The inferior train must usually be clear five minutes before a scheduled superior train is due, called the "clearance time," to allow a margin of safety, so the superior train will not have to approach prepared to stop. Trains of the same class, and extra trains, must approach such points prepared to stop, unless a block system is in use and signals are clear.

All train orders are essentially alike in that they are prepared, transmitted and delivered according to strict rules that ensure correctness and timely delivery. All are recorded in writing in the dispatcher's order book and are numbered. They are addressed to those who are to execute them, which includes operators in some cases, expressed in the same words, and repeated after transmission to ensure accuracy. The only real distinction between what are called Form 31 and Form 19 orders is the method of delivery. They are otherwise distinguished by being written on standard forms printed on yellow or green paper, respectively.

The original train order was similar to the later Form 31. The order was transmitted to the operator, who put out a train order signal and repeated the order to the dispatcher. The conductor and engineer would report to the telegraph office after seeing the signal, where there were formal readings of the order, by the operator or the trainmen, to ensure that the contents were understood. Then the conductor and engineer would sign the order, and their signatures would be telegraphed to the dispatcher. The dispatcher would give "complete," and this was noted on the order with the time and the superintendent's initials (symbol of authority). Now the order was in force, until it was fulfilled, superseded or annulled. An order is considered fulfilled when the train involved ceases to exist.

When a cumbersome rule that would hinder traffic is promulgated on a railroad, it is ignored or worked around in some way, and this rule was no exception. Getting the engineman off the engine and into the office with the conductor is exceedingly time-consuming and inconvenient. The engineman has important work to do and no time to stand about chatting. What was actually done was that the conductor would go to the office for the order, and forge the engineman's signature, then deliver the order to the engineman the next time he was near the engine. This alternative became the rule. The conductor would go to the office and sign for the order, together with any reading formalities (that became neglected, fortunately), and then deliver the order to the engineman. The engineman would then sign the conductor's copy as a receipt. At that time, orders were turned in at the end of a run so they could be inspected.

This procedure was still time-consuming, and it required the train to stop, and the conductor to walk the length of the train and back. To cure the problem, a method of delivery was devised that did not involve signatures. The order was made complete (effective) when it was repeated, and then the operator delivered the order to both the engineman and the conductor, as the engine and caboose passed the office. In fact, the train did not even have to stop, and the orders could be handed up attached to hoops or delivery forks. These new orders, delivered personally to engineman and conductor by the operator, were dubbed Form 19, and the traditional order called Form 31. The usual way to show that the train did not have to stop when the order signal was displayed was to show a yellow flag or yellow lamp, with the operator standing on the platform as the train approached. If the process miscarried, the train had to stop for the orders. Form 19 orders were not used to restrict a train, only to help it, and not to establish meets. A train was considered restricted when a 31 order addressed to it had been repeated, not when it was eventually made complete.

Further progress, if it can be so called, was made when the engineman no longer had to receipt for the Form 31 order. Additionally, orders were often handed up to firemen and brakemen, who were supposed to pass the orders on, which they usually did, if it was convenient (see Dewey, IN). When a train was stopped for orders, it remained the best practice for the conductor to hand orders to the engineman personally, and to discuss them, but this was not required, and often omitted, as many of the accidents below testify. Since conductors and enginemen were supposed to compare watches before starting, discussing the orders would not be inconvenient. At this point, there was really no difference between the security of delivery of 31 and 19 orders, as the 31 order no longer assured that the engineman would see and understand it. Generally, 31 and 19 orders were generally handed to the engineman in a single pile, distinguishable only by color.

It was soon realized that any difference in security between 31 and 19 orders was minimal, and the taking of signatures (except in special cases) was dropped, with the elimination of the Form 31, or by the adoption of a common train order form. There were indeed failures of delivery of train orders, but they would not have been cured by Form 31 in its later, toothless form. Many officials thought the taking of signatures had some formalistic significance that would attract the attention of trainmen more than the informal Form 19 order. Arguments along this line simply never got to the root of failure of delivery, which lay mainly with operators forgetting orders and enginemen not getting them at all. The signatures did not have to reach the dispatcher before he could help another train (except in certain situations, or when the order was to be delivered by a third party)--this would have been safe, but much too cumbersome.

The 19 order was first used to restrict trains in the "middle order" where 19 orders would be delivered on either side of the meeting point to the two trains involved, and also addressed to the operator at the meeting point, who would use the train order signal to ensure that neither train passed the meeting point. It was used for running orders for extra trains, and for train order register checks. Some roads retained 31 orders for meets, wait and run-late orders, to the trains restricted. Other trains affected received 19 orders. If a 19 order were not delivered, it could at most cause delay.

The chances of nondelivery were greatly reduced with the introduction of the clearance card as an order index. This form was originally used to authorize a train to pass a train order signal at stop when there were no orders for the train. It positively identified the train for which it was issued. The practice grew of listing the order numbers for a train on it as they were copied by the operator. Before clearing a train, the operator repeated the list of orders to the dispatcher, who gave his OK. When conscientiously done, this procedure practically eliminates nondelivery of orders. Now it was safe to restrict a train by a 19 order, and this has proved an acceptable practice. Any necessary signatures are simply forged in a blank area. It is actually the clearance card that eliminated the need for the Form 31 order.

In rare cases, simply placing an order at an office for delivery to a train is not enough to restrict it, and signatures should be obtained before the order is made complete to the train acquiring right. For example, when an extra is given a schedule and rights over all trains, it must be known that the order has actually been delivered to all trains on the road that are thereby made inferior, before the order is made complete for the extra. The reason is that the order may be sitting on an operator's desk and the extra passes before the train restricted has arrived and knows about the scheduled extra.

Trains are also sent messages, which are not subject to the rules for train orders, and must not contain information respecting the movement and safety of trains. Messages are never written on train order forms, but usually on small pieces of white or yellow paper. They may give information on cars to pick up or set out, number of cars in opposing trains, special treatment for passengers, lunch arrangements, and so on.

Index to the Accidents

Operational accidents, which are the only kind that will be discussed here, are not as common as accidents due to bad flagging, passing signals at stop, failures of equipment, track or structures, or highway crossings, and usually are not as costly in lives, except when aggravated by other factors, such as fire. Their causes involve human actions and psychological factors rather than material reasons, which makes them interesting to review and analyze. The 68 incidents discussed here are listed in the table below. These are doubtless not all of the accidents due to failures of time table and train order, but are a good sample of them, revealed by computer search of all available accident reports. Clicking on a link will take you to the part of the page that discusses that accident.

Causes of Accidents

It is possible to classify the 66 accidents under time table and train order operation by principal cause. In most cases, this is the cause as determined by the ICC investigators. We are dealing not with a sample here, but with all available accident reports, so the results may have some validity in showing what actually caused accidents in practice. Of the 66 accidents, 37 resulted from improper actions of train crew members, chiefly conductors and enginemen, who are jointly responsible for the safety of their trains. 16 resulted from improper actions of operators, who are the intermediaries between dispatcher and train crew. 13 were caused by erroneous actions by the dispatcher in creating orders and seeing that they were sent to the proper addresses at the proper places.

The dispatcher's errors were quite various, each occurring only once. The errors were: failure to send a restricting order to the train restricted when another train had been helped against it; when going on duty, not checking wording of previous orders; failing to send an annullment to a concerned train; completing an order to a train already past the station; forgetting a returning helper extra; issuing a lap order; OK'ing a clearance omitting a restricting order; overlooking an extra on the road; wording orders improperly; handling sections improperly; and issuing an erroneous check of overdue trains. Among all the possible errors, these were the ones that actually caused accidents. Dispatchers were often called upon to handle 30 or 40 trains a day on a hundred miles of single track, or on 10 or 15 miscellaneous branches aggregating hundreds of miles. In general, they performed a marvelous job. One accident reveals single-order dispatching with the knowledge of officials.

Single-order dispatching is the practice of sending orders to individual trains, telling them only what they are to do. It is the method that would be devised by experts at the present time, and may already be in use, judging from accident reports. It is also very dangerous and faulty, second in fallacy only to having several dispatchers independently giving orders to trains on the same track. The Standard Code explicitly prohibits it, requiring that all orders must be issued in the same words to all involved.

The most frequent operator's error was the failure to deliver a restricting order, either because the train was forgotten or the train order signal was improperly at Clear, which was responsible for eight accidents. In two further cases, the order was not legible, and misreading led to accident. Single accidents resulted from: forgetting that a second train was to receive orders; displaying the train order signal for the wrong direction; clearing a train order signal used as a block signal when the block was reported clear, while orders were held; failing to display the train order signal at Stop before an order was copied; failing to deliver order when engine was past train order signal; copying a train number incorrectly and not correcting it upon repetition. In several cases, an order was held restricting a train at the station of delivery, but the operator took no measures to ensure that the train crew was aware of the fact, and later cleared the train order signal. Whereas in dispatcher's errors, the dispatcher is often overworked and harried, with operators it seems more often that the operators are careless and negligent to a surprising degree. This may be a result of the introduction of the telephone for dispatching, which eliminated the requirement for skilled Morse operators and their training. Operators also had an amazing variety of duties, and were often distracted from train orders by mail, baggage, express, tickets, car reports and so forth.

The most common error of train crews is to misread an order that is perfectly legible and properly composed, which caused 13 accidents. A surprising number of these involved mistaking section numbers, though they were clearly written. In five cases, a meet or wait order was ignored by the engineman, and the conductor took no effective action to stop the train. In two cases, there was faulty delivery of orders to a relieving engine crew or engineman, and again the conductor took no effective action. In two cases, the engineman seems not to have read an order in time, but again the conductor took no effective action. In two cases, the conductor, being near the office, received orders but did not deliver them promptly to the engineman. In three cases, one of which was quite serious, an incorrect train identification was made, in which the conductor took no interest. In one case, the train register was misread and a section missed, and in two more the conductor did not check the register at all. One accident was caused by the failure of the conductor to observe Rule S-90, and another by leaving his train on the main track without protection. The time of a superior train was not properly cleared in one accident, and in another, a train did not take siding but held the main track. In one curious case, verbal orders were confused in a rare case when interurban methods were used by a steam railway. In another, a local practice at certain meets where a train had to double in was not properly observed by an engineman.

The conductor appears to be largely responsible for errors of train crews, through the common habit of leaving the running of the train up to the engineman, and taking more interest in tickets and waybills than train orders. In every case when a meet was overlooked, or a waiting time not observed, the conductor might as well have been on another planet. Junior crew members, the firemen and brakemen, are often browbeaten and excluded from consideration, when participation in the running of the train forms a valuable part of instruction of inexperienced employees. Even when they suspect that disaster looms, they are hesitant to "pull the air" in case they might face the wrath of the conductor if they are wrong. We do not know how many times cooperation has averted disaster, but from the reports it does not seem that it is the rule. In most cases, the reliance on enginemen seems well-placed, since they must realize how much depends on them alone, and they often suffer for the faults of others.

