A Bad Day At SI

The collision at Silver Lake, Ohio on 31 July 1940

"SI" was the telegraphic call for block station Silver Lake, which was on the Pennsylvania's Columbus-Hudson, Ohio line about a mile north of the Cuyahoga Falls station, on the northeastern edge of the Akron metropolitan area. (These letters were later used for a different station in Ohio.) At about six in the evening of a hot, humid midsummer's day an appalling tragedy occurred just south of the unattended block station, as block-limit stations were then known. Silver Lake had become an unattended block station in 1931. Pullman-Standard gas-electric car 4648 was skimming southward at about 50 mph as Train No. 3380, coming from the connection with the Cleveland-Pittsburgh main line at Hudson. A pair of I1s 2-10-0's with 73 cars, approaching from the south, had shut off steam and was coasting to a halt. The gas-electric car smashed itself against the lead engine and was pushed back 537 feet as the heavy freight train stopped, its brakes in emergency. The 250-gallon fuel tank of the gas-electric car was ruptured; it still contained about 150 gallons of gasoline, which immediately caught fire and enveloped the car. Of 46 individuals on the car, only three escaped alive: the engineman, the conductor, and a deadhead track worker, who jumped just before the collision. The date was 31 July 1940.

A report of the ICC investigation of the accident can be found as Investigation 2440 at DOT, from which my knowledge of the affair comes. This is an interesting investigation for several reasons. First, we are curious how the dual protections of a highly-developed train order system, and a superimposed manual block system, failed on this occasion. Secondly, the fatal effects of the combustible fuel made this accident as costly as an air crash or ship sinking, where such a high proportion of fatalities is the rule, rather than the usual small toll of a typical railway accident. Finally, the manual block system was an economical form pioneered by the Pennsylvania in the early 1930's, where unmanned block station, or block limit stations, were substituted for manned block stations. In this accident, we see how the system was worked in practice.

We have Optical Character Recognition programs to thank for making the accident reports available to us on the internet. However, the DOT has not taken the trouble to proofread them, so there are frequent errors. In most cases, fortunately, these errors are easily recognized and corrected. The accident reports themselves contain inconsistencies and apparent errors here and there, and the analysis, though earnest and well-meaning, is not always the best. The pertinent train order in this case was Order No 99, given in the report as "Eng 4454 run Extra Arlington to Hudson and meet No. 3380 Gas Eng 4648 at Switch 1 Silver Lake." The term "Gas Eng" is a little unusual--one would expect "Motor," but there seems to be an omission. It is stated that everybody knew that the motor car would take siding at Switch 1 (switches to sidings are numbered from north to south on the Pennsylvania here), but the order as stated does not imply this. A more proper order would read "Eng 4454 run extra Arlington to Hudson hold main track and meet No 3380 Motor 4648 at Switch 1 Silver Lake."

Under order 99 as stated, No 3380, a first class train, would hold the main track at meeting points with extra trains. However, the modified order is quite correct, the dispatcher knowing it is much easier to get the motor car on and off the siding than the heavy freight train. The motor train should have let itself onto the siding at Switch 1, then pulled ahead to the block limit sign, where there was a telephone. The conductor or engineman would telephone the block station that controlled SI, which was Hudson, and report themselves clear of the block from Hudson to SI.

Extra 4454 would have pulled up to the SI block limit board when it arrived, and its lead engineman would have called the operator at Hudson to get permission to enter the next block, which extended to Arlington. The Hudson operator would know the block was unoccupied from his block record, and would have contacted Arlington to inform the signalman there that he intended to allow a train to enter the block. Arlington would then not allow any train to enter the block from that direction, and make a note in the block record. Since Extra 4454 required permission for only one block, this could be given verbally. Once this was done, Extra 4454 would move off towards Hudson. Strictly speaking, it would have to report itself clear of the AR-SI block before No 3380 could be given permission to occupy the block, but this would probably be omitted, since No 3380 could see Extra 4454 and could not proceed until it had cleared Switch 1.

We have just seen how two trains could meet at an unstaffed block-limit station between two staffed block stations. If No 3380 had been cleared between Hudson and Arlington, and there was no opposing train, it would have been given authority to pass SI by a Form K. The Form K is always used if permission is given to occupy a block in advance. It can be furnished with the train orders at an open block station, or the engineman can fill out a Form K in the telephone booth at a block-limit station. At least in 1940, the engineman did not have to furnish a copy to the rear end of his train. With train orders, both ends of the train had to have copies of each order. On the Cleveland division, there was a practice of the engineman's making two copies, and leaving one copy at the switch where his train entered the main track. When this switch was closed, the conductor could retrieve this copy and verify that the train had authority to enter the main track. If the train was already on the main track, this was not done.

