Tay Bridge disaster

Bridge collapse and train wreck
William McGonagall's poem on this subject

The Tay Bridge disaster occurred during a violent storm on Sunday 28 December 1879, when the first Tay Rail Bridge collapsed as a North British Railway (NBR) passenger train on the Edinburgh to Aberdeen Line from Burntisland bound for its final destination of Dundee passed over it, killing everybody on board. The bridge—designed by Sir Thomas Bouch—used lattice girders supported by iron piers, with cast iron columns and wrought iron cross bracing. The piers were narrower and their cross bracing was less extensive and robust than on previous similar designs by Bouch. Bouch had sought expert advice on wind loading when designing a proposed rail bridge over the Firth of Forth; as a result of that advice he had made no explicit allowance for wind loading in the design of the Tay Bridge. There were other flaws in detailed design, in maintenance, and in quality control of castings, all of which were, at least in part, Bouch's responsibility. Bouch died less than a year after the disaster, his reputation ruined. Future British bridge designs had to allow for wind loadings of up to Bouch's design for the Forth Bridge was not used. As of 2024, it remains the fifth deadliest railway accident in the history of the United Kingdom, as well as the second deadliest rail accident in Scottish history, being surpassed by the UK's deadliest: the Quintinshill rail disaster.

Construction of the original Tay Rail Bridge began in 1871. In its initial design, the bridge was to be supported by brick piers resting on bedrock. Trial borings had shown the bedrock to lie at no great depth under the river. At either end of the bridge, the bridge girders were deck trusses, the tops of which were level with the pier tops, with the single track railway running on top. However, in the centre section of the bridge (the "high girders") the bridge girders ran as through trusses above the pier tops (with the railway inside them) in order to give the required clearance to allow passage of sailing ships to Perth. The bedrock lay much deeper than the trial borings had shown, and the bridge's designer, Sir Thomas Bouch, redesigned the span with fewer piers and correspondingly longer girders. The pier foundations were now constructed by sinking brick lined wrought iron caissons onto the riverbed, and filling these with concrete. To reduce the weight these had to support, Bouch used open lattice iron skeleton piers; each pier had multiple cast iron columns taking the weight of the bridging girders. Wrought iron horizontal braces and diagonal tiebars linked the columns in each pier to provide rigidity and stability. The basic concept was well known, but for the Tay Rail Bridge, the pier dimensions were constrained by the caisson. For the higher portion of the bridge, there were thirteen girder spans. In order to accommodate thermal expansion, at only three of their fourteen piers was there a fixed connection from the pier to the girders. There were therefore three divisions of linked high girder spans, the spans in each division being structurally connected to each other, but not to neighbouring spans in other divisions. The southern and central divisions were nearly level, but the northern division descended towards Dundee at gradients of up to 1 in 73. The bridge was built by Hopkin Gilkes and Company (Gilkes), a Middlesbrough company which had worked previously with Bouch on iron viaducts. Gilkes, having first intended to produce all ironwork on Teesside, used a foundry at Wormit to produce the cast iron components, and to carry out limited post casting machining. Gilkes were in some financial difficulty; they ceased trading in 1880, but had begun liquidation in May 1879, before the disaster. Bouch's brother had been a director of Gilkes, and all three had been colleagues on the Stockton and Darlington thirty years previously; on Edgar Gilkes's death in January 1876, Bouch had inherited shares valued at £35,000, but also owed for a guarantee of £100,000 of Gilkes borrowings and had been unable to extricate himself. The change in design increased cost and necessitated delay, intensified after two of the high girders fell when being lifted into place in February 1877. The first engine crossed the bridge seven months later. A Board of Trade inspection was conducted over three days of good weather in February 1878; the bridge was passed for use by passenger traffic, subject to a speed limit. The inspection report noted:

When again visiting the spot I should wish, if possible, to have an opportunity of observing the effects of high wind when a train of carriages is running over the bridge. The bridge was opened for passenger services on 1 June 1878. Bouch was knighted in June 1879 soon after Queen Victoria had used the bridge.