The move that contributed more to safety than any other was the restriction of the hours of service to 8 hours for dispatchers and operators, and 16 hours for train crews. Early accident reports reveal atrocious and heroic conditions; one dispatcher was on duty for 12 hours after substituting for another's 12-hour trick. Operators normally worked 12-hour shifts, and changing shifts sometimes meant 24 hours on duty. Their work, which was hard, skilled and demanding, was considered light and comfortable by railroad managers, because it was performed indoors with a nice stove nearby. In train service, everybody except the engineman could get nice, refreshing sleep if they were bored by the scenery.

Two accidents are discussed that happened under Centralized Traffic Control, one largely the fault of the control operator and the other of the engineman, to show how accidents are not confined to time table and train order operation, and that CTC is no panacea, simply a different can of worms. Most accidents today are caused by passing signals at Stop, not by the literary exercises of train orders.

One characteristic of accident investigations should be pointed out to the unwary. Usually, those most culpable, individually or as a committee, create a "fairy story" to explain events in the most favorable way possible, throwing the blame, if possible, on inanimate objects, and if not, on persons not present. As a last resort, the most junior participants are held at fault. In all these accidents, there are no failures of the air brakes or signals. All were due entirely to human causes, and most occurred in good weather. Many were due to understandable lapses, fatigue and errors of perception, very few to culpable negligence. There is no interest here in assigning blame for these long-past events, and we revere those who suffered.

Shaw, who deprecates time table and train order operation, forsaw a golden era when CTC would make all the uncertainties disappear and banish collisions to limbo. In December 1959, CTC was brought into service between St. Joseph and Napier, MO, superseding the double track that had existed between those points about 37 miles apart. This was part of the Ottumwa Division of the Chicago, Burlington and Quincy. The control operator was located in St. Joseph, and worked under the supervision of the dispatcher.

Just after midnight on the 22nd of August, 1960, regular freight train No. 70 was rapidly moving southward with only 31 cars and a way car, passing Napier at 1.30 am. Passenger train No. 23 had just left St. Joseph with 14 cars, including 7 sleepers, a flat car and a way car. The 2.1-mile siding at Nodaway, 16.6 miles north of St. Joseph, looked appropriate as a meeting place for the two trains. By close cooperation between the dispatcher and the control operator, they managed to get both trains on the same siding, headed up, where they bumped at 1.59 am, in clear weather. It was probably best that this embarrassment happened at night, when nobody could see. There were 12 injuries, 2 of them passengers.

The dispatcher had called the control operator at 1.30 am and directed that No. 70 should be put on the siding. He changed his mind, and phoned again at 1.40 directing that No. 23 be put on the siding instead, the control operator telling him that he had not yet lined the route for No. 70, as he wanted to do that when No. 70 was close to the end of the siding. Both men thought that No. 70 was proceeding on the main. The control operator heard the flagman of No. 70 report clear of the main track, which should have opened his eyes.

The engineman of No. 70 dimmed his headlight when he was told that the train had cleared, and proceeded toward the south end of the siding at what he said was 20 mph. There was no proof after the accident, since the speed recorder had no tape in it. No. 23 acted about the same way, dimming its headlight and crawling along, hoping that it would get a clear signal at the north end of the siding after No. 70 had passed. Each train probably saw the other, but did not appreciate that they had been put on the same track until it became obvious.

There was no track circuit on the siding, which was usual. The crewmen seemed to believe that if there were already a train on the siding, they would get a red aspect, not a red-over-yellow, but the belief was erroneous. If the control operator intended to put two trains on a siding, he had to inform them by telephone at the ends of the siding. Of course, nobody intended to have two trains on the siding here, so there were no such conversations.

When the CTC board was inspected after the accident, the control operator said it was just the way it was at the time of the accident. The switch lever was set for the main at the north end of the siding. The control operator swore that he had never lined the switch for the siding, but of course he had, and was trying to see if a brazen lie would work. One also wonders where the speed recorder tape went. The report does not say anything about a seal that I think would have to be broken.

The siding had to be used at Reduced speed, which on the Burlington seems to have been the same as Restricted speed, except that you didn't have to worry about switches not properly lined. The turnouts could be negotiated at 30 mph, but this would be a little fast for Reduced speed. The engineman of No. 70 was probably going more than 20 mph, but this was no doubt normal practice. This provision was insufficient for two opposing trains not expecting to be on the same track, however.

In the ICC website, this report is filed under a heading for Southern Pacific, Clotho, CA. That accident, whatever it was, seems lost. There is a pretty good description of how to operate a CTC panel in the report.

On 25 November 1951, two passenger trains collided in CTC territory at Woodstock, AL, killing 15 passengers and 2 employees. This accident is discussed here to show how mishaps occur under operation by signal indication, and contrast them with train order accidents. On this line, CTC had just been brought into service (10 October) between Burstall and Tuscaloosa, 44 miles, the first phase of installing CTC over the whole division from Meridian to Birmingham, 152 miles. The Woodstock siding was 1.62 miles long, with controlled signals at each end, of searchlight type. Intermediate automatic signals were one-arm semaphores. Intermittent inductive train stop was also provided, and the passenger speed limit was 80 mph.

The trains involved were northbound No. 48 with 8 cars, that had taken siding at Woodstock to meet First 47 and was moving slowly. First 47 was carrying green signals for a following section, and whistled them to No. 48, which acknowledged them. This was not required in CTC territory, but was probably a good idea since CTC did not extend over the whole division. Second 47 was a heavy L&N train of 13 cars, detoured because of a defective bridge on its own line, closely following. The engineman of No. 48 had forestalled at the restrictive aspects approaching Woodstock to prevent a penalty brake application.

Approaching the north end of the siding, the engineman of No. 48 misread the dwarf searchlight signal governing admission to the main track, and trailed through the switch, which was set for the main line, stopping 100' north of the siding switch after an ATS penalty application. The fireman went to the phone booth at the end of the siding to report events. This happened just as Second 47 passed the last intermediate signal at clear, and approached at about 65 mph. The fireman of No. 48 made a futile attempt at flagging with an orange cloth, which was all he had on hand, while the engineman jumped. The speed of Second 47 was reduced to about 45 mph at the moment of collision. The engineman was killed in the wreckage, so his testimony was not available.

There was no tape in the speed recorder of the engine of No. 48, but officials at the CTC console estimated his average speed on the siding at 26 mph. The recording paper had not been installed yet on the CTC machine. A test a day later with a similar train moving at 15 mph on the siding showed that the ATS penalty brake application could stop the train in 242' while the distance from the dwarf signal to the fouling point was 270'. This shows why the ATS did not prevent the accident.

The dwarf signal's beam was inclined upwards so it could be seen clearly at a short distance, within 800'. The high signals in the same location were easily read at 2000'. When the engineman of No. 48 first saw the signals, the high signals were clearly red, while the dwarf signal probably showed mainly an uncolored reflection from its cover glass. The fireman testified that the signal seemed Clear, and he and the engineman called the aspect to each other. It seems that nobody paid very close attention to the signal as it was approached.

This accident shows that operation by signal indications puts the responsibility on engineman in particular, instead of on the conductor and engineman mutually. Calling of signals to each other if there is more than one man in the cab, if conscientiously performed, does prevent mistakes and negligence, but is by no means infallible. In this accident, the wrong aspect was called. It seems that one set of hazards is only replaced by another set. Automatic train stop complements operation by signal indications as naturally as automatic block signals complement time table and train order operation, giving nearly perfect safety, but it still can be thwarted by determined effort.

Paynesville, MN is on the Soo Line about 85 miles northwest of Minneapolis. Early in the morning of the 13th of July 1950 a dispatching problem arose at this point on the prairie. Eastbound passenger train No. 108 was approaching from the direction of Glenwood, while westbound passenger train Second 3 was coming from Minneapolis. Eastbound freight No. 90, with 105 cars, had just arrived at Paynesville, which had a 103-car siding. The dispatcher, evidently an optimist, had expected No. 90 to make more rapid progress, and had issued Order No. 142 before midnight: "Second 3 Eng 4001 take siding and meet No 90 Eng 205 at Eden Valley." Eden Valley was 9.3 miles east of Paynesville, and Second 3 was taking siding there because the siding was not long enough for No. 90. With this taken care of, he issued another order directing No. 108 and Second 3 to meet at Regal, a siding west of Paynesville.

This should have worked fine, except that No. 108 was late, and No. 90 was about on its time as Paynesville was approached. No. 90 received a clearance at the station, which was at the west end of the siding, the train order signal was cleared, and No. 90 went on the siding to clear No. 108 before it was due. Some 105-car trains had fit on the 5162' siding, but this one did not, by about two car-lengths. When the rear end cleared, the front was only 35' from the switch, well past the clearance point.

Now we have a fully-developed problem. No. 108 cannot pass Regal. No. 90 cannot proceed against No. 108's time (we are not allowed to assume that No 108 cannot get by Regal somehow!). Second 3 cannot leave Eden Valley until No. 90 gets there. Checkmate. The distances are too great for a flagman to walk. The dispatcher, seeing that No. 90 was glued to Paynesville, issued Order No. 24: "Order No 142 of July 12th is annulled." The conductor of No. 90 received the order, which was made complete at 2.12 am. Then the order could be made complete for Second 3, which allowed it to pass Eden Valley under time table rights and froze No. 90 at Paynesville, since not all sections of No. 3 had passed. Second 3 would now get to Regal, No. 108 would proceed, and No. 90 would follow it from Paynesville. Problem solved.

There was a small catch, however. No. 90 did not fit on the siding, which they knew up at the head end a mile from the station, but did not at the rear end. They decided that the only thing to do was to protect against No. 108 by a flag, and when it arrived, to back a little and let it "saw" by. Since nobody wanted to walk all the way to the head end to tell them, they did not know that Order No. 124 no longer protected them from Second 3, and that they were encroaching on its rights by about 100 ft. While worrying about No. 108, they acquired a new worry when Second 3's headlight suddenly appeared in the east. A man started out waving a fusee, which did some good, since Second 3 had reduced to about 30 mph when it struck the units of No. 90 a glancing blow. There were no serious injuries.

The dispatcher was at fault for not getting the signatures of both conductor and engineman before making the order complete, especially as the engine had gone by a clear train order signal by about a mile. Optimism that the train was clear was not enough in this case. He did not get both signatures because he was in a hurry to move Second 3, and afterwards the conductor was unwilling to make the long trek in the darkness. One should never depend on a conductor in situations such as these, even when it is not snowing.

