1954 NKP Head-On Collision
Dunkirk, New York

Nickel Plate Road Magazine


Peter K. Shepherd

Nineteen fifty-four was not the best of years for the Nickel Plate. The end of the Korean War had brought a business recession, as NKP management had foreseen. Until November, each month produced less favorable operating and financial results than the corresponding month of 1953; only November and December were improvements on the year previous. This fourth quarter could, in another sense, have been said to have come in with a bang and gone out with a bang.

The “bangs,” unfortunately, were the sounds of collisions, one as the quarter began and the other as it ended. They were on geographically opposite ends of the Nickel Plate. One involved brand-new diesels [at Gibson City, Illinois], the other NKP's best steam power [at Dunkirk, New York]. Thankfully, neither involved fatalities, but one injured a considerable number of Nickel Platers, and both caused considerable property and equipment damage.

Fall turned to winter. The I.C.C. accident report on the Gibson City derailment was completed on December 2, and blamed the event on an open switch. Twenty-six days later, the I.C.C. had another Nickel Plate misfortune to investigate, a head-on collision on the Buffalo Division at Plate interlocking, near Dunkirk.

The Conneaut dispatcher wasn't having an easy time of it early in the morning of December 28, 1954. It was raining, which was a break from the snow one could usually expect. An eastbound freight had been delayed east of Brocton due to a defective car, thus tying up the eastbound main track. East of Brocton, however, was the Nickel Plate-Pennsylvania paired track, which was operated as a double-track railroad. Thus, the dispatcher routed eastbound trains on the westbound main track against the current of traffic. A PRR freight was so routed between BM interlocking (Brocton) and Plate, where it was routed to number 3 track. (Number 3 track, Nickel Plate-owned, ran between the eastward and westward mains from Plate to Silver Creek and was often used to allow a train in either direction to overtake another without affecting traffic in the opposite direction).

As the Pennsylvania train occupied the westward main track, Extra 776 West received a stop signal at signal R-6, just east of the Erie crossing at Plate. The signal did not clear as the PRR train completed its move to track 3, as the dispatcher intended that eastbound Nickel Plate train 90 would follow the same route.

The interlocking at Plate was controlled by the operator at AK tower, one mile west. An eastbound train on the westbound track, such as 90, would encounter signal LA-6, 659 feet west of the switch which would divert a train to track 3. A yellow aspect on signal LA-6 indicated "proceed at restricted speed" and was intended to authorize movement through the interlocking only; it would display this aspect regardless of the route lined through the plant. To set up a route through the plant the operator had to move a switch lever to the desired position and depress a button, which transmitted a code, moving the switch. The same procedure was followed to display a signal.

The interlocking was equipped with time, route, and indication locking. This feature prevented an operator from changing a route once the train for which it had been lined had entered the interlocking. Similarly, signals on a properly lined route could not be changed once a train had entered the interlocking.

Ninety was a rare but by no means unknown eastbound doubleheader, consisting of S-3 770, S-2 740, 65 cars and 3 cabooses. Train order number 5, authorizing 90 to move on the westward main track to Plate and then on to track 3 to Silver Creek, was delivered to 90 at BM.

Ninety entered the plant without incident and passed signal LA-6, which displayed a yellow aspect. When the 770 entered switch 23, which controlled entry to track 3, the engineer realized that he was not entering that track. He immediately applied his brakes in emergency and closed the throttle, but could not stop in the 450 feet between switch 23 and Extra 776 West.

The extra's engine and tender were shoved backwards some 200 feet. The consist, ironically, was a wreck train, but this earned the extra no special treatment. The crane, immediately behind the tender, was uncoupled and stopped 100 feet east of the tender. The third of three tool cars was destroyed. Both of 90's engines derailed, as did the lead truck of the second car and the seventh through the fourteenth cars. Four of these cars were demolished, and three other cars were destroyed by a fire which broke out shortly afterward. Three other cars were damaged.

Twelve crew members, six from each train, and twelve members of the wreck train crew, were injured. Fortunately, there were no fatalities.

A highly unusual and complex set of circumstances was discovered as a result of the investigation and was detailed in the I.C.C. report

As we have seen, switch 23 was controlled from AK tower, a mile from the switch. To operate the switch, the switch lever was moved to the desired position and a code button was depressed, transmitting the code that operated the switch. The operator said that he had followed this procedure soon after 90 had left BM tower. Immediately thereafter, he copied a train order, which required that he turn his back to the control machine. He then followed a procedure similar to the switch operation to cause signal LA-6 to show a "proceed at restricted speed" aspect, and the panel indicated that LA-6 displayed the desired indication. The operator became involved with other duties and did not check the board again until after the collision. Among these duties was the delivery of a train order to 90, and the inspection of another eastbound on the eastward main (Although not specified in the report, this may have been the train with the defective car). After returning to the tower, he did observe the visual indicators showing switch 23 lined for the westward main while the lever was lined for the intended movement to track 3. The operator thought this discrepancy was due either to a defect in the plant or because the switch lever had not been fully moved to the desired position before the code button had been depressed.

The switch control apparatus was arranged so that if the lever was not moved fully into the desired position, the code transmission would have no effect on the switch. The operator stated that after the accident, he relined the switch to its normal position, permitting continuous movement on the westward main. A signal maintainer who inspected the equipment verified that the last code transmitted to the switch motor would have restored it to this position.

Apparently the switch had been lined for the westward main after the Pennsylvania train had crossed from the westward main to track 3. Signal LA-6 was caused to show "proceed at restricted speed" even though the switch was not properly lined for the intended movement.

The head-end crew of 90 didn't escape its share of responsibility. Signal LA-6 was intended only to give authority to enter the interlocking. The "restricted speed" it required should have been slow enough to permit a train to stop short of another train, obstruction, broken rail or a mis-aligned switch. Obviously, even though 90 was proceeding at or under 15 M. P. H., it couldn't stop short of switch 23 or the waiting wreck train. Thus 90's crew was held partly to blame.

Magazine Article Photos
Nickel Plate Road Berkshires 776 and 770 both show damage after collision at Dunkirk, New York on December 28, 1954
Both Photos by R. Grainger
The 740, the 3rd engine involved in the accident, on a better day near Avery, Ohio
Photo by Don Wood

Three very unhappy Berkshires made an unscheduled trip to Conneaut for repairs. The 740, second engine of the doubleheader, entered the shops on January 3 - upon its emergence on the 17th, she sported a new bell and frame, Mars light, headlight, numberboards, pilot, air pump shield and spindle guide bushings. Ninety's lead engine, 770, required considerably more front-end work, including a coupler, pocket and lift pins, smokebox front, steps, drawbar, front side rods and both air pump shields. She was released on February 10. Sister 776 suffered even more damage - among her requirements were a new cab side, tank deck, engine truck frame, one cylinder head and pilot as well as the same front-end parts that were replaced on 740 and 770. At the same time 770 and 776 exchanged tenders. The work on 776 was completed on February 23.

(The NKP Berkshires could count themselves lucky. Steam still had a place on Nickel Plate, and management would spend whatever was necessary to put the three engines back in service. A year earlier, similarly-aged Louisville and Nashville M-1 Berkshire 1981 tangled with a rock slide, causing $36,000 damage. The L & N Berk never ran again.)

Three days later, the fourth quarter of 1954 was over. Doubtless no Nickel Platers hoped for a quiet 1955 more fervently than the wreck crews.

Special thanks go to Timothy Adang, who made available records kept by his grandfather, A.H. Adang, on shop work done at Conneaut in the 1950's. These records detail the repairs on 740, 770 and 776.


  • Shepherd, Peter K. 1994. This is an extract of an article that covered two Nickel Plate train wrecks. Only the Dunkirk wreck is addressed in this web page. The article appeared in the Winter 1994 edition of the Nickel Plate Road Magazine, Volume XXVIII, No. 1, on pp 5 - 9. The Nickel Plate Road Magazine  is published by the Nickel Plate Road Historical and Technical Society (NKPHTS) who has kindly given permission for use of this material. This uniformly, high-quality magazine is recommended to anyone interested in the Nickel Plate - the former New York, Chicago, & St. Louis Railroad and its predecessor. Anyone interested in joining the NKPHTS should contact them at:

    NKPHTS, Inc.
    P.O. Box 381
    New Haven, Indiana, 46774-0381