On the evening of Sunday 28 December 1879, a violent storm (10 to 12 on the Beaufort scale) was blowing virtually at right angles to the bridge. Witnesses said the storm was as bad as any they had seen in the 20–30 years they had lived in the area; one called it a 'hurricane', as bad as a typhoon he had experienced in the China Sea. The wind speed was measured at Glasgow – (averaged over an hour) – and Aberdeen, but not at Dundee. Higher windspeeds were recorded over shorter intervals, but at the inquiry an expert witness warned of their unreliability and declined to estimate conditions at Dundee from readings taken elsewhere. One modern interpretation of available information suggests winds were gusting to
Use of the Tay Rail Bridge was restricted to one train at a time by a signalling block system using a baton as a token. At 7:13 p. m. a North British Railway (NBR) passenger train from Burntisland (consisting of a Class 224 locomotive, its tender, five passenger carriages, and a luggage van During the inquiry, testimony was heard that the wind was pushing the wheel flanges into contact with the running rail. John Black, a passenger on the previous train that crossed the bridge, explained that the guard rails protecting against derailment were slightly higher than and inboard of the running rails. At that point "there was a sudden bright flash of light, and in an instant there was total darkness, the tail lamps of the train, the sparks and the flash of light all disappearing at the same instant." Not only was the train in the river, but so were the high girders and much of the ironwork of their supporting piers. Divers exploring the wreckage later found the train still within the girders, with the engine in the fifth span of the southern 5 span division. There were no survivors; only 46 bodies were recovered out of 59 known victims. Fifty six tickets for Dundee had been collected from passengers on the train before crossing the bridge; allowing for season ticket holders, tickets for other destinations, and for railway employees, 74 or 75 people were believed to have been on the train. It has been suggested that there were no unknown victims and that the higher figure of 75 arises from double counting in an early newspaper report in the Dundee Courier, but the inquiry did not take its casualty figures from the press; it took sworn evidence and did its own sums.

A court of inquiry (a judicial enquiry under Section 7 of the Regulation of Railways Act 1871 "into the causes of, and circumstances attending" the accident) was immediately set up: Henry Cadogan Rothery, Commissioner of Wrecks, presided; supported by Inspector of Railways William Yolland and William Henry Barlow, President of the Institution of Civil Engineers. By 3 January 1880, they were taking evidence in Dundee; they then appointed Henry Law, a qualified civil engineer, to undertake detailed investigations. Whilst awaiting his report they held further hearings in Dundee (26 February – 3 March); having got it they sat at Westminster (19 April – 8 May) to consider the engineering aspects of the collapse. By then the railway, the bridge's contractor and Bouch had separate legal representation, and the NBR had sought independent advice from James Brunlees and John Cochrane, both engineers with extensive experience of major cast iron structures. The terms of reference did not specify the underlying purpose of the inquiry – to prevent a repetition, to allocate blame, to apportion liability or culpability, or to establish what precisely had happened. This led to difficulties (culminating in clashes) during the Westminster sessions. When the court reported their findings at the end of June, there was both an Inquiry Report signed by Barlow and Yolland and a minority report by Rothery.

Two witnesses, viewing the high girders from the north almost end on, had seen the lights of the train as far as the 3rd–4th high girder, when they disappeared; this was followed by three flashes from the high girders north of the train. One witness said these advanced to the north end of the high girders with about fifteen seconds between first and last; the other that they were all at the north end, with less time between. A third witness had seen "a mass of fire fall from the bridge" at the north end of the high girders. A fourth said he had seen a girder fall into the river at the north end of the high girders, then a light had briefly appeared in the southern high girders, disappearing when another girder fell; he made no mention of fire or flashes. 'Ex Provost' Robertson had a good view of most of the bridge from his house in Newport on Tay, but other buildings blocked his view of the southern high girders. He had seen the train move onto the bridge; then in the northern high girders, before the train could have reached them, he saw "two columns of spray illuminated with the light, first one flash and then another" and could no longer see the lights on the bridge; the only inference he could draw was that the lit columns of spray – slanting from north to south at about 75 degrees – were areas of spray lit up by the bridge lights as it turned over.