Waterville, MN was on the Minneapolis & St. Louis south of Minneapolis, on the way to Albert Lea, at the point where the CGW's Faribault-Mankato line was crossed. The station was at the west end, the west switch of the siding was 1127' east of it, then came the 1308' siding, and finally the CGW crossing, protected by an automatic interlocking with home signals, about 2134' east of the east switch of the siding.

At 3.35 pm on the 17th of July 1953, local freight Extra 546 East (compass direction south) arrived at the station, and began to do some switching. At Montgomery, 15.6 miles west, it had received train order No. 96: "No 1 Eng GE25 wait at Palmer until 415 four fifteen pm for Extra 546 East." Palmer was 4.4 miles east of Waterville. It was pretty obvious that Extra 546 would not finish its work in time to make Palmer for No. 1, so they decided to remain at Waterville for No. 1. The dispatcher did not let things take their natural course, but wanted to expedite matters. Although Extra 546 had passed the train order signal at clear, he directed the operator to display Stop, and issued order No. 101: "Order No 96 is annulled No 1 Eng GE25 meet Extra 456 East at Waterville Extra 456 East hold main track at Waterville." As at Paynesville, the conductor assumed that Extra 456 was pinned. The order was completed to No. 1 at Waseca, east of Palmer, at 3.59 pm. When No. 1 received order no 101, it proceeded immedately. The conductor of Extra 546 received the order at 4.10 pm.

The train was then moved to the siding, east of the station. The engineman detached the engine at 4.09 pm and moved it forward to deposit a flagman near the CGW crossing to flag for No. 1, saving a half-mile walk. He had no idea of the annullment, which gave the track he was occupying to No. 1. No. 1 was very close by this time, in fact past the signal for the CGW crossing, which Engine 546 would have sent to Stop when it entered the track circuit for the crossing. No. 1 collided with Engine 546 at 4.12 pm, moving at about 5 mph. There were no serious injuries.

It was definitely improper to help No. 1 before the order was complete to Extra 546, as well as not to have the signatures of the conductor and engineman of Extra 546, which had passed the train order signal. As at Paynesville, mischief was caused by the head end not knowing what the rear end knew. If the engine had not been detached, or the train had fit on the siding at Paynesville, no one would have been the wiser in either case.

Enright, TX was on the Missouri Pacific line from Waco to Houston, which crossed the Palestine to Austin line at Valley Junction, about 35 miles south of Valley Junction, as shown on the sketch at the right. This line was operated by time table and train orders, and the maximum speed was 65 mph passenger, 45 mph freight. The morning of 21 March 1947 was clear. Northbound second-class freight No. 70, with 47 cars and caboose, and southbound third-class local freight No. 395, with 6 cars and caboose, were to pass through this area between 9 and 10 am.

At 5.49 am train order No. 227 was made complete for northbound trains at Spring, 65 miles south, and for southbound trains at Valley Junction. This order read: "Motor car RO1 works extra 801 am until 601 pm between Cawthon and Fountain protect against second and third class trains. No. 70 Eng 1106 wait at Cawthon until 945 am Koppe 955 am." Both No. 70 and No. 395 received this order. No train was specified for No. 70 to wait for, so it could not pass the named sidings until the times given.

When No. 70 made better progress than expected, at 6.32 am the dispatcher issued order No. 231 to No. 70 at Navasota, and to No. 395 at Valley Junction, which said: "No 70 Eng 1106 wait at Enright until 920 am Koppe 930 am for No 395 Eng 334. There is a problem here, that shows the dispatcher did not have a clear grasp of what he had done. This order does not supersede order No 227, and No. 70 must still wait the later times, whether No. 395 has appeared or not. The dispatcher has forgotten the earlier order No. 227, it would appear, and assumes that order No. 231 will permit No. 395 to proceed.

At 7.46 am, the dispatcher seems to have realized his blunder, and issued order No 238 to No. 70 at Navasota, and to the work extra R01 at Cawthon, reading: "That part of order No 227 reading No 70 Eng 1106 wait at Cawthon until 945 am Koppe 955 am is annulled. No 70 Eng 1106 wait at Enright until 920 am Koppe 930 am for Work Extra R01." Now he thought he had fixed up Extra R01 as earlier he had fixed up No 395. The order was not delivered to Extra 395, since he had overlooked that Extra 395 was running on order 227. Now, the conflict between orders 227 and 231 should have raised a question, but they did not, and it was legal for No. 395 to take advantage of the situation.

Had No. 395 received the annullment, as it should, what the dispatcher intended would have taken place. Since it did not, No. 395 was running for the 9.45 time at Cawthon, while No. 70 waited at Enright only until 9.20. It was no surprise that they met at 9.26 am a mile and a half north of Enright. Speeds were reduced to 20 mph at impact, but six employees were injured.

The report incorrectly states "there is [sic] no sidings at Enright Koppe Cawthon." It normally makes no sense to wait at a point that is not a junction or siding for an opposing train, for obvious reasons.

A more fatal accident occured on the Midland Valley at Bokoshe, Oklahoma on 1 Feb 1958. The Midland Valley ran from Wichita to the vicinity of Fort Smith, and was operated in three subdivisions, Western, Middle and Eastern. The Eastern Subdivision ran from Muskogee to Fort Smith, with entry to Fort Smith over 15.4 miles of joint track with the Frisco. This was a single track, operated by time table and train order, with day offices at Porum, Stigler, Panama and Rock Island. The office at Muskogee was open continuously, since the dispatcher was located here at the headquarters of the company. Trains originated at Shopton, 1.1 miles east of Muskogee, but called for orders at Muskogee. The speed limit was 30 mph.

Eastward trains were superior to westward trains of the same class on the Eastern Subdivision. The only scheduled trains were No. 42, eastward, and No. 41, westward, both local freight trains, scheduled to leave their initial stations in the morning and complete their 100-mile runs a little after noon. 1 February 1958 was a Saturday, so normally there would be no open offices, except Panama for a couple of hours around noon. However, the dispatcher had notified the operator at Stigler to go on duty at 10 am to handle some paperwork there.

No. 41 had departed Rock Island at 9.35 am, 2 hours and 20 minutes late, and arrived Panama at 9.55 am. The time table meet with No. 42 was at Stigler, but No. 41 could not leave Panama because it could not clear at Keota in time. No. 42 left Shopton at 8.30 am, 30 minutes late, with work to do at Warner, Briartown and Keota. Stigler was the next station west of Keota, and this morning the dispatcher had an operator there, so he decided to restrict No. 42 at Keota so that No. 41 could leave Panama. Order No. 27 said: "No 41 Eng 152 meet No 42 Eng 153 at Keota No 41 hold main track at Keota." The order was made complete at 10.34 am at Panama and No. 41 left at 10.35.

Of course, before this occurred, the dispatcher asked the operator at Stigler if No. 42 had arrived. The operator said he had not seen it. The operator had gone on duty at 10 am, so all he could say is that it had not passed in the last half hour. In fact, No. 42 had passed Briartown (the station next west of Stigler) at 9.40 am, Stigler at 9.55 am, and arrived in Keota at 10.20 am. After doing its work, it left at 10.30 am. No. 42 was almost to Panama before No. 41 left there! The lack of open offices meant that the dispatcher had no idea of what progress the trains were making, and he made an incorrect assumption.

The Fort Smith and Van Buren crossed the Midland Valley at Bokoshe, which had no siding, only a station sign. The crossing was gated against the FS&VB, so MV trains could cross at restricted speed without stopping. No 41 and No 42 collided on a curve 1213' west of the station sign, No. 41 moving at 30 mph and No. 42 at 25 mph. No. 41 had 4 cars and a caboose, No. 42 12 cars and a caboose. The collision split the diesel tanks of both locomotives, and ruptured a tank car loaded with gasoline. The fire destroyed both locomotives and several cars of their trains. The engineman and front brakeman of No. 42, and the fireman of No. 41 were killed.

Rule 36 of the company's rules for dispatchers stated that a train must not be restricted on an operator's statement when the operator has been absent from the office. It is a rather fundamental principle that a train cannot be restricted at a station that it has passed, and this accident illustrates the danger very well.

This appears to be the only serious accident in the history of the Midland Valley system, which consisted of the MV, the Kansas, Oklahoma and Gulf, and the Oklahoma City-Ada-Atoka. The MV also had an encounter with the CRI&P at Midland Tower, KS. The OC-A-A had a collision between a passenger train and a freight at Konawa in 1927, but no passengers were killed, and a work train difficulty at Ada in 1929. The Missouri, Oklahoma and Gulf, a predecessor of the KO&G, had a freight derailment on unballasted track in 1916, and the KO&G a head-end collision of freights at Kenefick, OK in 1931, discussed next. That's all there is in the ICC investigations.

Kenefick, OK was 11.5 miles north of Durant, on the Southern Division of the KO&G, which extended from Muskogee (South Junction) to Dension, and was an important but independent link in the Missouri Pacific's route to north Texas. Like most U.S. railroads, and especially small ones, the KO&G was capital-starved and had to make do. This was especially noted in yard and siding capacity. Though trains grew in length in the 20th century, siding lengths did not. At Kenefick, for example, the siding was 1986' long, holding at most 58 cars (the 1950 timetable lists the capacity at 34 cars). There was also a house track (for cars to be unloaded at the freight house) 1862' long, connected only at the north end, the switch located 954' north of the north siding switch.

A northbound train too long to fit into the siding could "double over" by pulling onto the siding until the caboose cleared at the south end, then cutting the train short of the fouling point at the north end. The engine and front part of the train would then pull forward, and back in on the house track until in the clear. This was not quite as convenient for a southbound train, but could still be done with more backing. When a meet was to take place, the northbound inferior train, the one taking siding, had no right to the track north of the north siding switch. In order to pass this switch to back in on the house track, it was necessary under the rules to send a flagman forward at least half a mile to warn the opposing train, even if it was slowing for the meet.

Flagging used up at least an additional quarter-hour in this case, waiting for the flagman to walk forward, although everyone involved understood what had to be done and could do it safely if they knew a train was doubling over. The Chief Dispatcher handled this by putting the notice in the meet order. For example: "Eng 503 run extra Denison to South Jct take siding meet Extra 501 South at Kenefick. Extra 501 South look out carefully for Extra 503 North doubling over at meeting point." The non-standard clause meant that Extra 501 should approach Kenefick expecting Extra 503 to be north of the siding switch by up to a train length. Extra 503 would have its headlight displayed brightly until it was clear of the main track. This form of order served the purpose quite well at other points as well as at Kenefick.

