Design

Amasa Stone, the bridge's designer

Amasa Stone's bridge was, by his own admission, experimental.^[16] He had constructed only one all-iron Howe truss bridge before, a 5-foot (1.5 m) high, 30-foot (9.1 m) long railroad bridge over the Ohio and Erie Canal in Cleveland.^{[10][lower-alpha 4]}

Joseph Tomlinson, a well-known bridge builder and designer, was hired to flesh out Stone's design and create the fabrication drawings for all the bridge components.^[17] Tomlinson designed the bridge's lower chord to have a camber^{[lower-alpha 5]} of 6 inches (150 mm). When the falsework supporting the bridge was removed and the dead load of the bridge came into play, the camber would drop to between 3.5 to 4 inches (89 to 102 mm).^[9] Tomlinson was alarmed when Stone demanded that the bridge be constructed completely of iron, rather than a combination of wood and iron.^{[lower-alpha 6]} An all-iron bridge would have a much greater dead load, reducing the bridge's live load (its ability to carry trains).^[10] He also concluded that the beams and posts Stone intended to use were undersized.^{[3][lower-alpha 7]} Tomlinson proposed riveting plates to the I-beams to strengthen them, but Stone angrily refused.^[19] Stone demanded that Tomlinson make the changes he required. Tomlinson refused, and was fired from the design effort.^[3][10] Stone then ordered the CP&A's chief engineer, Charles Collins, to make the desired changes to the bridge design. Collins refused, and was fired from the design effort.^[20] Stone then made the changes to the design.^{[21][lower-alpha 8]}

Stone made additional changes to the design. In a Howe truss bridge, the vertical posts connect the upper and lower chords (main parallels) in the truss. The deck on which the train travels usually hangs from these posts; the greater the live load, the greater the tension on the posts. The bracing reacts in compression, counteracting the tension. Amasa Stone inverted this design so that only the upper chord (now at the bottom of the bridge) provided tension. Where diagonal braces did not receive the extra compression from a live load, inverted Howe truss bridges had a tendency to buckle where the vertical posts were attached to the deck with cast iron angle blocks.^[21] Stone's other major change involved the end panels. In the traditional Howe truss bridge, the end panel on each side of each end of the bridge has three vertical posts and three diagonal braces. Only five Howe truss bridges ever built by 1863 had just one vertical post and two diagonal braces in the end panels. These were known as "Single Howe" bridges. Amasa Stone used the "Single Howe" design for the end panels at Ashtabula.^[11] Thus, the bridge's entire structure relied on just 12 beams and posts (three at each end).^[11][25]

Design of the angle blocks

The chord, diagonals, angle blocks, and vertical posts of the Ashtabula Bridge, as drawn from original plans by Charles MacDonald in 1877

Gasparini and Fields claim that the exact design of the angle blocks and the ends (the "bearings") of the diagonals are lost to history.^[26]

Civil engineer Charles MacDonald, who inspected the bridge's original plans in 1877,^{[lower-alpha 9]} described and made drawings of part of the angle blocks. He noted that the vertical posts were made of iron pipe 3 inches (76 mm) in diameter with a wall 0.5 inches (13 mm) thick. Inside the pipe ran an iron rod 2 inches (51 mm) thick. The top of the rod passed through the space between the members of the chord at the top of the bridge and then through a gib-plate.^{[lower-alpha 10]} A nut and washer screwed onto the upper end of the rod, creating tension as well as securing the gib-plate in place.^[29] Those angle blocks at the top of the bridge had vertical, squarish lugs. Those members of the chord which ended atop an angle block had their bearings placed against the lug. These lugs served to transmit stress from the chord to the angle block and thence to the diagonals. These upper angle blocks also had lugs facing inward, to which were attached (by means MacDonald did not describe) the lateral braces.^[30] The interior side of each upper angle block also had a recess to accept a lug and a tap bolt.^{[lower-alpha 11]} The tap bolt was used to connect the lug on the end of the sway rod to the angle block.^{[32][lower-alpha 12]}