Ex provost Robertson had bought a season ticket between Dundee and Newport at the start of November, and became concerned about the speed of north bound local trains through the high girders, which had been causing perceptible vibration, both vertical and lateral. After complaining on three occasions to the stationmaster at Dundee, with no effect on train speed, after mid December he had used his season ticket to travel south only, using the ferry for north bound crossings. Robertson had timed the train with his pocket watch, and to give the railway the benefit of the doubt he had rounded up to the nearest five seconds. The measured time through the girders (3149 feet (960 m)) was normally 65 or 60 seconds, but twice it had been 50 seconds. When observing from the shore, he had measured 80 seconds for trains travelling through the girders, but not on any train he had travelled on. North bound local trains were often held up to avoid delaying expresses, and then made up time while travelling over the bridge. The gradient onto the bridge at the northern end prevented similar high speeds on south bound locals. Robertson said that the movement he observed was hard to quantify, although the lateral movement, which was probably , was definitely due to the bridge, not the train, and the effect was more marked at high speed. Four other train passengers supported Robertson's timings but only one had noticed any movement of the bridge. The Dundee stationmaster had passed Robertson's complaint about speed (he had been unaware of any concern about oscillation) on to the drivers, and then checked times from cabin to cabin (at either end of the bridge the train was travelling slowly to pick up or hand over the baton). However he had never checked speed through the high girders. Painters who had worked on the bridge in mid 1879 said that it shook when a train was on it. When a train entered the southern high girders the bridge had shaken at the north end, both east–west and, more strongly, up and down. The shaking was worse when trains were going faster, which they did: "when the Fife boat was nearly over and the train had only got to the south end of the bridge it was a hard drive". A joiner who had worked on the bridge from May to October 1879 also spoke of a lateral shaking, which was more alarming than the up and down motion, and greatest at the southern junction between the high girders and the low girders. He was unwilling to quantify the amplitude of motion, but when pressed he offered When pressed further he would only say that it was distinct, large, and visible. One of the painters' foremen, however, said the only motion he had seen had been north–south, and that this had been less than .

The North British Railway maintained the tracks, but it retained Bouch to supervise maintenance of the bridge. He appointed Henry Noble as his bridge inspector. Noble, who was a bricklayer, not an engineer, had worked for Bouch on the construction of the bridge. Whilst checking the pier foundations to see if the river bed was being scoured from around them, Noble had become aware that some diagonal tie bars were 'chattering', and in October 1878 had begun remedying this. Diagonal bracing was by flat bars running from one lug at a column section top to two sling plates bolted to a lug at the base of the equivalent section on an adjacent column. The bar and sling plates all had a matching longitudinal slot in them. The tie bar was placed between the sling plates with all three slots aligned and overlapping, and then a gib was driven through all three slots and secured. Two "cotters" (metal wedges) were then positioned to fill the rest of the slot overlap, and driven in hard to put the tie under tension. Noble had assumed the cotters were too small and had not been driven up hard in the first place, but on the chattering ties the cotters were loose, and even if driven fully in would not fill the slot and put the bar under tension. By fitting an additional packing piece between loose cotters and driving the cotters in, Noble had re tightened loose ties and stopped them chattering. There were over 4,000 gib and cotter joints on the bridge, but Noble said that only about 100 had had to be re tensioned, most in October–November 1878. On his last check in December 1879, only two ties had needed attention, both on piers north of the high girders. Noble had found cracks in four column sections – one under the high girders, three to the north of them – which had then been bound with wrought iron hoops. Noble had consulted Bouch about the cracked columns, but not the chattering ties.