On the evening of 20 May 1931, Extra 501 South had 90 cars and caboose, and received order No. 53, Form 19, at Tupelo, 34.3 miles north of Kenefick, worded as in the preceding paragraph. Extra 501 left there at 9.01 pm, left Bromide Jct with 72 cars, and proceeded to Kenefick. Approaching Kenefick, Extra 503's headlight was seen, so Extra 501 slowed to 10 or 15 mph with a service application. Just then, the headlight disappeared, so it seemed that Extra 503 was safely in the hole. The engineman of Extra 501 released the brakes and opened the throttle. At that instant the headlight reappeared, bright as ever--the engine of Extra 503 had only gone behind the cotton gin, and was actually just pulling out of the siding. An emergency application of the brakes was ineffective so soon after they were released (a problem with type K triples), so the speed was not reduced much at the point of collison. The engineman and fireman of Extra 501 were injured, but these were the only casualties, fortunately.

This is one of the better ICC investigations, where all relevant factors were properly considered, and a just conclusion reached. Under the rules, the engineman of Extra 503 was at fault for being beyond the siding switch without flag protection. Under KO&G operating practices, however, the engineman of Extra 501 was at fault for not taking sufficient care to be sure Extra 503 was in the clear, though his mistake was understandable. The examiners found a certain laxness in making brake tests, failure to compare watches daily with standard clocks, and too-infrequent examinations on the rules, which they recommended that the company correct, and, for once, made no silly recommendations such as calling for automatic train stop.

There is a safe way to handle cases when a train has to double over or back into a siding. For example, "Extra 501 South hold main track meet Extra 503 North at Kenefick Extra 503 double over protecting according to rule 99." If Extra 501 was not seen coming, engine 501 might creep forward a half-mile or so to leave a flagman, then return to get its train on the siding. This could be made legal with an order like: "Extra 501 South wait at Cain until nine fifteen 915 pm for Extra 503 North," in which case the flagman could be taken out at any time up to 9.15 pm. Incidentally, the siding at Cain, 8.8 miles north, held 75 cars, so Extra 503 would not have had to double over there.

At 3.36 am on 14 June 1951, eastbound mail train No. 210 collided with westbound freight train No. 17 at Cray, MN on the Chicago St.Paul Minneapolis and Omaha (CNW). Then engineman of No. 17 was killed, and seven other employees were injured. The accident was created by events at Lake Crystal, a station 26 miles west of Cray. Eastbound No. 210 left St. James, 22 miles west of Lake Crystal, at 2.35 am, on time.

This was the first day on the job for the operator at Lake Crystal, but he had had previous experience as operator for another company. At about 3.05 am, he heard his address and "31 east" on the dispatcher's wire. He replied "SD east" after operating the train order signal lever. He then received and repeated order No. 14, addressed to No. 210 at Lake Crystal, and to No. 17 at Mankato, reading: "No 210 meet No 17 at Cray No 210 take siding." No. 17 had 90 cars and caboose, so it would not fit on the Cray siding. At 3.19 am, No. 17 left Mankato in possession of the order. At 3.24, No. 210 arrived at Lake Crystal. The operator went out to help with handling mail, and at 3.27 No. 210 began to move. He thought that the train was only pulling up until the single coach was opposite the station for the convenience of the conductor. Instead, the train gathered speed as he ineffectively waved stop signals at it with a flashlight.

He then noticed that the eastbound train order signal was at Clear, while the westbound signal displayed Stop. He had displayed the wrong train order signal, though the lever he operated showed a label "West" before and a letter W on a red background after. In his previous job, the levers had been in the opposite order, so he pulled the wrong one, acting out of habit.

The Louisiana and Arkansas line between Greenville, TX and Shreveport, LA was single track, operated by time table and train order. Sulphur Springs is near the western (timetable direction: northern) end of the line, with the sidings of Brashear and Como on either side of it. Hughes Springs was the first open office east (south) of Sulphur Springs. On the 11th of October, at 10.25 pm, northbound freight No. 54 received orders there, among which were order No. 96: "No 54 Eng 73 meet Extra 76 South at Como" and order No. 97: "No 54 Eng 73 meet Extra 76 South at Brashear instead of Como." At 10.51, order No. 106 was sent simultaneously to Hughes Springs and to Hunt, the initial station on the subdivision at Greenville. Order No. 106 said: "Order No. 95 is annulled." Order No. 95 dealt with movements affecting Extra 76 and a work extra, and had nothing to do with No. 54.

At 11.15 pm, Extra 76 called for orders at Hunt, and received order No. 96, among others, but not Order No. 97 superseding it. The operator at Hunt had unaccountably written down order No. 97 instead of order No. 95, and the variation was not noticed in the repetition by the dispatcher, and neither of the operators heard the others repeat. When the clearance card was OK'd, the dispatcher did not notice the inclusion of order No. 95 or the absence of order No. 97. The operator at Hunt had filed Order No. 97, believing it annulled.

No. 54 believed it could go to Brashear for Extra 76, while Extra 76 believed it could go to Como for No. 54. The lap was caused when Extra 76 did not receive the superseding order No. 97. They collided 2.36 miles south of Sulphur Springs at 12.40 am, injuring six employees.

There was a collision near Norton, Kansas on the Rock Island on 3 September 1944 that injured a number of soldiers on a troop train and several employees. The ICC website has the place down as Morton, but I suppose that is close enough for government work.

The stations are shown in the sketch on the right. Extra 2310 received order No. 291 at Phillipsburg: "Eng 2310 run extra Phillipsburg to Goodland meet First 96 Eng 5061 at Almena." First 96 was a second-class train, a troop train running as a section of a freight train, with 11 cars including troop sleepers. At Selden, the conductor reported to the dispatcher that a journal was running hot, and would require attention at Norton. The expected delay made the dispatcher want to help Extra 2310, but he did not want to delay First 96 further waiting for orders, so he let it go on to Norton while he issued order No. 295: "First 96 Eng 5061 meet Extra 2310 West at Norton East Siding instead of Almena Order to First 96 at Norton."

Order No 295 was delivered to Extra 2310 at Prairie View, the last open station before Norton, after the dispatcher impressed on the operator at Norton of the importance of holding No. 96, which would certainly stop there. Extra 2310 left Prairie View at 6.10 pm. On this line, manual block protected following movements, and the train order signal was also the block signal. The operator at Norton had displayed Stop for First 96, and the signal also protected a passenger train that had just gone east. When this train cleared the block at Prairie View, the operator cleared the train order signal as he usually would do, forgetting the order for First 96. First 96 then came through, the hot bearing not having given any more trouble, and right by the signal, to the horror of the operator, who must have realized what had happened in an instant.

The collision occurred on a curve about 6 miles east of Norton, where the view was obscured, so the speeds of the trains could not be reduced substantially before impact.

An automatic block system is supposed to catch lapses in time table and train order operation and prevent accidents that would otherwise occur. An example of how this works is furnished by the accident near Port Henry, New York on 24 October 1954. This station was on the Delaware and Hudson line between Rouses Point and Whitehall, beside Lake Champlain. It was operated by time table and train order, with ABS. The speed limits for passenger trains was 65 mph, freight 45 mph. On this morning, passenger trains No. 18 and No. 34 were moving southward, No. 18 with 9 cars preceding. The train order office at Westport is 63 miles south of Rouses Point and 11 miles north of Port Henry. The siding at Port Henry is called Sherman, and its north switch is 2968' south of Port Henry.

At 10.58 am, the dispatcher transmitted order No. 9 simultaneously to No. 18 and No. 34 at Westport, and to Extra 4123 North at Whitehall, which read: "No 18 take siding meet Extra 4123 North at Gunnisons No 34 meet Extra 4123 North at Sherman." Extra 4123 had 113 cars and caboose. No. 18 received this order, and finally left Westport at 11.20. At 11.35, Extra 4123 left Whithall. Somehow the operator forgot that the order was addressed to No. 34 at all, and its omission on the clearance card was overlooked by both the dispatcher and the operator. The dispatcher had not told the operator how many copies to make, which may have contributed to the confusion. So, No. 34 left Westport without the order and proceeded on its schedule. There was now a lap of authority south of the south switch of Sherman.

As the two trains neared each other, the signals responded properly and both received Approach and Stop-and-Proceed aspects. No. 34 did indeed stop, and was standing when struck by Extra 4123 North at 12.45 pm. The extra had passed a Stop signal at 37 mph, and was even accelerating at the time of collision, as the speed recorder tape showed. No explanation was given for this behavior. The front brakeman of Extra 4123 was killed in the collison.

Inexperience is often the reason for incorrect actions, but it is often aided and abetted by more experienced superiors. The operator at Goreville, IL on the night trick on New Year's Day, 1944 was a young man who had been examined on the rules 11 months earlier, at age 16, and found not sufficiently proficient to handle train orders. In spite of this, wartime shortages of experienced operators led to his assignment at Goreville on 26 December. Telegraph was used for dispatching on this division, and the new operator was skilled enough to handle this, although the sense of a message often was not evident until after he had read the message he had copied again. This young man needed help on this night, and he did not get it.

Engines 2001 and 941 were ready to depart Thebes for Salem Yard with 61 cars and caboose. They received order No. 222: "Engine 2001 run extra Thebes to Salem Yard meet Extra 1917 South at Omar," and left. At Cypress, 16 miles south of Goreville, order No. 106 was received: "Extra 2001 North meet No 185 at Omar and meet Extra 1917 South at Goreville instead of Omar Extra 2001 take siding at Goreville Extra 1917 South gets this at Goreville." The dispatcher wanted to help Extra 2001 as much as possible, so it was necessary to restrict Extra 1917 at Goreville, since it had already passed the last open office north of Goreville, West Frankfort, 24 miles north, where it had received order No. 222 on Form 31. The northbound train was ordered to take siding, leaving the main line for the southbound train, who would arrive at Goreville not knowing better. All of this is quite correct, and properly protected.

Except, of course, for the green operator at Goreville. Two southbound trains were to receive order No. 106 there, and both were restricted. As order No. 106 was transmitted, No. 185 was waiting impatiently at Goreville at the train order signal at Stop, its normal position on the C&EI. There was pressure, and the operator did not perceive that he was to make 5 copies, not the usual 3. He copied and repeated the order, and handed it up with a clearance card to No 185, which departed. The third copy went on the hook under the desk. A while later, he was copying another order for a northbound train when Extra 1917 approached. Seeing no orders to be delivered, he cleared the train order signal, apparently not breaking the dispatcher and asking him if it was all right to clear Extra 1917.