MacDonald described (but did not publish a drawing of) the angle blocks at the bottom of the bridge. The bottom of the rod in the vertical posts screwed into these angle blocks. The members of the chord at the bottom of the bridge were flat bars, not I-beams, each bar measuring 5 by 0.375 inches (127.0 by 9.5 mm). Where a member of the chord ended at an angle block, a 3-by-1-inch (76 by 25 mm) lug was forged at the base of the bar. This lug fit into a slot in the angle block. The angle blocks which made up the chord at the bottom of the bridge also had lugs facing inward, to which were attached (by means MacDonald did not describe) the lateral braces.^[30]

MacDonald and Gasparini and Fields noted that the diagonal I-beams were designed to connect to both the upper and lower angle blocks with the flanges of the I-beam in a vertical position. The web of the I-beam fit into a horizontal slot between two lugs.^[26][32]

It is also known that, at the ends of the bridge, only half of each angle block received load because Stone used only a single diagonal in the end panel. This put enormous shear stress on the bridgeward side of these angle blocks.^[33]

Construction

One design for a half-angle block. The attached chord puts immense downward (shear) stress on one side of the block only, for which the block is not designed.

The Ashtabula River bridge was erected in 1865 using Stone's design and plans and partly under his supervision.^[11] Tomlinson was the bridge's original construction supervisor, but Stone said he fired him for "inefficiency" at some point during the bridge's construction. Tomlinson was replaced by A. L. Rogers.^[23]

When construction began, Tomlinson observed that the I-beams intended for use as diagonals were smaller than the fabrication plans called for.^[26]

The amount of camber created a problem during construction. At Congdon's suggestion, Rogers built falsework to support construction of a bridge with a 5-to-7-inch (130 to 180 mm) camber. Stone, now himself supervising Rogers' work, ordered the camber reduced to 3.5 inches (89 mm). With the members of the upper chord now too long, Rogers had the bearings shaved down. It is clear Rogers ordered other changes as well, but it is uncertain what these included. Gasparini and Fields suggest he had the lugs atop the angle block planed down as well.^[15] When the falsework began to be removed, the dead load caused the bridge to bend about 2.5 inches (64 mm) below horizontal.^[34] The bridge was jacked up and the falsework put back in place. Stone then ordered the chord members to be returned to their original lengths, restoring Tomlinson's intended camber. Rather than ordering new I-beams, Rogers used shims to close the space between the bearings and the lugs.^[35]

When the falsework was removed a second time, the bridge buckled where the vertical posts connected to the deck.^[14][21] Several diagonals also buckled.^{[35][lower-alpha 13]} Once more, the falsework went back in place.^[35]

To correct this problem, Stone added more iron I-beams to the diagonals to strengthen them.^[36] The placement, size, and number of beams added is not clear, but Stone likely added two I-beams to the brace in the end panel, two I-beams to the brace in the first panel from the end, and one I-beam to the second panel from the end.^[35] This worsened the bridge's dead load problem.^[11] Collins, Congdon, Rogers, and Stone all later testified that the I-beams making up the diagonals were now turned 90 degrees, so that the flanges were horizontal. Congdon says that he realized the I-beams would carry more live load if they were rotated.^{[lower-alpha 14]} Collins, Rogers, and Stone believed workers had installed the beams incorrectly (on their sides).^{[35][lower-alpha 15]} To make the change, Stone had workers cut away portions of each diagonal I-beam's web at the bearing, enabling the web to fit over the lugs. This weakened the new diagonals.^[11] There is also some evidence that the angle blocks were damaged while the braces and counter-braces were rotated.^[37]

The bridge was prestressed again.^[35] In every other panel connection, the diagonal braces were fitted to the angle blocks using shims rather than by tightening the vertical posts and putting the diagonals under compression. This meant that the shims carried the weight of a live load, rather than the braces themselves.^[38] It is also possible that the shims created uneven contact, causing angle blocks to undergo both bending and shear stress.^[39] Nevertheless, the bridge did not sag this time.^[35]

Upon completion, the bridge was tested by having three locomotives run over the bridge at speed. A second test had the three engines stand still on the bridge. Deflection was minimal and the bridge rebounded satisfactorily.^{[40][41][42][lower-alpha 16]}