The workers at the Wormit foundry complained that the columns had been cast using 'Cleveland iron', which always had scum on it—it was less easy to cast than 'good Scotch metal' and more likely to give defective castings. Moulds were damped with salt water, cores were inadequately fastened, and moved, giving uneven column wall thickness. The foundry foreman explained that where lugs had been imperfectly cast; the missing metal was added by 'burning on'. If a casting had blowholes or other casting defects considered to be minor faults, they were filled with 'Beaumont egg' (of which the foreman kept a stock for that purpose) and the casting was used.

Gilkes' site staff were inherited from the previous contractor. Under the resident engineer there were seven subordinates including a foundry manager. The original foundry manager left before most of the high girders pier column sections were cast. His replacement was also supervising erection of the bridge, and had no previous experience of supervising foundry work. He was aware of 'burning on', but the use of Beaumont egg had been hidden from him by the foreman. When shown defects in bridge castings, he said he would not have passed the affected columns for use, nor would he have passed columns with noticeably uneven wall thickness. That was on the instructions of the resident engineer, who had little foundry experience either and relied upon the foreman. Whilst the working practices were the responsibility of Gilkes, their contract with NBR provided that all work done by the contractor was subject to the approval of the workmanship by Bouch. Hence Bouch would share the blame for any resulting defective work in the finished bridge. The original foundry foreman, who had been dismissed for drunkenness, vouched for Gilkes personally testing for unevenness in the early castings: "Mr. Gilkes, sometimes once a fortnight and sometimes once a month, would tap a column with a hammer, first on one side and then on the other, and he used to go over most of them in that way sounding them." Bouch had spent over £9,000 on inspection (his total fee was £10,500) Bouch told the court that Paterson's age was 'very much mine' but, in fact, Paterson was 12 years older and, by the time of the Inquiry, paralysed and unable to give evidence. Another inspector appointed later The completed bridge had been inspected on Bouch's behalf for quality of assembly, but that was after the bridge had been painted (though still before the bridge opened, and before the painter witnesses were on it in the summer of 1879), which hid any cracks or signs of burning on (though the inspector said that, in any case, he would not know those signs on sight). Throughout construction, Noble had been looking after foundations and brickwork.

Henry Law had examined the remains of the bridge; he reported defects in workmanship and design detail. Cochrane and Brunlees, who gave evidence later, largely concurred.

Samples of the bridge materials, both cast and wrought iron, were tested by David Kirkaldy, as were a number of bolts, tiebars, and associated lugs. Both the wrought and cast iron had good strength, while the bolts "were of sufficient strength and proper iron". However, both ties and sound lugs failed at loadings of about 20 tons, well below what had been expected. Both ties

Bouch had designed the bridge, assisted in his calculations by Allan Stewart. After the accident Stewart had assisted William Pole in calculating what the bridge should have withstood. On the authority of Stewart they had assumed that the bridge was designed against a wind loading of 'with the usual margin of safety'. Bouch said that whilst had been discussed, he had been "guided by the report on the Forth Bridge" to assume and therefore made no special allowance for wind loading. He was referring to advice given by the Astronomer Royal, Sir George Biddell Airy in 1873 when consulted about Bouch's design for a suspension bridge across the Firth of Forth; that wind pressures as high as might be encountered very locally, but averaged over a 1600 feet (488 m) span would be a reasonable allowance. This advice had been endorsed by a number of eminent engineers. Bouch also mentioned advice given by Yolland in 1869 – that the Board of Trade did not require any special allowance for wind loading for spans less than 200 feet (61 m), whilst noting this was for the design of girders not piers. Pole referred to Smeaton's work, where high winds were said to give , with higher values being quoted for winds of or above, with the caveat that these were less certain. Brunlees had made no allowance for wind loading on the Solway viaduct because the spans were short and low – if he had had to, he would probably have designed against with a safety margin of 4–5 (by limiting strength of iron). Pole had ignored it because no reference was given; he did not believe any engineer paid any attention to it when designing bridges; he thought a reasonable allowance; this was what Robert Stephenson had assumed for the Britannia Bridge. Benjamin Baker said he would design to with a safety margin, but in 15 years of looking he had yet to see wind overthrow a structure that would withstand He doubted Rankine's pressures because he was not an experimentalist; told that the data were observations by the Regius Professor of Astronomy at Glasgow University, he doubted that the Professor had the equipment to take the readings.