There is considerable discussion of this point in the report, although it was wonderfully irrelevant to the case at hand. Although it may be of some help for both ends of the train to know block conditions, it is always primarily the engineman's responsibility. If a conductor sees his engineman pass a Block Occupied signal, he should, of course, pull the air, but the main thing is that entering a block does not require memory, but response to an immediate situation. With automatic block signals, they change as soon as they are passed, so the rear end cannot usually see them. I have never seen a passenger conductor worry the least bit about what block signals say.

It is different in the case of train orders. They may require some delayed action, carried out at some distant point, which must be recalled later from memory. Reliance on one memory is very dangerous in this case, and the train order rules explicity provide for safety by making the conductor and engineman jointly responsible. That is, the conductor is as concerned about meeting points as the engineman. On roads without the block system, this was often enforced by prescribing that the engineman give a distinctive whistle signal before approaching a meet, and for the conductor to respond with a hand or communicating signal. This was not done on the Pennsylvania, the block system taking its place in catching lapses.

What went wrong at SI? The first thing one notices is the deplorable practice of the conductor, in handing a copy of the order to the engineman, of stating its contents. So the engineman may just roll up the orders and stick them in his overall bib with at the most a glance. No problem here, however. The engineman repeated the intent of the order correctly, and said in evidence that he understood. He had no form K, of course, but he usually got one, it must be assumed, and never looked at it, either. Habit is of great influence in railroading.

As at all block-limit stations, SI had a block-limit signal consisting of red and yellow lights and targets. Unless you happen to have a Form K for that station, this signal means STOP. I, personally, would always glance at the order clipboard and verify that I had one before I even saw the board in the distance. From the engineman's evidence, I would assume that somebody had tried to make this very clear to him. There was a rule on the Pennsylvania that employees in the cab would call out signals to each other, and, strange as it may seem, this seems to have been done (at least, always in accident evidence). This engineman said that he always called out to himself the block permission at block-limit stations. Of course, he was alone in the motor cab. We can be pretty sure that he didn't carry out this goofy response to the block-limit board, either on July 31 or any other day.

What the engineman said was that he became concerned about the delay to his train, which would make it late for a B&O connection in Akron, and the next thing he knew he was whistling for the road crossings in Silver Lake with the big twin air tanks of an I1s looming a few hundred feet ahead. He hit the cinders. This is not an unknown human condition--he simply let his mind wander, for all we know. There is no need for a reason, although the examiner had a theory about carbon monoxide poisoning that he favored. On this trip, the engineman was in the rear vestibule, since the motor was running backwards to avoid turning in Hudson. Maybe he got gassed on the way up to Hudson when he was in the engine compartment. I have actually travelled in the motor end of a motor car on several occasions, and was never impressed by any carbon monoxide. On the contrary, there was always lots of noise and fresh air.

The conductor was doing his accounts as the motor approached Silver Lake, and didn't notice anything wrong until SI went by at 50 mph. Then he quickly left the smoker (then at the rear of the passenger compartment) and baled out of the baggage door. A deadhead track laborer in the baggage compartment knew nothing about train orders, meets and blocks, but seeing the conductor bale out made him copy the action at once. The baggageman, who was also the flagman, must have been astonished to see this precipitate action as the brakes set in emergency, but hesitated a moment too long, and became the only employee to lose his life in the accident. The engineman, conductor and track laborer, skinned up on the ballast, saw their late train passing them in the reverse direction in flames. The track laborer was an innocent bystander, but the others were accomplices in the disaster. Maybe the conductor pulled the air as he left; the brakes were said to be in emergency, but the engineman just let his hand off the brake lever, and I don't think this gives an immediate emergency application.

We conclude that the train order was forgotten by one man, the one who had to take action, and was ignored by the others, who did not. However much the joint responsibility of engineman and conductor was taught, the practice seems to be that the engineman runs the train. This is the single fact that makes systems like automatic train control desirable. The block system was completely ineffective here, because of the habit of running by a block-limit station with a Form K, and not connecting the two facts at the proper time. This is a matter of education, and of instilling proper procedure. I wonder how many trainmasters ever examined this by having a Form K issued with the wrong name on it. For example, at Arlington, just hand up a Form K to a passenger train for the block at IS. The Form K is the authority to pass a block-limit sign, memory is not. The only exception is when the block-limit sign is immediately passed after telephone communication, which is completely reasonable.

Many companies provided for an exchange of signals between conductor and engineman approaching meeting and waiting points, specified in Rule S-90. After the engineman sounds the station whistle (one long sound), the conductor on a passenger train gives a single sound on the communication signal, which is answered by the engineman with two long sounds and a short sound (- - o). On other trains, the engineman gives the meeting point whistle without being prompted. The Pennsylvania does not prescribe such a meeting signal, relying on the block system to protect against forgetfulness. However, as we see here, exactly the same man that is in danger of forgetting is also the man that provides the block protection! If the engineman forgets one, he will almost certainly forget the other, as happened in this case.