Blizzard conditions

Train No. 5^[43] of the Lake Shore and Michigan Southern Railway,^{[lower-alpha 17]} known as The Pacific Express, left Buffalo, New York, at 2 PM on December 29, 1876, 1 hour and 8 minutes behind schedule.^[46] A powerful blizzard began hitting northern Ohio, northwest Pennsylvania, and western New York two days earlier.^{[47][lower-alpha 18]} More than 20 inches (510 mm) of snow had already fallen,^[49] and winds 24 to 54 miles per hour (39 to 87 km/h)^[50] were creating heavy snowdrifts on the railroad tracks^[51] 6 feet (1.8 m) deep in places.^[52] The snow was so heavy that, shortly after leaving Buffalo, a second engine was added to help pull the train.^[46]

The train left Erie, Pennsylvania, at 5:01 PM,^[51] an hour and 16 minutes behind schedule.^{[lower-alpha 19]} Its two locomotives, Socrates and Columbia, were hauling two baggage cars, two day-passenger coaches, two express coaches, a drawing room car (the "Yokohama"), three sleeper cars (the "Palatine", which originated in New York City and was bound for Chicago; the "City of Buffalo", which originated in Boston and was bound for Chicago; and the "Osceo", a sleeper for passengers going to St. Louis), and a smoking car^[53][59][60] with about 150 to 200 passengers and 19 crew aboard.^{[lower-alpha 20][lower-alpha 21][lower-alpha 22]} Two additional locomotives were needed to push the train away from the station due to the heavy snow.^[53][55]

Initial collapse and survival of the "Socrates"

Illustration of the bridge collapse

The No. 5 was due to arrive in Cleveland at 7:05 PM,^[58] but at about 7:30 PM^[68][73] it was just reaching Ashtabula—an hour and 53 minutes behind schedule.^[58] About half an hour earlier, the No. 8 passenger train of the LS&MS passed over the Ashtabula River bridge without incident, heading east.^[60] The bridge over the Ashtabula River was about 1,000 feet (300 m) east of the Ashtabula station,^[75] and the locomotives shut off their steam (cutting off power to the drive) about 66 to 99 feet (20 to 30 m) east of the bridge to allow the train to glide into the station.^[76] As the Pacific Express crossed the bridge, it was doing 10 to 15 miles per hour (16 to 24 km/h) (according to the locomotive engineers).^[60][76] Visibility was practically nonexistent,^[77] at most one or two car lengths.^[60]

As the "Socrates" neared the western abutment, engineer Daniel McGuire heard a crack and felt his locomotive drop slowly downward. Realizing the bridge was collapsing beneath him, he opened the throttle to maximum. The "Socrates" lurched ahead, just as the weight of the 11 cars began to pull on the "Columbia" behind it. The connection between the two locomotives snapped, enabling the "Socrates" to make it off the bridge.^[77] The rear trucks on his tender hung in the air, but the forward movement of the "Socrates" pulled the tender forward and it regained the rails and solid ground. McGuire brought the "Socrates" to a halt about 100 yards (91 m) down the track, and began repeatedly sounding his whistle and ringing the train bell in alarm.^[60]

Collapse of the remaining bridge

Postcard photo of the wreck. The view is toward the west abutment; the locomotive "Columbia" can be seen in the wreckage at left.

The "Columbia" and the 11 cars behind it acted like a linked chain load. The bridge's collapse was therefore not sudden, but rather somewhat slow and piecemeal.^[77] The "Columbia" struck the abutment, the engine supported by the stonework while the tender hung downward toward the river valley.^[76] The first express car fell into the ravine,^{[lower-alpha 23]} crashing nose-first into the ground at the base of the abutment.^{[78][lower-alpha 24]} The "Columbia" slipped backward off the abutment, landing upside-down and backward atop the first express car.^[60][79] It then fell onto its side, its trucks pointing north.^[80]