Baker argued that the wind pressure on the high girders had been no more than , from the absence of damage to vulnerable features on buildings in Dundee and the signal cabins at the south end of the bridge. The Inquiry felt that these locations were significantly more sheltered, and therefore rejected this argument. Baker's subsequent work on wind pressures at the Forth Rail Bridge site showed meteorologists were overestimating, but his might have over interpreted the data.

Both Pole and Law had calculated the wind loading needed to overturn the bridge to be over (taking no credit for holding down bolts fastening the windward columns to the pier masonry) and concluded that a high wind should have overturned the bridge, rather than cause it to break up (Pole calculated the tension in the ties at windloading to be more than the 'usual margin of safety' value of 5 tons per square inch but still only half the failure tension.) Pole calculated the wind loading required to overturn the lightest carriage in the train (the second class carriage) to be less than that needed to overturn the bridge; whereas Law – taking credit for more passengers in the carriage than Pole and for the high girders partially shielding carriages from the wind – had reached the opposite conclusion.

Law concluded that the bridge as designed if perfect in execution would not have failed in the way seen (Cochrane went further; it 'would be standing now'). The calculations assumed the bridge to be largely as designed, with all components in their intended position, and the ties reasonably evenly loaded. If the bridge had failed at lower wind loadings, this was evidence that the defects in design and workmanship he had objected to had given uneven loadings, significantly reduced the bridge strength and invalidated the calculation. (Bouch's own view that collision damage to the girder was the sole cause of bridge collapse found little support).

Bouch's counsel called witnesses last; hence his first attempts to suggest derailment and collision were made piecemeal in cross examination of universally unsympathetic expert witnesses. Law had 'not seen anything to indicate that the carriages left the line' (before the bridge collapse) nor had Cochrane nor Brunlees. The physical evidence put to them for derailment and subsequent impact of one or more carriage with the girders was limited. It was suggested that the last two vehicles (the second class carriage and a brake van) which appeared more damaged were those derailed, but (said Law) they were of less robust construction and the other carriages were not unscathed. Cochrane and Brunlees added that both sides of the carriages were damaged "very much alike". Bouch pointed to the rails and their chairs being smashed up in the girder holding the last two carriages, to the axle box of the second class carriage having become detached and ending up in the bottom boom of the eastern girder, to the footboard on the east side of the carriage having been completely carried away, to the girders being broken up, and to marks on the girders showing contact with the carriage roof, and to a plank with wheel marks on it having been washed up at Newport but unfortunately then washed away. Bouch's assistant gave evidence of two sets of horizontal scrape marks (very slight scratches in the metal or paint on the girders) matching the heights of the roofs of the last two carriages, but did not know the heights he claimed to be matched. At the start of one of these abrasions, a rivet head had lifted and splinters of wood were lodged between a tie bar and a cover plate. Evidence was then given of flange marks on tie bars in the fifth girder (north of the two rearmost carriages), the 'collision with girders' theory being duly modified to everything behind the tender having derailed. The tender coupling (which clearly could not have hit a girder) had also been found in the bottom boom of the eastern girder. Two marked fifth girder tie bars were produced; one indeed had 3 marks, but two of them were on the underside. Dugald Drummond, responsible for NBR rolling stock, had examined the wheel flanges and found no 'bruises' – expected if they had smashed up chairs. If the second class carriage body had hit anything at speed, it would have been 'knocked all to spunks' without affecting the underframe. Had collision with the eastern girder slewed the frame, it would have presented the east side to the oncoming brake van, but it was the west side of the frame that was more damaged. Its eastern footboard had not been carried away; the carriage had never had one (on either side). The graze marks were at above the rail, and 11 feet (3 m) above the rail and did not match carriage roof height. Drummond did not think the carriages had left the rails until after the girders began to fall, nor had he ever known a carriage (light or heavy) to be blown over by the wind.