Extra 2001 and Extra 1917 met 2.7 miles south of Goreville, and 2.0 miles north of Omar, at 2.30 am. A deadheading conductor and the fireman on engine 941 were killed. The failure to deliver order No. 106 had resulted in a lap of authority between those points. Although Extra 1917 was restricted at the point at which it would receive the order, the dispatcher did not instruct the operator on what to do in this case--to be on the platform with stop signals, and to inform the engineman immediately that his train was to be restricted there. This would have detected the lack of copies for Extra 1917. Also, the operator should know to consult the dispatcher before clearing Extra 1917. So little noise was made that it was forgotten than an order was held for it. The dispatcher should have enquired about this order when he got the OS for No. 185, and reminded the operator of it. The pressure of the moment also contributed, but all the blame should not fall on the inexperienced operator.

The ICC report states that the inclined position of the 3-position train order signal was Caution, which, of course, it was not. It simply means that a 19 order is to be handed up without stopping the train. There never was a satisfactory definition of Caution. Perhaps the best was "proceed prepared to stop short of train, obstruction or switch not properly lined, but not to exceed restricted speed," or "proceed prepared to stop within one-half the range of vision." There were also very specific meanings of Caution in connection with certain signals or events.

When a train order says a train is to be restricted at the point that it receives the order, the operator should take special action, which includes putting a red lamp on the platform, and placing torpedoes 700 ft or so in rear of the train order signal, which should be observed at Stop and lighted. He should meet the train as it stops and tell the engineman about the restriction. Compare this prescription with the actions of the operator at Custer City, OK, for example. Where the train order signal only displays Stop and Clear, the operator on the platform with the signal at Stop usually means that Form 19 orders are to be handed up, so Josserand recommends that the operator not go onto the platform if the train is to be restricted at that point. Several of the accidents discussed here show this is not safe. At Goreville, the train order signal was three-aspect, so this stricture does not apply.

On 14 September 1944, southbound First No. 95 ignored a meet with northbound No. 90 at Atherton, IN, and collided with it at Dewey, IN, 6.3 miles south of Atherton and 3.7 miles north of Terre Haute, IN on the Evansville Subdivision of the Chicago and Eastern Illinois. There were automatic block signals, but they failed to prevent the collision as the engineman of First 95 neglected to observe them as he had neglected the meet at Atherton. First 95 was a 14-car limited overnight between Chicago and Nashville, while No. 90 was a solid mail and express train of 15 cars. The C&EI had an excellent safety record, and this was one of the very few accidents to its passenger trains, but 26 passengers and 3 employees lost their lives.

The C&EI was a heavy passenger hauler on its Chicago-Evansville main line, which connected with Louisville and Nashville lines to all points south, carrying six or seven trains a day in each direction. It brought coal up from southern Illinois as well, on a line originally intended to be the Chicago main of the Frisco, crossing the Mississippi at Thebes, but this dream was defeated by the alliance of business enemies and the government around the turn of the century. The main line was double or triple track from Chicago to Clinton, IN, single track beyond to Evansville. Miller system automatic train stop, installed by the C&EI in 1911 at its own instance, not that of the ICC, extended from Dolton Junction near Chicago to Clinton. This system used lineside ramps and a shoe on a tender truck. The passenger speed limit on the main line was 80 mph. In 1944, CTC was only a few years in the future for the Evansville Subdivision, and trains were operated by time table and train order.

The night of 13-14 September was foggy along the Wabash, which was crossed on a rickety truss bridge at Clinton (10 mph speed restriction). No. 90 received the following three orders at Terre Haute just after midnight: Order No. 3, Form 31: "First and Second 95 engs 1019 and 1012 have right over No 90 eng 1011 Clinton to Terre Haute." Order No. 11, Form 19: "First 95 eng 1019 meet No 90 eng 1011 at Dewey First 95 take siding Second 95 eng 1012 gets this order and wait at Clinton until 2 40 am." Order No. 13, Form 19: "First 95 eng 1019 take siding and meet No 90 eng 1011 at Atherton instead of Dewey." These orders are clear and easy to understand, except that the reason for reversing the right and having First 95 take siding is unclear, since order No. 3 seemed to have the purpose of requiring No. 90, a less-important train, to take siding although it was superior by direction. Order No. 11 established a meet with First 95, and helped No. 90 to the end of two main tracks, as if it were more important. Order No. 13 moved the meet farther north to Atherton. No. 90 passed Haley, 2.7 miles south of Dewey, at 2.14 am, 1 hour 1 minute late. Approaching Dewey, Stop signals were encountered and the train stopped, then proceeded at restricted speed in the murky fog.

First 95 had received order No. 3 at Cory, just south of Danville. At Clinton, order No. 13 was received, but in the modified wording: "First 95 eng 1019 take siding and meet No 90 eng 1011 at Atherton." The dispatcher simplified things for First 95 by not issuing order No. 11 and then immediately superseding it, instructing the operator at Clinton to drop the "instead of" part. This got the job done, and was not dangerous in this case, but it was a blatant violation of the Standard Code to issue orders in different words to those who were to execute them. The ICC investigator did not object. First 95 passed Otter Creek Junction at 2.18 am, 37 minutes late. The Pennsylvania line from Crawfordsville came in here, 1.9 miles from Dewey. First 95 passed several signals at Proceed or Approach until the Stop signal caught the engineman's or fireman's eye and brakes were applied in emergency, reducing speed to about 35 mph at the instant of collision, so it was a very hard impact.

The orders were handed up at Clinton to the engine, and to the front brakeman, riding in the sixth car. The brakeman read and understood the order, but did not give the order to the conductor and flagman. Instead, he made his way forward so he could drop off and open the north switch at Atherton, which was only a couple of miles away by that time. He took no action as the train went on and passed Otter Creek Junction, and was still inert when the brakes went into emergency. We cannot know what went on in the engine cab, but the events are consistent with the engineman and fireman falling asleep or into deep reverie just after Clinton, though with a meet at the next station, only 4 miles away, this is hard to believe. The lack of a meeting-point whistle seems to say that the meet was ignored. The conductor, of course, knew nothing of any meet, and cannot be blamed for not observing Rule S-90, only for not finding out at once what was in the orders received at Clinton. The brakeman should have known, but he seems to have been witless.

The ICC investigator makes a great to-do over the block signals and rule 509 pertaining to stop-and-proceed actions. The situation at Dewey as First 95 approached the standing No. 90 is shown in the diagram. Signal 172-7 would have been at stop, and Signal 171-9 at Approach as soon as No. 90 entered the overlap south of Signal 173-8, and No. 90 was then moving at restricted speed. First 95 would have encountered Signal 171-9 at Approach, and then had 9587' to stop, quite ample for the maximum speed. Even if No. 90 had moved farther, the stopping distance would not be less than about 4000'. It seems to me that there was nothing whatever at fault with the automatic block signals, except that it always reduces protection against opposing trains to stop-and-proceed, and there is no way to correct this with overlap signals. See Quincy, MS for a similar example. If First 95 had observed the signals properly, there would have been no accident. Apparently, no action was taken at the Approach signal, so it is no wonder that an emergency application when the Stop came into view through the fog would not suffice. The ICC report mistakenly assumes that signal 171-9 would not display Approach until No. 90 had passed Signal 173-8, which is untrue. No comment is made about the alteration of order No. 13, and the lack of a whistle signal (though the pertinent rules are quoted). However, the causes of the accident are correctly stated.

This tragic accident was one of two involving gas-electric motor cars in which the consequent fire led to a heavy toll. The other, at Cuyahoga Falls, OH, is discussed elsewhere. Tipton Ford was a result of ignoring a meet order, but the cause is completely unknown.

Motor car 103 made up Missouri and North Arkansas train No. 209, running from Joplin, MO to Harrison, AR. Between Joplin and Neosho, the M&NA used KCS tracks. In the twenty miles, there were two sidings, Saginaw near Joplin, and Tipton Ford a little more than halfway between Joplin and Neosho. Tipton Ford no longer appears on road maps, though Saginaw does. The M&NA was a direct, branchless line from Joplin to Helena, AR, 368 miles long, through beautiful but impecunious Ozark country, paralled as a through route by a Missouri Pacific line not far to the north of it, and by the Frisco's Springfield-Memphis line. When it was largely abandoned except for a few short segments after World War II, the 65 miles between Seligman and Harrison survived for a while as the Arkansas and Ozarks Railway.

Motor car 103 was of light steel construction, 70' long and seating 65. It had a 150 gal. fuel tank, which contained 105 gal. of gasoline at the time of the accident. It was totally consumed by fire when telescoped by the engine of northbound KCS First 56, a passenger train with engine 805 and 7 cars. The collision occurred on 5 August 1914 at 6.00 pm, 3500' south of the south siding switch at Tipton Ford, with both trains running at about 35 mph. 38 passengers and 5 employees, among them the entire crew of No. 209, were killed.

No. 209 left Joplin at 5.30 pm, 1 hour and 15 minutes late (probably after waiting for a connection). First 56 left Neosho at 5.40 pm, 3 hours and 15 minutes late, superior to No. 209 by direction. It was necessary to help No. 209 from Joplin against the delayed train, so dispatcher Sebring at Pittsburg, KS composed order No. 84: "First No 56 meet No 209 at Tipton Ford and wait at Tipton Ford until 5 50 pm Saginaw 6 pm for Extra 563 South." Engine numbers are not given in the ICC report, so perhaps they were not used on the KCS at the time. Also, the time 6 pm, an even hour, was later not permitted; it would be 6 01 pm instead.

Extra 563 had arrived in Joplin at 4.10 pm, and was to follow No. 209 out. Both Extra 563 and No. 209 were given copies of order No. 84 on Form 31. Conductor Gibson of Extra 563 signed for it, and the order was made complete for him, and then Conductor Nicholas of No. 209 signed for it, and received complete at 5.08 pm. A carbon copy of the order was produced in evidence, with the signature that was not proved not to be Nicholas'. Operator Hadley was regarded as an excellent operator, but it still had to be established that he did not fake the signature himself after failing to deliver the order. Hadley had been fired from the AT&SF years earlier after he failed to deliver an order at Holly, CO, but his career at Joplin had been faultless.

The single irregular event was that Conductor Nicholas had not registered out, the first time he had been observed not to do so. This may have been evidence of some mental irregularity, but he was not noticed to act in an unusual manner. The engineman would have required a clearance card, on which Order No. 84 would have appeared, and he would have read it. Something was required on No. 56 to have left Joplin at any rate. The reason for the simultaneous lapses by conductor and engineman is completely unknown, but it is consistent with other similar accidents that if the engineman alone forgot the meet, the conductor would not have taken the required action. All the men involved were experienced and reliable.

Traffic was not handled at Saginaw and Tipton Ford--they were simply sidings, and could not economically be open offices. If they were open offices, a "middle order" could have been used, and this would have prevented the accident, as the ICC investigators pointed out. The KCS recommended middle orders, in fact, but they could not be used here.