The second express car and the two baggage cars landed largely upright, slightly south of the bridge. The second baggage car was slightly askew, its nose resting against the western abutment and its rear pointing southeast. Most of the bridge's upper chord (the bottom of the bridge) crashed to the ground north of the bridge.^[78] The lower chord (at the top of the bridge) and what remained of the deck held for a moment, then fell directly down to land atop the locomotive, express cars, and baggage cars.^[81] Momentum pulled the rest of the train into the space where the bridge used to be. The first passenger coach landed upright in mid-stream atop the wreckage of the bridge and the second express car.^{[lower-alpha 25]} The second passenger car twisted in the air as it fell, landing on its side atop the bridge and first baggage car. The smoking car, having broken free of the passenger coach ahead of it, moved more freely. It struck the forward part of the second passenger coach, crushing it, before being propelled into the first passenger coach. (It is widely believed most people in the first passenger coach died when the smoking car fell on them.)^[83]

Momentum also carried the parlor car "Yokohama" and the three sleeper cars into the chasm. All of them landed about 80 feet (24 m) south of the bridge. The "Yokohama" landed upright in mid-stream,^[83] and the sleeper "Palatine" landed mostly right-side up beside it to the north.^[60][83] The sleeper "City of Buffalo" then nose-dived into the rear of the "Palatine", partially crushing it and killing several people.^[60] It continued through the "Palatine" into the rear of the "Yokohama", pushing the parlor car onto its side. The "Buffalo" smashed forward along the length of the parlor car, likely killing everyone inside.^[84] The rear of the "Buffalo" lay atop the "Palatine", high in the air.^[85] An eyewitness said no one in the "City of Buffalo" survived the crash.^[60] The final sleeper, the "Osceo", landed on the east bank of the river, mostly upright.^[83]

Fire and deaths

1878 drawing based on a January 1877 photograph of the ruins of the bridge

Disaster site in 2015

The crash was heard at the railroad's W. 32nd Street Station^[86] (just 100 yards (91 m) to the west of the bridge)^[79] and in the town, and the alarm was raised.^[87] Rescuers first on the scene included railroad employees, those waiting on the platform at the station, and residents of Ashtabula who lived near the bridge.^[86] The only access to the valley floor was a set of steep, narrow steps, covered in snow.^[88] Most people slid down the steep incline rather than take the steps, and several people brought axes to help free survivors.^[86]

The wooden cars burst into flame when their coal- and kerosene-fed heating stoves and oil lamps overturned.^[8] An early report said fire broke out in the "Osceo" and in at least three other places, and within a minute the entire wreck was ablaze.^[60] Historian Darrell E. Hamilton says the fire broke out at either end, and moved toward the middle.^[86] According to rear brakeman A.H. Stone, those still left alive in the wreckage were dead within 20 minutes.^[89] By the time rescuers reached the bridge, many wounded passengers had already made their way to the shore^[90] and the fire was burning fiercely.^[91]

The response by the Ashtabula Fire Brigade was minimal. G.W. Knapp, the city's fire chief, was an alcoholic who even when sober was slow to make decisions and easily confused.^[86] The Lake Erie Hose Company's hand- and steam-pumped horse-drawn fire engines arrived first,^{[92][lower-alpha 26]} but Knapp never gave any orders to fight the flames. He told a bystander that there was no use in fighting the fire,^[86] even though it was plain that some survivors were still trapped in the wreck.^[95] Railroad employees also told Knapp that his firefighters should get the wounded out and clear a pathway up the side of the ravine.^{[96][lower-alpha 27]} At least one member of the town begged Knapp to put water on the flames, but he refused. Instead, the townspeople secured buckets and (with the help of some members of the fire brigade) tried to put out the blaze.^[92] The Protection Fire Company's hand-pumped engine and the Neptune Hose Company's steam-pumper (both horse-drawn) were hauled more than a mile through town to the bridge, but arrived too late to stop the spread of the fire.^{[94][lower-alpha 28]}

The darkness and blinding snow made it difficult for surviving passengers to orient themselves and get out of the wreck.^[79] A number of passengers drowned in the river,^[79] while others escaped the blaze only to die of smoke inhalation.^[97]

The injured and dying were either carried up the steep steps or hauled up the incline on sleds or sleighs pulled up by rope.^[98] There was no hospital in Ashtabula. The injured were first taken to the railroad engine house, to the filthy and run-down Eagle Hotel adjacent to the station, or to the nearby Ashtabula House hotel. As these places filled, residents opened their homes to the survivors.^[98] Ambulatory injured were the last to be hauled up from the valley. By midnight, all the survivors had reached safety.^[98] The 10 doctors in the village attended to the wounded. About 1 AM, a special train arrived from Cleveland carrying railroad officials and five more surgeons.^[98]

