The 08:44 passenger service from Glasgow Queen Street to Dundee, despite running late and experiencing technical difficulties, left Invergowrie station without incident. However, the brake on the leading bogie of locomotive25083 was binding, although the driver carried on as Dundee was only a few miles away. As the train was running along Invergowrie Bay a traction motor caught fire and the train was stopped. Approximately ten minutes later, the stationary train was run into at around by the 7 coach 09:35 express from Glasgow to Aberdeen hauled by locomotive 47208. The impact threw the last four coaches of the Dundee train over the sea wall. The last 2 broke away completely and ended up in the Firth of Tay; fortunately, the tide was out. The class 47 loco was subsequently scrapped due to damage. Both passengers in the rear carriage and the driver and secondman of the Aberdeen train were killed instantly. A further passenger died later and a total of 51 people were injured.
Investigation
The signalman at Longforgansignal box stated that he put the mechanical starting signal correctly back to Danger behind the Dundee train. Around ten minutes later, the Aberdeen train arrived at his box and drew up to the Home signal, which was then cleared for it. The train continued to draw down to the Starting signal but, after a few moments, began to accelerate. It passed the Starting signal which, as far as the signalman could see, was still at Danger. He went down onto the track and saw that the arm of the Starting signal was slightly raised; about 4°. Subsequent investigations showed that it was possible for the arm to have been raised roughly 8°. The guard of the Aberdeen train said that he had looked out of the window of the rear coach at Longforgan as the train picked up speed. He saw the starting signal giving "a poor off" but assumed that it had already been put back to Danger after the locomotive had passed it and perhaps had not quite returned to the horizontal position. It is not clear if the guard could have seen the starting signal exactly as the driver would have seen it. The subsequent public inquiry found that the guard was not to blame. Why the driver passed the signal remained a mystery. The inquiry speculated that he may have been looking back towards the signal box, or checking that the train was clear of the level crossing. As he then looked up towards the signal he might have concluded that it had moved since he had last seen it and that it had, therefore, been cleared by the signalman. Various operating staff who saw the signal before and after the accident also gave evidence that the arm was not properly horizontal, including some who said that the degree of elevationappeared to increase as they got closer to it. It was later found that the signal post bracket was badly bent. The bracket may have been struck by a chain hanging from a wagon, or perhaps by engineers' machinery working on the lineside. In addition, the signalling at Longforgan was basic and lacking in many safety features. The Starting signal had no AWS that would have warned the driver of the Aberdeen train, nor was there an adjuster for the pull wire. There was also no repeater in the signal box, nor was a detonatorplacer provided.