The single track of the Nickel Plate parallelled the multiple-track main line of the New York Central System along Lake Erie in this area, which was the Buffalo Division, Buffalo to Conneaut, 116.8 miles. In 1928, it did not have automatic block signals, but a manual block was used for passenger trains only. The Form 31 train order was not used, only Form 19 and Form 17 (the latter peculiar). With both forms, complete was given upon repetition. A Form 19 had to be used to restrict a train. With this form, the operator was required to read the order to the conductor and engineman after delivering it personally to them, and then to take their signatures. This was not required for a Form 17, but this form could be used for helping orders only.

These procedures were unfortunate. Reading the order to the conductor and engineman did not ensure that they understood it, and usually meant that the order might not be read by them at all, just stuffed in a pocket. Most rule books require reading to the operator when this is required. Secondly, getting both conductor and engineman in the office simultaneously is difficult, especially on long freight trains. In fact, usually only the conductor had to sign for Form 31 orders, filling in the engineman's signature. These difficulties are underlined by this accident.

The stations in the area of interest are, from the west, Thornton Junction, then 5.9 miles to Fairview Pit siding, with Fairview station just east, 3.6 miles to Swanville, and 8.5 miles more to Erie. Thornton Junction, Fairview and Erie were train-order offices. Second 58, engine 611 and 94 cars and caboose, received order 48, Form 19, at Thornton Jct.: "2nd No 58 eng 611 will wait at Fairview Pit until four forty 440 pm for No 51 eng 640 No 13 and 2nd No 58 will meet at Swanville No 13 will take siding." This order was received at 4.12 pm, and Second 58 was 5 hours and 17 minutes late leaving. This order shows the old "will wait" and "will meet" phrasing rather than the later "wait" and "meet." It is a relatively complex order, but easily comprehensible.

What happened was that the engineman formed the impression that he had until 4.40 pm to reach Summerville, where No. 13 and No. 51 would take siding, blotting Fairview Pit from his memory. So, he approached Fairview Pit at 4.29 pm and kept going. No. 51, with engine 640, 89 cars and caboose, and running 1 hour 36 minutes late, received order No. 48, Form 17, at Erie at 4.11 pm, making good time at 30 to 40 mph. No. 51 did not reach Fairview Pit, which it could have made by 4.40 pm, but collided with Second 58 a short distance west of Swanville at 4.32 pm. Although each train was running at 15-25 mph at the instant of collision, there were fortunately no fatalities, and lots of people to give evidence later.

It did not contribute to the accident, but the method of delivery by operator Ryan at Thornton Junction to Second 58 is curious. He said that engineman Hites jumped off the engine, without stopping the train, so the order could be read to him and his signature could be obtained, then clambered back aboard somehow (a likely story!). When the caboose came by, he handed up the order with a hoop to the flagman, who was on the rear platform. We suspect that both Form 19 and Form 17 were usually handed up without stopping the trains, the reading being omitted and the signatures forged. An impractical rule will usually call forth such ingenuity. The order did, anyway, get to everyone concerned, and was read and understood by all but the engineman, who formed a deviant interpretation.

Conductor Friend had numerous reasons why he did not pull the air as his train passed Fairview Pit before 4.40 pm without seeing a train there. First, he said the engineman may have received orders at Fairview that he did not know of (but Fairview was beyond Fairview Pit!). Second, he was worried about a possible dynamiter in the train--a car brake that would go into emergency on any brake application--and this would break his train in two. Third, when he saw No. 51 coming, he was afraid the two engines would meet on the bridge if he did so. This creativity demonstrates that he knew his duty, but just did not do it. Flagman Stearns also said he thought about it, but by the time he got to the valve, the brakes were already in emergency. The story is drearily familiar.

Largo, California on the Northwestern Pacific was the site of two accidents, one in 1929 and the other in 1952. The earlier one involved northbound passenger train No. 2, which was a mail-baggage combine, baggage, smoker and coach, pulled by engine 141, and a southbound freight extra, Extra 184 South, with 31 cars and a caboose. At Hopland, the first open office south of Largo, No. 2 received order No. 32 on Form 31, which read: "No 2 Eng 141 meet Extra 184 South at Largo," and left at 11.40 am. Extra 184 received this order on Form 19 at Ukiah, 10 miles north of Largo, and departed at 11.27 am.

Every crewman on No. 2 saw the order, except for the flagman, Allen, who could not leave the last car. Though they had met other trains on the way from Tiburon, and exchanged the meeting point signals, this time the whistle was silent at Largo, and No. 2 simply passed by. The engineman and fireman of Extra 184 were killed in the subsequent collision, as well as the engineman of No. 2. We note that the two orders were in correct form, properly delivered, and were not delivered far from the point at which they were to be executed.

By the time of the later accident, the line was controlled by the Southern Pacific, so north became east and south became west, since on the SP the direction toward San Francisco is called west, whatever it may be by the compass.

On 29th October 1952, the 3rd trick dispatcher issued order No. 40 at 7.19 am: "Engine 2541 run extra Willits to Tiburon Engine 2348 help Extra 2541 West Willits to Largo then run extra Largo to Ukiah. Extra 2541 left Willits at 8.50, reached Ukiah at 1.48 and Largo at 2.15 pm, after which engine 2348 started back towards Ukiah.

At 10.38 am, the first trick dispatcher issued order No. 57: "Engine 5261 run extra Willits to Schellville." Schellville was on a branch from Ignacio, near the southern end of the line. Extra 5261 left Willits at 11.55 am, and passed Ukiah at 2.05 pm. Although the transfer had gone normally, the first trick dispatcher had overlooked the returning helper. The collision was at 2.30 pm in clear weather, both trains moving at about 20 mph. The engineer and swing brakeman of Extra 5261 were killed, and four other employees injured.

This is Little Rock, Washington, on the Northern Pacific branch line from Gate to St. Clair, 28.6 miles. As the diagram shows, it is about halfway between Gate and Olympia. A collision between two freight extras occurred here on the 20th of September, 1944. The collision was on a curve, and the train speeds could not be substantially reduced below the 35 mph at which they were running previously.

The dispatcher responsible for this subdivision had 11 subdivisions under his supervision, mostly branches, but which amounted to 465 miles, and there were 29 active movements at the time of the accident. At 1.07 pm he completed order No. 29 to C&E engine 1903 at Olympia: "Eng 1903 run extra Olympia to Blakeslee Jct." Extra 1903 departed with 60 cars and caboose at 2.00 pm. At 1.51 pm he completed order No. 32 to C&E engine 1802 at Gate: "Eng 1802 run extra Gate to Saint Clair." Extra 1802 then started out immediately at 1.55 pm. Nothing could now be done. The fireman and front brakeman of Extra 1903 were killed.

This is a classic case of lap running orders for extras, one of the greatest hazards faced by a dispatcher. He knows he is solely responsible for the outcome, and it is a heavy burden. This dispatcher was overworked, but this was not unusual.

Jesup was an important junction on the Atlantic Coast Line, between Waycross and Savannah in eastern Georgia. On 24 March 1949, a collision of a passenger train and a light engine occurred just south of the south switch of the siding at Jesup, on the Folkston-Jesup via Waycross line, 73.3 miles, which was operated by train orders and had no automatic block signals. The traffic on this line amounted to 12.4 trains per day, which was easily handled by train orders. The north switch of the 1.28 mile siding at Jesup was 792' south of the station. The siding was within yard limits. The next siding south of Jesup was at Slover, 4.9 miles south, and the next open train order office was at Offerman, 19.2 miles south.

No. 231 was a southbound third-class freight that arrived Jessup at 6.30 am with 93 cars and caboose, pulled by engine 1720. A clearance card and two train orders were received at the station. One of the orders, No. 34, said: "2nd 58 wait at Slover until 745 am for No 231." First 58 had not yet been met. The train stopped on the main line clear of the south switch of the siding, whistling out a flagman to protect the rear. While a switcher removed 13 cars from the rear of the train, the engine pulled forward and backed onto the siding, where it took water and coal, rejoining the train about 6.50 am.

The engine crew wrongly interpreted the order to refer to No. 58, which they had not yet met, and which was much on their minds. Not knowing that No. 58 was running in sections, they assumed the dispatcher had helped them against the schedule. No. 58 was due at Slover at 7.02 am, Jesup 7.18 am, which meant they would have to be in the clear by 6.57 am at Slover and 7.13 am at Jesup. Without the order, they would have to remain at Jesup for No. 58. However, they now believed they had until 7.40 am to make Slover, which they could easily do in the half-hour interval. The flag was whistled in at 6.56 am, and preparations made for departure.

There were delays in leaving, and time grew short. By twenty past seven, it was clear that there was not sufficient time to make Slover, so it was decided to flag No. 58. At 7.23 am, the engine was detached from the train and run south within yard limits about a mile from the south switch to put a flagman there. Just as the flagman was getting down, the headlight of First 58 was seen through the fog and the engine rapidly reversed. The flagging was too late, and the engine reached about 15 mph in reverse before it was struck at 7.27 am by First 58, running at 59 mph. The collision killed the engineman and fireman of First 58. First 58, with engine 1509 and 6 cars, had left Offerman at 7.06 am, 27 minutes late, and had made up about 15 minutes over the 19 miles.

Not much is said about what the rear end crew was doing during this time, except that they had understood the order and thought their train was remaining at Jesup for First 58. They should have fretted about their train's presence on the main line, and have seen that it was protected after 6.57 am, if they could not get it on the siding, which would have been inconvenient. Apparently, a belated effort was made to flag First 58, which we have just examined.

The claim was made that somebody's thumb covered the important word "Second," as in the accidents at Mango, also on the Coast Line, and David on the Western Pacific, in all of which an order was taken as referring to the schedule, when it only referred to a single section. This was probably a metaphorical thumb, representing an attempt at a logical explanation for a mental process that actually did not register the modifier. As has been frequently pointed out in these studies, once a perception has been formed, it is held in spite of evidence to the contrary. The mind really does not see the "Second" or similar word, though it is physically there.

Perhaps it would have been a good idea to have mentioned First 58 as well, either by delivering a copy of the order to carry signals to No. 231, or by mentioning it explicitly. On the Canadian Pacific, we have noted the practice of writing an order like this as" "First 58 on time Second 58 wait at Slover until 7.45 for No. 231." This makes the situation crystal clear, and would have avoided the misunderstanding that led to this accident. Unfortunately, this practice does not seem to have caught on.

Now it is 25 years earlier than Jesup, on the Tampa District, Third Division, which ran from Lakeland to Port Tampa, 41 miles, via Plant City. Northbound trains are superior by direction, and there are no automatic block signals. There was a short double-track segment from Tampa Union Station to Uceta, 3.2 miles. See the accident at Nashville for a discussion of how these short double track sections are operated. Mango was located 5.5 miles north of Uceta, and Seffner 2.2 miles north of Mango.