Thieves moved among the dead and wounded,^[99] stealing money and valuables. A large crowd of the curious gathered at the wreck site the next morning, and some in the crowd looted the train until Ashtabula's mayor, H. P. Hepburn,^{[94][100][lower-alpha 29]} stationed a guard at the site.^[97] Some of the money and valuables were discovered following investigations by local police, and a few arrests made.^[97] Hepburn later issued a proclamation promising amnesty for anyone else who turned over stolen items. Money and valuables worth about $1,500 ($41,222 in 2022 dollars) were collected,^[101] but most of the stolen money and goods were never recovered.^[97]

The number of people killed in the Ashtabula bridge disaster will never be known.^[66] The number could be as low as 87 or as high as 200, although the official count is 92 dead.^{[6][8][37][66]} Another 64 people were injured.^[6][8][37] The number of deaths is inexact, in part because the number of passengers on the train is difficult to estimate and in part because many remains were partial (a hand, a leg, a torso). Most (although not all) remains recovered from the wreck were burned beyond recognition and could not be identified from clothing or personal items.^[102] An unknown number of the dead were essentially cremated in the blaze.^[103] Among the dead was hymn writer Philip Bliss.^{[92][lower-alpha 30]}

Identifying and burying the dead

Identification of the dead took a week or more. There were full or partial remains of about 36 bodies in the railroad freight house, with concerned families encouraged to come by and try to identify corpses.^[97] For several days after the wreck, townspeople and railroad employees used their hands and feet, hoes, rakes, and shovels to dig through the ash, ice, mud, and snow to find any personal items they could. These items—which included partially burned train tickets, diaries, photographs, watches, jewelry, unique or rare items of clothing, or keepsakes—were kept by the railroad. When a corpse could not be identified, grieving families sometimes were able to use these "relics" to confirm that a loved one had been aboard the train.^[105] Even so, identifying objects or papers often were separated from remains, and misidentification of remains was common.^[106]

Concerned friends and family members sent letters and telegrams in the hundreds to railroad and civic authorities, seeking knowledge of their loved ones. These contained descriptions of the alleged passenger, as well as any identifying personal effects. Some of these letters were fraudulent, sent by people seeking gold watches, jewelry, or other items as "loot". Fraudulent letters tended to be spotted fairly easily, and were not answered.^[107]

Wreck investigators were still turning up remains as late as mid-January.^[108]

A burial service for the unidentified dead was held at Ashtabula's Chestnut Grove Cemetery on January 19, 1877.^{[109][lower-alpha 31]} A mile-long procession conveyed the dead to the cemetery.^[110] The railroad purchased a burial plot in which 18 coffins, containing the remains of an estimated 22 people, were placed.^[103]

Three coffins with three corpses remained at the freight house in the hope that they could still be identified. When these remains went unclaimed, they were buried about a week later in the same plot at Chestnut Grove.^[103]

Conclusions by the coroner's jury

There being no coroner in Ashtabula,^[112] a coroner's jury of six men from the town was convened by Justice of the Peace Edward W. Richards on December 31.^[113] The jury took testimony from 20 railroad officials and employees (including both locomotive engineers and the rear brakeman), nine members of the Ashtabula fire department, 10 residents of Ashtabula, six passengers, and eight civil engineers and bridge builders.^[94] The coroner's jury submitted its report on March 8, 1877.^[94]

The coroner's jury blamed the collapse of the Ashtabula River bridge and the deaths by fire on five factors:^[94]

