A word in time
The Southall Disaster 19 September 1997
A word in time
by Ray State
Continuing the saga of the Southall Accident
"Accidents are rarely caused by a single event but by a combination of a number of events each one of a possibly minor nature which act cumulatively"
As the Southall Inquiry grinds on, information about the events in the days leading up to the accident are beginning to emerge. What is evident is that the dissemination of information about the condition of the HST trainset was less than perfect. The timely notification of crucial defects would have triggered warnings in the minds of those who could have taken action at several points along the route. Like the Titanic (see later) a word in time could have been all that was needed to prevent the accident.
During the evening of the 18th September 1997 the 19.47 train from Oxford to Paddington was formed from a HST trainset which included power car 43173.
As the train drew into Oxford the AWS horn sounded and could not be cancelled by the driver by depressing the rest button. The reset button is a very tough switch having been designed for locomotive operation. The contacts are silver and have a rolling/wiping action which is designed to wipe away any dirt so that contact is always achieved. When AWS was fitted to existing stock this button sat on the surface of the driving desk in a grapefruit size domed case with a push button which could be driven down by the flat of the hand. However, when the reset button was incorporated in new designs the button was flush mounted below the desk surface with just the push button showing. Whilst this was tidier and aesthetically pleasing it did assist in liquids, dirt and other debris migrating down the push button sleeve. In particular liquids like tea or coffee containing sugar were lethal. The liquid would evaporate leaving the sugar as a sticky residue to which dust or powder would adhere. The contact wiping action had extreme difficulty in removing such deposits. The HST design was flush mounted.
What fouled the contacts on 43173 has yet to be declared but whatever it was it prevented the AWS from being reset.
In order to move the train the driver isolated the AWS. The driver believed that the GWT control at Swindon was somehow notified of the fact that the AWS was isolated. How this was done is not stated. What he did do was to write "AWS isolated. Unable to cancel" in the Driver's Repair Book in the cab. These log books are carried in all driving cabs and are intended to be communication from the footplate to the maintenance staff of problems.
He then drove the train to Paddington. The set was due maintenance that night and was driven to Old Oak Common. The maintenance was an "A" examination which did not involve work on the AWS. However, part of the examination is to read the Drivers Repair Book and carry out any repairs. For 43173 now at the back of the set, this was done. Finding a complaint about the AWS the fitter de-isolated it and gave it a full test. Unfortunately, it worked perfectly. This is not unusual. Changes in atmosphere between inside and outside can be just enough to give a contact. What the depot did not do was record the fact that there had been a fault on the RAVERS computer system. This system is Mainframe based and will within seconds disseminate a fault to all the people who have a need to know. For whatever reasons the depot did not see a need to record that the complaint had been investigated and "no fault found". The fitter recorded the test and "no fault found" in the Drivers Repair Book. At this point, if the book is full it is removed and taken to the foreman for storage and replaced by a new one. The book on 43173 was full or near full but it was not replaced.
Thus as the 19th September 1997 dawned an apparently serviceable HST stood waiting for its morning duties. At 0550 the early shift driver took the empty stock into Paddington with 43173 leading towards Paddington. However a problem developed with the driver guards buzzer which was sounding continuously. On arriving at Paddington, he walked to the rear power car 43163 and tried the buzzer. It would not work. Returning to 43173 he placed his key in the desk and found the AWS fault had re-appeared. The horn sounded but would not cancel. The driver isolated the AWS in 43173.
The driver then went to the supervisors office and reported the problems. Attempts were made to find the station fitters to no avail. The supervisor then spoke to someone on the telephone which the driver assumed to be Swindon Control. There is no record of any call in the Swindon Control. However, by 0735 they were aware of the defective buzzer. The timing of this record puts it just after Reading and it is assumed it was telephoned in from there. The driver was driving from 43163 which had an operational AWS. Reading station staff would be aware of the buzzer fault because of the fact they would have to have given the driver the start signal by hand but no-one on Reading would be aware that the AWS was isolated in the rear power car.
On arrival at Swansea the driver changed ends. He was now in power car 43173. He de-isolated the AWS and found that it still wouldn't cancel. No fitters had boarded the train en-route and none were waiting for the train at Swansea. On the driver enquiring about this from Swindon Control they told him that nothing was known about the request. The Station Manager at this point arrived to find out what the problem was. Neither the Manager or the driver can recall what they talked about but fitters were called from Landore depot.
Two fitters arrived to look at the power car. At this point there is confusion as to who told what to whom. What is known is that the buzzer was silenced by cutting the wire to the sounder. Nothing was done about the AWS. Records in Landore only show only the work on the buzzer. RAVERS computer system shows defect 972620363 19/9/97 D/Guard Buzzer Sounding Continually (Disconnected) . The fact that the word "Disconnected" appears would tend to indicate that this was entered after the visit of the Landore fitters. There is no mention of the AWS isolated in the depot records or the RAVERS system. Swindon Control vaguely remember something being isolated but what ever it was, was written on a piece of paper that got thrown away. In Swindon something being isolated rang no warning bells.
The driver tried to enter the AWS fault in the Drivers Repair Book but it was full. He wrote a note which said "AWS isolated, repair book full" and attached to the cab clipboard. In accordance with the rule book the signalmen en route should be made aware of the inoperability of the AWS. This should be done by the driver who should tell the nearest signalman who in turn, notifies Railtrack Control. It is without doubt that Railtrack Control would not have let the train depart had they known. However, they did not know as the driver took no action assuming that Swindon Control had told the signalman at Swansea.
The train could have been turned at Swansea. This option does not seem to have occurred to the those on site at Swansea. In reality this would have taken time. Had Swansea been aware of the need then the turning could have been carried out in time but time was lost in bringing the fitters down to the station which meant that if turned now the train would be late departing. In addition, when the train reached Paddington again 43173 would be the lead power car and as there were no turning facilities at Paddington the next journey leg would be lost. These aspect must have occurred to those on the ground.
The driver took the train to Cardiff and handed over to Larry Harrison. He was told the AWS was isolated, that the buzzer was disconnected and that the Drivers Repair Book was full hence the note attached to the clipboard.
The rest is history.
In this discourse what is incredible is that the key fact that the AWS was isolated was never told to those who could have made a decision to stop the train. If the Swindon Control was told they either did not understand the seriousness or the facts were forgotten. The buzzer fault on the other hand was disseminated because its immediate affect on train operation was more acute. The action of the maintenance staff at Landore is even more inexplicable. They duly recorded the details about the buzzer disconnection but failed to report anything about the AWS.
It is a sad fact that had any one of these dozen or so people taken time to say just two or three words to the right person then it would have triggered the warnings in those who could have taken action.
Titanic sinks - most passengers saved
This file last updated: Tuesday, 09-Nov-1999 06:00:43 EST