On the clear evening of 10 October 1925, southbound No. 89 with engine 473 and 7 cars departed Lakeland at 9 pm, 3 hours and 10 minutes late. At Plant City, it received the following orders: No. 137 (Form 19), "1st No 82 meet No 89 on double track at Uceta," and No. 139 (Form 31), "No 89 pull by back in siding meet 1st No 82 at Seffner instead of on double track Uceta." The first order really meant that First 82 was to wait at the end of double track for No. 89, while the second superseded this, putting the meet at the blind siding at Seffner. It would be interesting to know whether the second order would be interpreted as permitting No. 89 to move south of the siding switch at Seffner without flag protection (see the KO&G accident at Kenefick for a similar case). According to the Standard Code, it would not. No. 89 left Plant City at 9.33 pm, now 3 hours 23 minutes late.

Second 82, engine 457 and 3 cars, departed Tampa Union Station at 9.40 pm, 10 minutes late, and received order No. 139 at Uceta, permitting it to proceed, which it did at 10.03 pm. Note that Second 82 was closely following First 82. No. 89 met First 82 at Seffner on order No. 139. The green signals were whistled and acknowledged. Although it had nothing whatsoever against Second 82, No. 89 proceeded southward. The fateful thumb had covered the "First" in order No. 137, so it was supposed that all sections but the first would be at the end of double track at Uceta. The conductor and engineman both agreed to this, and did not show the orders to any other members of the crew. The innocent engineman of Second 82 was killed.

The dispatcher probably thought that mentioning First 82 would show that No. 82 was running in sections, and would keep No. 89 at Seffner. However, this was quite unsafe, and order No. 139 should have read "No 82" not "First 82," which would imply that some other arrangment had been made for Second 82. It seems that the only safe way is to mention all the sections or none of the sections, to avoid the dreaded thumb.

We note that trains were directed to meet "on double track" at Uceta. Under the Standard Code, a meet is always a meet, and if the trains are directed to meet at Uceta, then the train going on double track must wait there for the train coming off. In this case, the nonstandard phrase "on double track" surely meant that the train would proceed at once, effectively making the double track a long siding. In fact, these short sections of double track, as is remarked elsewhere, are just that, and trains meeting must identify one another to avoid having to check a register at the end of double track (which is not provided). The purpose of these sections was to reduce congestion, and they could not do that unless operated in this manner, which was relatively safe, in spite of the disaster at Nashville. Josserand (p. 309) says that the train going on to double track can proceed at once if ordered to meet there, but such a meet order should not be issued. It is better to give the train right over the other to the end of double track.

Shaw discusses the accident on the Frisco at Marshfield, MO on 17 September 1918 fairly well, but since this is such an important example of the rare dispatcher's error it is worth comment here. Shaw does not give a totally coherent account of the train orders involved, for one thing. The fact is that Second 39, a westbound freight had no knowledge of Passenger Extra 1260 until they collided at Marshfield, while Extra 1260 held an order helping it against Second 39. This could be either a failure to deliver the order at Conway, or a failure of the dispatcher to transmit the order to Conway compounded by clearing the train at that point. The preponderance of evidence indicates that the latter was the case.

An extra train can either clear an opposing regular train's schedule time or be "helped" against the possibly delayed regular train by train order. An important extra will usually be instructed to meet a less important regular train by an explicit order, such as: "Second 39 eng 56 meet Extra 1260 East at Marshfield." This order restricts Second 39, not permitting it to pass Marshfield until Extra 1260 arrives there. It simultaneously helps Extra 1260 East by permitting it to ignore the schedule of Second 39 as far as Marshfield, where it must take siding at the west switch as the inferior train. Meet orders are the safest way to handle these situations, safer than run-late or wait orders.

The dispatcher must ensure that the order will be delivered to the superior train (Second 39) before the inferior train (Extra 1260) is allowed to act on it. This is done by having the operator who will deliver the order to Second 39 either repeat the order, or give the "X" response. This ensures that the train order signal is against Second 39, and that the operator is aware of his responsibilities. It would be best, of course, to actually have the order delivered and the signatures of the conductor and engineer sent to the dispatcher in receipt of it. The "31" order form provides for these signatures, and "complete" is given after they are sent. The "19" order form does not require signatures, and complete is given after it is correctly repeated. A completed order may be acted upon. In practice, there is no significant difference in the degree of safety of the two forms, and a train is considered "held" as soon as the order is repeated or X'd.

There is one case in which an order must not be made complete to the inferior train until a Form 31 order has been made complete, and that is when the order is sent in care of someone other than the operator. Only the signatures can guarantee that delivery has been made. This form is always available in case of need (signatures can be written on the Form 19 order if the Form 31 order is not used). In the present case, under Frisco rules of that time, a Form 31 should be sent to No. 39 at some station east of Marshfield, such as Conway, and a Form 19 to Extra 1260 at some station west of Marshfield, such as Springfield. It would have been even better if a copy of the order had been transmitted to the operator at Marshfield, who would keep his train order signal at stop until one or the other of the trains had arrived and it was obvious that they were aware of the meet. This is called a "middle order," and is excellent practice.

The actual case in 1918 was a little more complicated. The actual order helping Extra 1260, order No. 115, was a schedule wait order between Springfield and Newburg giving right over all but first class trains (and therefore over Second 39) and prescribing a maximum speed of 30 mph. Therefore, a middle order would not have been appropriate, since the meeting point was unknown. Engine 1260 was a 2-8-0, and its train was 6 sleepers, 1 baggage, 6 sleepers and a caboose. At Strafford, west of Marshfield, it received order No. 123 to run 15 minutes late on order No. 115. At Marshfield, it picked up order No. 127 stating: "No 9 eng 1069 meet passenger extra 1260 east at Conway." For its part, Second 39 received order No. 115 at Newburg and order No. 113 at Lebanon, reading: "Second 39 eng 56 has right over No 32 to Marshfield and hold main line." This order was imperfectly worded, since it should have given both limits of the right--"Lebanon to Marshfield," as well as "Second 39 hold main track at Marshfield." All the details of the orders are not given in the ICC report, and we have the digitization garble to contend with as well. These orders could be much more clearly worded.

The operator at Conway testified that he was out of the office for supper at the time the order was transmitted, and that the train order signal was fastened at clear, since he held no orders for trains in either direction. This is normal practice when the station is closed. The report does not say if Frisco practice was to keep the train order signals at stop, or to fasten them at stop only when an order was to be transmitted. He received no comment when he OS'd the train to the dispatcher. This account was supported by an insurance agent who had eaten dinner with him, and accompanied him back to the office. No record of the order was found at Conway, but then again, the evidence could very easily have been destroyed.

The dispatcher testified that he had transmitted the order to Conway for Second 39, and that it had been made complete there. The time given for the transmission of the order was when the operator was out of the office. Again, the record in the order book could have been manufactured after the accident. The dispatcher had a number of other trains to deal with, including the Meteor, No. 9. It apparently was his practice to ask operators to repeat only the words relevant to their stations, not the whole order, to save time. This practice is strongly deprecated.

What probably happened was that Extra 1260 was ready to go at Springfield, but could not leave without something on Second 39, since the schedule was overdue. The dispatcher tried to raise Conway, but could not because the operator was out of the office. He would no doubt return before Second 39 passed Conway, so the dispatcher completed the helping order at Springfield at once so that Extra 1260 could get on its way, intending to transmit the order to Conway as soon as the operator returned. Unfortunately, by that time he had forgotten all about the matter, and not even the OS of Second 39 at Conway restored his memory. The call from Marshfield did, however. The order book is supposed to prevent this sort of thing (and it does), but the critical error was completing the helping order before the superior train was held.

There were automatic block signals on this line that would have prevented the accident had they been properly observed. A signal displaying Approach was run through without slackening speed by Extra 1260, and so no stopping distance remained at the Stop signal. One possible reason was that the automatic signal was connected to the train order signal, and showed Approach when "19" orders were to be handed up. The engineman of 1260 took this to be the reason for the Approach, not the Stop aspect at the next signal. There was testimony that after the order was received, the throttle was opened. Neither the ICC report nor Shaw points this out. Second 39 was expecting to meet eastbound No. 32 at Marshfield, and so was not running rapidly (the engine, a 2-10-2, could not run fast, anyway). Three men on engine 1260 died in the collision, but not the engineman, and 12 soldiers in a wooden tourist sleeper at the front of the train.

It is never good for a signal aspect to have more than one meaning. If automatic signals are connected to the train order signal in this way, a rule is necessary stating something like: "Proceed prepared to stop at next signal or to receive train orders without stopping. A train exceeding medium speed must immediately reduce to medium speed. Speed must not be increased until the automatic block signal in advance is seen to indicate Proceed." Shaw mentions a similar happening on the Missouri Pacific in 1922 at Sulphur Springs, MO directly after he discusses Marshfield.

The most tragic railroad accident in Oklahoma happened on the SL-SF at Kellyville, OK, 8.4 miles west of Sapulpa, on 28 September 1917, almost an exact year before the Marshfield collision. All of the 32 killed, and most of the injured, were travelling in the Jim Crow coach of westbound Train No. 407. No. 407 was composed of engine 1035, a 4-6-2, a steel baggage car, mail car, two coaches, a chair car, three sleepers and an office car, all wood except for the baggage car. In the accident, the mail car was driven up and over the first coach, which it telescoped. If all the cars had been steel, there would probably have been no fatalities. No. 407 was en route from St. Louis to Lawton, OK (the location of Fort Sill), and left Sapulpa at 2.12 pm.

This accident was a classic example of misidentification. Order No. 56 stated: "Extra 1322 East has right over No 407 Depew to Kellyville" while Order No. 61 read: "No 406 gets this and meet No 407 at Kellyville." There was also an Extra 1343 moving eastward against No. 407's time in the area. Passenger trains No. 406 and No. 407 had stops at Kellyville. Extra 1322 was an empty troop train returning from Fort Sill, with 29 tourist sleepers, 2 loaded box cars and a caboose, which left Oklahoma City at 10.30 am, and passed Bristow, 13.4 miles west of Kellyville, at 2.12 pm. Extra 1322 was, like the Extra 1260 at Marshfield, a troop train pulled by a fairly new 2-8-0, but empty in this case. Extra 1343 was a freight that had some work to do in Kellyville.