1. The bridge was poorly designed. An all-iron Howe truss was unsuitable for such a long bridge. Some of the members of the chords and braces were not designed to be of sufficient strength, and were poorly placed. The lateral cross-bracing was so underdesigned as to be of little value. The angle blocks had too few and poorly designed lugs, which did not keep the braces and counter-braces from slipping out of place.^{[lower-alpha 33]}
2. The bridge was poorly constructed. Each member of the bridge acted independently instead of being positively connected to its neighbors. Some of the vertical posts, braces, and cross-braces were put in the wrong places. To accommodate design changes during the bridge's strengthening, the lugs on the angle blocks and the flanges on the I-beams of the chords were cut away in part, reducing their strength and effectiveness. The construction of the yokes used to bind the braces and counter-braces was poor, and shims were used to compensate for chord members which were too short. The modifications made to the bridge before its completion used thick members where thin ones were required, and thin ones where thick ones should have been used.
3. Railroad bridge inspectors did not perform their jobs adequately. Inspectors should have noted the severe deficiencies in the bridge's design and construction, and did not. They also should have observed problems with members becoming loose over time.^{[lower-alpha 34]}
4. The railroad failed to use self-extinguishing heating stoves, as required by state law passed on May 4, 1869.
5. The fire was worsened by the failure of those on the scene to attempt to douse the flames. The Lake Erie Hose Company's hand pump and steam pump, first on the scene, could have saved lives. The Neptune Hose Company's steam pump and the Protection Fire Company's hand pump were hauled more than 1 mile (1.6 km) through snowdrifts, but arrived too late.

Amasa Stone was held personally responsible for the bridge's poor design, and the railroad company for the lax inspection of the bridge and the failure to use self-extinguishing heating apparatus. G.W. Knapp was held personally responsible for failing to fight the fire in a timely fashion.^[94]

Conclusions by the state legislative joint committee

Sketch of the Ashtabula Bridge wreckage

On January 1, 1877, the Ohio General Assembly adopted a joint resolution appointing a committee to investigate the causes of the Ashtabula River bridge collapse, and to make recommendations to the legislature. The committee consisted of five members of the Ohio House of Representatives and three members of the Ohio Senate.^[116] Senator A.M. Burns chaired the committee.^[117] The joint committee hired three civil engineers (Benjamin F. Bowen, John Graham, and Thomas H. Johnson) to examine the wreckage and report to the committee regarding the bridge's design and performance. The committee also received written reports from civil engineers Albert S. Howland^{[lower-alpha 35]} and W.S. Williams, and took personal testimony from civil engineer M.J. Becker and from railroad officials and employees Amasa Stone, Charles Collins, Albert Congdon, A.L. Rogers, and Gustavus Folsom (engineer of the "Columbia").^{[lower-alpha 36]} Technical advice was provided by civil engineers D.W. Caldwell and J.E. Wright. The coroner's jury granted the joint committee full access to all of its testimony and reports as well. In appendices to its report, the joint committee printed in full the coroner's jury testimony of civil engineers A. Gottlieb, John D. Crehore, and Joseph Tomlinson.^[119]

The legislative joint committee issued its report on January 30, 1877.^[40] The committee blamed the collapse of the Ashtabula River bridge on three factors:^[115][120]

1. The bridge's design was flawed. Nothing in the design prevented the lateral buckling of the braces or the upper chord.^{[lower-alpha 37]} Only a part of the upper chord was designed to transmit load to the angle blocks, and the lugs on the angle blocks were poorly designed (so that strain could not be fully transmitted to the braces and counter-braces). The bridge was apparently designed for a non-moving live load, and little evidence existed to show it was designed to accommodate jarring, oscillation, vibration, or the force of wind.
2. There were significant errors in construction. The members making up the braces were not connected to act in unison, greatly reducing the bridge's ability to carry a live load. In the lower chord, lateral braces were placed only at every other panel connection, extended across two panels (not one), and were not connected to one another where they crossed. The sway braces were too small to prevent sway, and placed only at every other panel as well. Neither the braces nor counter-braces were attached to the angle blocks, and their ends were not square.
3. The defects in design and construction could have been identified by careful inspection at any time, but the railroad's inspectors did not identify these problems. Many braces and counter-braces had fallen out of position before the bridge collapsed, something inspectors either did not notice or the railroad did not repair.

The committee found no defects in materials, and rejected the idea that application of the air brakes by the second locomotive caused the bridge to collapse.^[49]

Conclusions of the MacDonald investigation

A third investigation was conducted independently by civil engineer Charles MacDonald, who likely was commissioned to study the bridge's collapse by the American Society of Civil Engineers (ASCE).^[115]

MacDonald blamed three factors for the bridge's failure:^[122]

1. Improper inspection by railroad officials. He noted, however, that the LS&MS was not alone in infrequent inspection by largely untrained men.
2. An inadequate knowledge of the strength of iron. The whole industry suffered from a lack of this knowledge, he wrote.
3. The failure of a mis-cast lug.