According to Yolland and Barlow "the fall of the bridge was occasioned by the insufficiency of the cross bracings and fastenings to sustain the force of the gale on the night of December 28th 1879 the bridge had been previously strained by other gales". Rothery agreed, asking "Can there be any doubt that what caused the overthrow of the bridge was the pressure of the wind acting upon a structure badly built and badly maintained?" He therefore concluded – with (he claimed) the support of circumstantial evidence – that the bridge might well have failed at the north end first; he explicitly dismissed the claim that the train had hit the girders before the bridge fell. Their report is therefore consistent with either a view that the train had not hit the girder or one that a bridge with cross bracing giving an adequate safety margin against windloading would have survived a train hitting the girder. Yolland and Barlow noted "there is no requirement issued by the Board of Trade respecting wind pressure, and there does not appear to be any understood rule in the engineering profession regarding wind pressure in railway structures; and we therefore recommend the Board of Trade should take such steps as may be necessary for the establishment of rules for that purpose." Rothery dissented, feeling that it was for the engineers themselves to arrive at an 'understood rule', such as the French rule of or the US .

Rothery's minority report is more detailed in its analysis, more willing to blame named individuals, and more quotable, but the official report of the court is a relatively short one signed by Yolland and Barlow. Rothery said that his colleagues had declined to join him in allocating blame, on the grounds that this was outside their terms of reference. However, previous Section 7 inquiries had clearly felt themselves free to blame (Thorpe rail accident) or exculpate (Shipton on Cherwell train crash) identifiable individuals as they saw fit, and when Bouch's solicitor checked with Yolland and Barlow, they denied that they agreed with Rothery that "For these defects both in the design, the construction, and the maintenance, Sir Thomas Bouch is, in our opinion, mainly to blame."

No further judicial enquiries under Section 7 of the Regulation of Railways Act 1871 were held until the Hixon rail crash in 1968 brought into question both the policy of the Railway Inspectorate towards automated level crossings and the management by the Ministry of Transport (the Inspectorate's parent government department) of the movement of abnormal loads. A Section 7 judicial enquiry was felt necessary to give the required degree of independence. The structure and terms of reference were better defined than for the Tay Bridge inquiry. Brian Gibbens, QC, was supported by two expert assessors, and made findings as to blame/responsibility but not as to liability/culpability.

The locomotive, NBR no. 224, a 4 4 0 designed by Thomas Wheatley and built at Cowlairs Works in 1871, was salvaged and repaired, remaining in service until 1919, nicknamed "The Diver"; many superstitious drivers were reluctant to take it over the new bridge. The stumps of the original bridge piers are still visible above the surface of the Tay. Memorials have been placed at either end of the bridge in Dundee and Wormit. A column from the bridge is on display at the Dundee Museum of Transport. On 28 December 2019, Dundee Waterfronts Walks hosted a remembrance walk to mark the 140 year anniversary of the Tay Bridge disaster.

Some in the Sabbatarian movement, which advocated for restricting activities on Sundays, pointed to the disaster as a punishment from God for traveling on the Sabbath. James Begg, a minister of the Free Church of Scotland, criticized the railway, stating that "The Sabbath of God has been dreadfully profaned by our great public companies", and also criticized the victims, posing the question, "Is it not awful to think that they must have been carried away when many of them must have known that they were transgressing the Law of God? ".

Various additional pieces of evidence have been advanced in the last 40 years, leading to "forensic engineering" reinterpretations of what actually happened.

The disaster inspired several songs and poems, most famously William McGonagall's "The Tay Bridge Disaster", widely considered to be of such a low quality as to be comical. The German poet Theodor Fontane, shocked by the news, wrote his poem It was published only ten days after the tragedy happened. C. Horne's ballad In Memory of the Tay Bridge Disaster was published as a broadside in May 1880. It describes the moment of the disaster as: