1995 Russell Hill subway accident

Signal lightbox at a switch of the Toronto subway

At the time of the accident, the TTC subway used conventional wayside signals without cab signalling. Thus drivers were expected to remember each signal aspect and act accordingly until they receive new information from the following signal. As of 2019, the TTC has converted some sections of the line, including the Russell Hill portion, to more modern automatic train control.[2] The following description is limited to aspects relevant to the accident.

Some signals are "automatic" and have a single head that shows one of the three basic colours. Red means stop and stay (until presented with a less restrictive signal aspect), yellow means that the next signal is red and the driver should prepare to stop, and green means that the next signal is not red and it is safe to proceed at normal speed.

"Interlocking" or "home" signals have a double head and show two coloured lights. The top light has the same meaning as that of an automatic signal. The second light is red if the top light is red; otherwise it shows yellow if the switches are set for a secondary route, and green for the main route or if no relevant switches exist. Thus the combination yellow over green means the same as a single yellow.

At locations such as downgrades, curves, and switches, the same signals additionally serve to enforce speed limits. A system called "grade timing" regulates speeds, and is indicated by an additional "lunar white" light below the one or two coloured lights previously mentioned. The white light means that the next signal is red (or double red) only because of grade timing, and if the train approaches at the proper speed, the signal will clear to a less restrictive aspect before the train gets there. If the train goes too fast, it will pass the red signal and be stopped by the train stop.

In 1995 the second signal did not provide any confirmation that it was about to change and the operator was expected to remember the previous signal aspect correctly. With the tolerant attitudes of management, it had become common practice for operators to travel at the fastest speed that the system allowed, passing the signals at the same instant they were clearing.

A long downgrade would therefore have several consecutive grade-timing sections. If all the signals were automatic and no trains were present, the first signal would be yellow over white, followed by a series of red over whites, and finally a red. When the train approaches each red over white signal, it would clear to yellow over white and the train stop would be lowered; and the final red signal would clear to green. Interlocking signals can also take part in grade timing, with aspects such as yellow over green over white (meaning the same as yellow over white).

Looking south from St. Clair West station. Signal SP77/X38 can be seen showing red over red.

From St. Clair West to Dupont the subway runs continuously downhill and grade timing is used. The signal at the end of St. Clair West station, SP77/X38, is an interlocking signal and therefore shows yellow over green over white whenever there is nothing to delay a train. As Jeffrey's train passed the signal, however, it was showing yellow over green. He should have stopped at the following signal, SP71, which was red because another train was occupying the block ahead.

After the accident Jeffrey said he remembered SP77/X38 as yellow over white (actually an impossible aspect for that signal) therefore he must have either "misremembered" it at the time, or else have been fooled by some reflection off the signal head into thinking he had seen the usual yellow over green over white. In any case, he believed that grade timing was in use and so approached signal SP71 at the same speed as he would normally drive on this section.

After this Jeffrey did not remember what he had seen, but it is easy to reconstruct the events. When the train passed SP71 without him seeing it change, he must have assumed that he had passed it just as it cleared. He continued coasting downhill, still thinking that he was under grade timing and violated the signal, without realizing that if this was true, the signal would have shown red over white as he approached.

He must therefore have expected the third signal, (the SP65) to clear in the same way. Again it was showing red, not red over white, but Jeffrey repeated his mistake and violated the signal. He braked only when he actually saw the train just ahead (it was stopped at the next signal, SP53); by then, a collision was unavoidable.

News media accounts of the accident often say that Jeffrey "passed three red lights." This count includes the repeater signal provided in advance of SP65 because a curve in the tunnel interferes with sighting. Drivers are not expected to stop at the repeater even if it is red.

The "trip" or "train stop" consists of a T-shaped arm placed just outside the right-hand rail near each signal (except a repeater). If the signal is red (or double red), with or without a lunar white, the trip arm rises enough for its crossbar to strike a "trip valve" or "tripcock" lever on the bottom of a train's first car as it passes; and by doing so triggers the emergency brakes of the train.

On the relevant class of trains, the trip valve is located near the centre of the right side of the leading bogie, where its position in relation to the rail is well controlled.

Wismer and Becker, Inc., the contractors who installed the signal system on the Spadina line portion of the subway, chose to use an Ericsson train stop that was designed to have the trip arm located in the middle of the track. A subcontractor, Gayle Manufacturing, designed a mechanical extension that would allow the device to drive a trip arm in its actual position outside of the rail.

What was not realized was that the mechanism they designed (when the trip arm was raised) did not have the clearance specified for the train and therefore would not have the desired effect of activating the train's emergency braking system.

When both the trip arm and trip valve were new, this did not matter; the train did not fill its entire clearance envelope, and the trip arm would contact the trip valve normally. However, with normal wear of the wheels and rails, and with the train rounding a curve at speed (and therefore riding off-centre) it was another matter. The first wheel on the right-hand side of the train hit a bolt in the mechanism attached to the trip arm and knocked the arm down. It would then rise again, but too late: with the train moving quickly, the trip valve, just a few feet behind the wheel, would already have gone past,[1] thus making the train stop useless for ensuring signal compliance.