By 1949, another warning had been added to the block-limit system. An "approach block limit" sign with reflectorized letters "ABL" was placed at a suitable distance on either side of the block-limit sign. If a train did not hold authority to proceed as if Clear Block was displayed, it had to reduce speed immediately to not exceeding medium speed, and approach the block limit sign prepared to stop. This meant that if the engineman did not show his intention of stopping by making a brake application at this point, the conductor could take steps to stop the train. It would seem to be implied by this that the conductor should be aware of block status in the same way as the engineman.

Another evil practice is mentioned in the testimony, though irrelevant to the case. It seems that when the pad of Form K's in a phone booth ran out, any old scrap of paper was used, or even the back of a timetable. This deplorable practice shows careless management. The signalman should be responsible for supplies at the block-limit stations he controls (one of the signalmen at a continuous station should be made responsible), just as he is at his own office, unless some other arrangement is made. This job probably fell between the cracks.

The report said that the motor car had a 250-gallon fuel tank, 20" x 6" x 7'11" near the middle of the car under the floor. A tank of these dimensions would have less than 50 gal capacity, so something is wrong. The 6" should probably be 2'6". About 150 gal of gasoline was supposed to have remained, which was responsible for the deadly fire. Though of 1/8" steel, the tank would have been ruptured by the 500-foot scraping and collision with the displaced motor truck with its heavy traction motors. There was a similar collision of a motor car followed by fire 26 years previously, on the Missouri and North Arkansas, at Tipton Ford, with a similarly high toll, though the number of passengers was greater in this accident, and many escaped. Although gasoline motor cars were widely used, these two seem to be the only catastrophic accidents in which the gasoline played a role. The gasoline engines were largely replaced by diesel engines, where the fuel is much less hazardous because of its lesser volatility. Nevertheless, liquid fuel of any kind is a definite hazard, even in modern equipment, as was seen at Ladbroke Grove in England in 1999, where fire from diesel fuel was responsible for many casualties.

The investigations themselves are of considerable interest. It must be realized that the different parties involved had different motives and circumstances. Most exposed were the employees involved, since the report could not help but lay blame, which would have an important effect on any legal proceedings or on the job security of the employee. The railway company made its own private, internal investigation, of course, and had to deal with the labor organizations working in the employee's interest. It is remarkable that neither the railway company nor the unions figure in the reports of these ICC investigations (though they may have attended the sessions).

The employee always attempts to deflect the blame, while as far as possible not accusing his fellows. This usually means that signals or brakes failed, but mysteriously cured themselves afterwards. Signals and brakes are actually very reliable, and never the causes of accidents unless thwarted by some signal maintainer's bridging a protective relay, or some brakeman's leaving an angle cock closed, or someone's not making a running brake test.

The ICC investigators were probably men of integrity, but do not appear in general to be familiar with practical railway operation, and could arrive at surprising conclusions, often straying from the precise matter at hand, and obscuring or improperly weighting the causes of accident. In the present investigation, the possibility of carbon monoxide poisoning is adduced without any supporting evidence. Also, the question of whether the conductor as well as the engineman should be aware of block conditions is raked back and forth inconclusively. The report does state the proper conclusion, but gives no hint as to possible causes and remedies, except from the usual recommendation that automatic train control be installed.

An example of this from a few years later is the accident at Dewey, Indiana on the Chicago and Eastern Illinois on 14 September 1944, at 2 a.m. on a foggy morning. The causes of this accident were obscure, and were not discovered by the investigation. A train ran by a meeting point and adverse signals before colliding with an opposing train, the brakes being applied only at the last moment. The engine crew did not live to explain their actions. The conductor was unaware of the meet--apparently his copies of the orders were taken by the front brakeman, who did not know where on the road he was. The examiner concentrated on the fact that the train that was struck had advanced past a signal at stop, after first stopping, according to rule within yard limits, and this reduced the stopping distance of an opposing train. However, the train had only advanced six feet, which had no significance at all, and the opposing train had ample stopping distance had it obeyed its signals. He recommended, as usual, that the company install automatic train stop, train control or cab signals. In addition to whatever was wrong with the engine crew, the handling of train orders and the conductor's response were contributory. If they had been proper, there would have been no accident. In my opinion, there was nothing wrong with the signals or rules at all, or they would have been responsible for other accidents, of which there were none.

Return to Railway Index

Composed by J. B. Calvert
Created 13 February 2002
Last revised 16 June 2002