MacDonald believed the bridge collapsed due to a flaw in the southernmost lug at the top of the first full angle block from the west end. He noted that the lugs atop the angle blocks on the bridge had, for some reason, been shaved down to 1.6875 inches (42.86 mm) from 2 inches (51 mm). This significantly weakened the ability of the lugs to transmit stress from the chord to the braces and counter-braces.^[32] According to his report, "The cast iron angle block at top of second set of braces had the south lug broken off close to the face, and the line of fracture disclosed an air hole extending over one half the entire section. ...[The] failure first began in the south truss, at the second panel point from the west abutment."^[123] The second full angle block^[124] "was so far impaired by an air hole as to be reduced in strength fully one half. ... At no other point were these lugs subjected to so great strain, except at the end casting and here they were heavily reinforced."^[125] This "defective detail" caused the collapse of the bridge.^[126]

Conclusions by the railroad

LS&MS president Amasa Stone categorically denied that there were any design or construction flaws.^[127] He initially blamed the bridge's collapse on the deraillment of one of the two locomotives pulling the train,^[37] or by a loose rail which caused the train's deraillment.^[128] The railroad also hypothesized that a tornado may have hit the bridge, causing it to come loose from the abutments.^[128]

Although the LS&MS refused to accept responsibility for the disaster, it paid out more than $500,000 ($13,700,000 in 2022 dollars) to victims and their families to quiet legal claims.^[129]

The angle block

The angle blocks were made of cast iron, which Åkesson says was appropriate, for the angle blocks were in compression.^[130] The vertical lugs atop the angle blocks in the upper chord of the bridge received shear stress, but this should have been minimal as shear stress was only induced by one of the adjacent members of the chord.^[33]

The critical issue, Åkesson says, is that the broken lug on the southern angle block atop the joint between the second and third panel had a void. The void itself weakened the integrity of the block. Voids also encourage the formation of large grains and can accumulate impurities like slag, both of which also increase the brittleness of iron.^[39] The void also worsened stress on the block through stress concentration effect.^[131] Gasparini and Fields conclude that this void, combined with metal fatigue, caused the lug to fail. Metal fatigue was an issue in cast and wrought iron which only a few metallurgists and engineers were aware of in the 19th century.^[132] The failure of this lug on this angle block caused the entire bridge to fail.^[133]

Neither the joint committee, nor the coroner's jury, nor any of the engineers employed by them identified the flaw in the angle block casting.^[134]

In retrospect, Gasparini and Fields say, the point where the lug transitioned to the main body of the angle block was extremely problematic. The solidification rates of the molten iron were so different, this was a likely site for the formation of cracks or voids.^[135]

Poor construction caused overwhelming thrust load

Poor construction of the bridge's diagonals worsened the stresses placed on the lugs on the angle blocks. Howe trusses rely on prestressing of braces and counter-braces to improve the way the bridge carries load. Tightening the nuts on the vertical posts (prestressing) puts the verticals in tension (stretches them). If the diagonals are already closely fitted to the angle blocks, prestressing compresses the diagonals. This allows them to carry more load. The Ashtabula Bridge diagonals, however, were only loosely fitted to the angle blocks. Prestressing brought the diagonals into a relatively close fit with the angle blocks, but did not put the diagonals in compression.^[14] The problem was worsened because shims were used to fill the space between the diagonal bearings and the angle blocks. The history of the bridge indicates that some of these shims had come loose over time and fallen away. The loss of shims induced uneven loading, as the more tightly connected diagonals absorbed load before the loose ones did.^[130] Åkesson points out that the shims themselves may even have created unequal pressure points between I-beams and the lugs, subjecting the lugs to bending forces as well as shear forces.^[39] With the diagonals not carrying the load they were intended to carry, extra stress was placed on the chords.^{[14][lower-alpha 38]} Unequal loading of the angle blocks worsened the metal fatigue.^[130]