At this time, Frisco cabooses displayed train identification in an illuminated box on the caboose. At this period, new Frisco locomotives had a small engine number on the cab side, with a large FRISCO on the tender. However, Frisco practice both before and after was to have a large engine number on the tender. The engine number was also painted on the sand dome, shown on a plate on the smokebox door or headlight bracket, and illuminated at the sides of the headlight. A few engines had illuminated engine numbers on the smokebox near the stack, a la Santa Fe. Illuminated numbers near the headlight are useless at night when the headlight is on bright, but are quite visible when the headlight is extinguished. The prominence of the engine number on most American locomotives is because of the use of the engine number for train identification. Train orders always state things like: "No. 407 Eng 1035." If you see a train with engine No. 1035, it will be No. 407. Extra trains are, of course, only identified by the engine number.

The right order made certain that No. 406 would reach Kellyville, or No. 407 would clear its time as far as Depew. The meet order made the meet certain, and it was there when No. 407 arrived. Since it was superior to No. 407 by direction, and also because it received the meet order there, it should have held the main line, but the report does not discuss this topic. Extra 1343 was on the siding, having decided to clear No. 407 there. When No. 407 arrived, it saw a passenger train and a freight train, just as one would expect from the train orders. Remarkably, nobody on No. 407 took care to notice the engine number on the waiting freight train, at any one of the five places where it was clearly shown on the engine and caboose. No. 407 made its station stop, with its engine near engine 1343, and left at 2.30 pm. It encountered Extra 1322 East at 2.42 pm, 2.3 miles west of Kellyville, with the results detailed above.

The dispatcher acknowledged that there was a chance Extra 1343 would be taken for Extra 1322 (the consist of the trains was not known to the crew of No. 407) when two trains were seen at Kellyville. He also observed that a "middle order" sent to the operator at the open Kellyville office would have prevented the accident. The operator would not have allowed No. 407 to leave before Extra 1322 arrived in this case. Although the middle order was recommended in the Standard Code, the Frisco had not adopted it. The reason given was that it would take too much time. The only additional time is that required for the operator at the meeting point to repeat the order. In this case, the right order instead of a meet order was a very poor choice. Unless some time order had been given to Extra 1322 (which is possible) it should be deprecated.

Again we see the neglect of the conductor's duty to be jointly responsible with the engineman for the operation of the train. He, as well as the engineman, is responsible for identifying that the proper train has been met at a meeting point. In this accident, the engineman was careless, and committed an error. The conductor, on the other hand, neglected his duty and should have been held principally responsible for the accident. The ICC report discusses the background of these two men in some detail. The conductor had been suspended, barred from passenger service, dismissed and later reinstated in passenger service due to a shortage of conductors. The engineman had been promoted to engineman, demoted to fireman, and later reinstated. It was becoming very difficult to find satisfactory men, especially in wartime.

The trains involved in this 31 May 1920 head-end collision on the SL-SF were No. 112, the overnight Oklahoma City-Kansas City train, and No. 403, the St. Louis-Oklahoma City all-stations local train that became simply No. 3 a few years later, and even later The Will Rogers, after most of the local stops had been eliminated. White Oak was between Chelsea and Vinita on the Cherokee Subdivision of the Southwestern Division, which extended from Monett, MO to West Tulsa, OK, 143.2 miles.

No. 112, consisting of engine 1049 and 8 cars, including 4 sleeping cars, was on time and due at White Oak at 12.12 am. At Claremore, order No. 86, Form 19, was received, stating: "No 112 engine 1049 wait at White Oak until 12.17 am for No 403 engine 1330." At Chelsea, 12.4 miles west of White Oak, order No. 88, Form 19, was received, which said: "Orders No 84 and No 88 are annulled. No 403 wait at White Oak until 12.38 am Witt 12.44 am Catale 12.50 am Chelsea 1.00 am Bushyhead 1.10 am Foyil 1.14 am Sequoyah 1.20 am Degroat 1.25 am." No 403, which was running late, had become even later, and a new schedule was given in this order for the benefit of opposing inferior trains. No. 112 was mainly interested in the annullment, which meant that it could proceed from White Oak on time, instead of waiting 5 minutes for No. 403. No. 112 was superior to No. 403 by direction, but it still had to approach all sidings prepared to stop when No. 403 was overdue until the the train was met, so the schedule also helped.

No. 403 had engine 1330, a 2-8-0, and 7 cars, including one sleeper. The Frisco seems to have used these new 2-8-0's, with 63" driving wheels, as dual-service engines. We noted them handling troop trains at Marshfield and Kellyville. No. 403's schedule, if it resembled the schedule of No. 3 of a few years later, allowed 14 hours and 45 minutes for the 424.6 miles from St. Louis to Oklahoma City, an average of 29 mph. However, there were 81 intermediate stops including three meal stops of 20 or 25 minutes, so allowing 4 minutes per stop, a net running time of 8.47 hours results, for an average of 50 mph, which is no mean achievement. It was no wonder that 1330 was a couple of hours late as it approached Vinita.

Order No. 86 was received at Afton by No. 403, giving it an extra 5 minutes to make White Oak. The order board was out at Vinita when No. 403 arrived, a little before midnight. The conductor of No. 403 went to the office and received order No. 88 there, with a clearance card, from the operator. Engine crews were changed at Vinita on No. 403, and the engineman being relieved handed the clearance card and orders he had received at Afton to the relieving engineman. The crew went over to the restaurant for refreshments, where the conductor joined them later. It seems that he had decided to leave Vinita at 12.05 in order to make White Oak by 12.17, 12 minutes for 7 miles. This was very short time, even though he only had to clear the departure time, since No. 112 was of the same class. This showed that, for some reason, the import of order No. 88 had not made an impression.

The fireman said the engineman had never received a clearance card and orders for Vinita, though the order board was still at Stop, so a clearance was required in order to leave. The conductor said he left everything up to the engineman about No. 112, and devoted himself to tickets and such. Apparently, the engineman proceeded on the basis of order No. 86, No. 112 passed White Oak at 11.12, and the trains collided 2 miles east of White Oak. One passenger, and both enginemen, were killed.

The conductor made the usual noises to excuse himself, saying that he had handed the orders and clearance to his engineman, tried the communicating signal, and pulled the air at a siding between Vinita and White Oak, none of which he probably did. The engineman, however, should never have left Vinita with the train order signal at Stop without a clearance form. The conductor should have carefully reviewed the orders in force with the new engine crew, to ensure that they were fully informed.

Quincy, MS was the next station east of Amory, the division point of the Birmingham Subdivision, Southern Division, 123.6 miles from Amory to Birmingham. This single-track line was equipped with overlap automatic block signals, 3-position upper quadrant with oil lighting. The passenger speed limit was 65 tangent, 55 curve, and the freight speed limit was 35 mph.

Northbound freight No. 238 left Thomas Yard, Birmingham, with engine 4208, a heavy coal-burning, stoker-fed Mikado built by Baldwin in 1930 with 64" driving wheels, 48 cars and caboose in the evening of 15 February 1939. After midnight, at Winfield, it was getting on the time of No. 107, the Sunnyland, so the dispatcher gave some help in order No. 9: "No 107 eng 1520 wait at Sulligent until 405 am for No 238 eng 4208." There was also a telegram for No. 238: "No 107 35 mins overdue Amory and not showing Call operator Sulligent as soon as you get there." While this was taking place, Conductor Connell walked up to the engine, inspecting his train, and when he got to the office, collected the order and telegram and climbed into the cab, anticipating a conference with the operator at Sulligent.

No. 238 received order No. 16 at Sulligent, giving more help: "No 107 eng 1520 wait at Quincy until 435 am for No 238 eng 4208." No. 238 departed Sulligent at 3.55 am, 14.3 miles south of Quincy. There would be time to make Quincy, and be in the clear by 4.40 am, since the second-class freight had to clear the first-class No. 107 by five minutes. The engineman claimed after the accident that he had read "Amory" for "Quincy" although the words are quite different,and the order was neatly written. The engine crew claimed that the order was read several times later, with the same error. Now Amory was 9.4 miles farther north, making 23.4 miles to go in 45 minutes. This would be smart running, since it would take 40 minutes running steadily at the speed limit to do it. Quincy, an open office, was reached at 4.18 am, an average of 37 mph, so it appears that 4208 could just have pulled it off, bending the speed limit a bit. Without restrictions, 4208 could roll at 60 or 65 mph with ease.

No. 107 had finally appeared at Amory, where it received order No. 16 and left at 4.08 am, 1 hour 43 minutes late on its scheduled time of 2.45 am. Oil-burning heavy 4-8-2 1520 with its 69" drivers, built by Baldwin in 1926, could also run, and was probably doing so with the nine cars of the Sunnyland. The dispatcher had underestimated its progress, and unless No. 238 were found on the siding at Quincy, it would have to cool its heels there for a quarter of an hour. No. 107 averaged 60 mph between Amory and Quincy, start-to-pass, so it was bending the speed limit itself.

The ABS used dual signals at the ends of sidings, with intermediate signals between sidings. There were long overlaps to ensure that opposing trains would receive restrictive aspects before they came together. The layout of the signals at Quincy was as shown at the right. Not all the overlaps are shown, only the ones given in the ICC report. The signals are shown with their aspects when the trains have their respective positions. In this case, No. 238 and No. 107 would each have accepted Clear aspects at signals 6228 and 6177 when they were about 4 miles apart. The next signal for No. 107 at signal 6203 would be Stop, and for No. 238 Approach at signal 6212. If No. 238 had been a little farther along, as was possible, then signal 6177 would also have shown Approach, more than two miles from the north switch, and signal 6212 Clear. Apparently, neither train responded to any signals at Approach. No. 107 arrived at too high a speed (especially in view of the expected wait). No. 238 responded only to the Stop at signal 6204, and overran it by 2158', a considerable amount, because of the 1.23% descending grade. There was no time to flag, and the collision occurred at 4.20 am in clear weather. The engineman and train porter of No. 107 were killed, along with a tramp on No. 238. There was a 3° curve at this point, but visibility was good.

It is a problem providing good protection for opposing trains with an overlap system. At the minimum, opposing trains will receive Stop aspects at the ends of a block, and at least one train will have received an Approach aspect. This appears to have been the case here, which was a severe test. Note that the overlap was extended to the east more than two miles. The report does not give equal information about the overlap to the east, but it is quite possible that both trains were given Approach aspects. A better solution is the use of Absolute-Permissive Block (APB; see the Attean, ME incident on the CP for an example), which effectively extends the overlap to the next siding. In this case, both headblock signals are at Stop, and preceded by Approach, so that opposing trains are stopped a block apart.

The ICC was perturbed that the ABS had not prevented the collision, but signals cannot do so if they are not observed. It was not mentioned that a middle order would have helped to slow down No. 238, and probably prevented the accident. Quincy was an open office, so this was a possibility. The other thing was the matter of the S-90 whistle signal 14(n