The construction of the upper chord of the bridge was also poor. This chord consisted of five I-beams running in parallel.^[9][14] Having all five members end at panel joints actually weakened a bridge, so Howe trusses were built so that three ended at one panel connection and the other two at the next panel connection.^[14] As with the connection between diagonals and angle blocks, it was critical that there be no space between the I-beams and the lugs on top of the angle block because these lugs transferred axial forces to the next member. Space between the member and lug would reduce the effectiveness of this transfer and introduce shear stress to the lug.^[9] The problems with camber led to members of the chords being shortened^[15] and the lugs being shaved down,^[136] actions which introduced space between the lugs and the chord members.^{[lower-alpha 39]} Construction workers used metal shims to fill the space between the lug and the chord members until a tight fit was achieved. Friction alone, rather than an active connection such as a yoke or bolt, kept the shims in place.^[137] For some years prior to the disaster, locomotive engineers reported hearing "snapping sounds" as their trains crossed the Ashtabula Bridge. This indicates that some shims had come loose and fallen off, reintroducing space between the chord members and angle block vertical lugs. This allowed members of the chord to thrust suddenly against the lugs, inducing even more metal fatigue.^[130] There is also an indication in the construction record that several chord members were misaligned. Even if their bearings had been flat, they would not have met the lugs completely. This, too, would have created uneven loading and worsened metal fatigue.^[130]

Gasparini and Fields conclude that the bridge might have survived the loss of the lug had the chords and diagonals been made stronger through active continuous connection.^[138] Active continuous connection was not used on the bridge: The members of the chords were connected to angle blocks at only every other panel,^[9][137] the five beams making up each chord did not have a continuous interconnection between them,^{[9][lower-alpha 40]} and the none of the parallel I-beams making up the diagonals were continuously interconnected.^[9] Åkesson points out that construction errors probably made the diagonals even less effective as thin members were placed where thicker ones should have gone and vice versa.^[29] The braces and counter-braces in a Howe truss must be the same size for the truss system to be robust and redundant. Making a brace stronger relative to a counter-brace, for example, actually reduces robustness and redundancy by changing the relative distribution of forces on the diagonals. These errors appreciably lessened the bridge's ability to withstand extra loading.^[139] Stone's strengthening of the bridge after the camber repairs also harmed the bridge's capabilities. By adding two I-beams to the end braces, Stone actually reduced the maximum stress the braces in the end panels could bear.^[140]

Low temperature exacerbated the flaw in the lug

Low temperatures the night of the disaster also worsened the metal fatigue in the already damaged lug.^[130] The role that low temperatures played in creating and worsening metal fatigue was also poorly understood in the 19th century.^[132] The temperature at the time of the accident was 16 °F (−9 °C).^[52][135] Cast iron is prone to fracture and fatigue cracks,^[130] and Gasparini and Fields suggest that a significant fatigue crack, originating at the void in the lug, existed at the time of the disaster, created by repeated unequal stress over the previous 11 years.^[135] The low temperatures on the night of the accident increased the brittleness of the cast iron. An existing fracture worsened in the cold and likely caused the lug's failure.^[130][135]

The issue of poor inspection

Modern analyzes of the bridge collapse conclude that the railway had inadequately inspected and maintained the bridge.^[127][133] Åkesson, however, says that better inspection of the bridge may not have prevented a collapse. An inverted Howe truss puts the superstructure below the track, where it is difficult to see and inspect, and the angle blocks were hidden by the surrounding I-beams. Better inspection may have corrected some construction errors and identified falling shims, but might not have improved the bridge's survivability.^[141]

Death of Charles Collins

Days after testifying before the state legislature committee, LS&MS chief engineer Charles Collins was found dead in his bedroom of a gunshot wound to the head.^[151] Having tendered his resignation to the railroad's board of directors the previous Monday and been refused,^[152] Collins was believed to have committed suicide out of grief and feeling partially responsible for the tragic accident.^[153]

Two official autopsies, both conducted in 1878, concluded that Collins had been murdered. For unknown reasons, law enforcement officials did not release these reports at the time. The reports were rediscovered in 2001.^[154][155]

Collins was buried in an elaborate mausoleum at Chestnut Grove Cemetery.^[108]