American Merchant – the ‘man versus machine’ case
Posted: 14/03/2025
This year marks the bicentennial of the founding of heritage maritime law firm Thomas Cooper in 1825. To celebrate this occasion, lawyers from Penningtons Manches Cooper will, for each month of 2025, be chronicling a different standout case conducted by Thomas Cooper across its two centuries of English legal practice.
The third instalment in this series focuses on the year of Thomas Cooper’s first centennial and the collision case of the American Merchant [1925] 23 LI L Rep 1; the first time the courts considered nautical evidence produced by a machine.
The roaring twenties
By 1925, the Great War was well over, and F. Scott Fitzgerald’s The Great Gatsby had just been published. The United States emerged rich, powerful and jazzy, into an era of new technology. Millions of American homes had started tuning in to the golden age of radio, assembly line-made automobiles were widely affordable, and the new electric traffic light was freeing up the labour of thousands of traffic officers.
Non-magnetic gyrocompasses were first installed on commercial vessels in 1923, having been used by various navies during the war. Early gyrocompasses were delicate mechanisms supporting the revolution of a suspended flywheel in universal rotational freedom. The gyrocompass was innovative in allowing accurate determination of true north at all times irrespective of the vessel’s motion, ferrous hull, or the weather. This breakthrough was soon followed by the first gyroscope-guided autopilot steering systems.
Navigated by an iron man
Owned by the United States Shipping Board, the 7,430gt, 436ft long steamship American Merchant was making 14 knots along the Thames on a voyage from New Jersey to London. It was a ‘very fine morning’ on 24 March 1924, and the vessel was under the control of a licensed Trinity House pilot.
The vessel had a gyrocompass along with an ‘iron man’ – a contrivance whereby a lever was inserted into a notch in the helm corresponding with the gyrocompass bearing. Another apparatus would print on a record card the changes in heading according to the gyrocompass. Relying on this gyropilot, American Merchant had crossed the Atlantic without upset, and the record card said so.
In the opposite direction was headed British steamship Matatua, at 8,000 gt and 448ft in length, bound for New Zealand. Matatua was making a speed over ground of 10 knots and was also in the hands of a Trinity House pilot.
There was nothing untoward in the weather. The river was well-marked with buoys. Both vessels had the regulation lights for a steamship underway, all burning brightly, and both pilots knew to pass one another port-to-port, red light to red light.
A ‘terrible ripping and rending’
At around 4.55 am, between Lower Hope Reach and Sea Reach off the Mucking Flats, near Stanford-le-Hope, the vessels collided, severely. American Merchant had gone hard astern immediately beforehand, reducing speed to 11 knots, resulting in a combined speed of 21 knots on impact.
American Merchant’s stem breached Matatua’s port bow and ploughed halfway through the forecastle. According to Matatua’s chief officer, it was ‘not like a knife cutting into cheese, but there was a terrible ripping and rending.’ A fire broke out on Matatua that American Merchant’s crew helped to extinguish, and the vessels remained locked together for a considerable time. Matatua had to be beached by tugs. Eight of her crew had been killed and another three injured.
Admiralty Court
The Admiralty President, the Right Hon Sir Henry Duke, heard witnesses for each vessel, along with the gyrocompass manufacturers and related experts.
Matatua’s witnesses claimed that when the vessels were 800 feet apart, American Merchant suddenly made a hard turn to port, placing her on a collision course.
The witnesses for American Merchant alleged that Matatua did not attempt to pass port to port, but instead kept a direct course and threw herself across the bows of American Merchant. It was speculated that Matatua had been trying to make for Holehaven, where she was due to load the balance of her cargo.
The record card also did not show a hard turn to port. In the collision aftermath, however, it showed the locked-together vessels revolving through more than 360 degrees, including turning 180 degrees within 10 minutes. As this was refuted by American Merchant’s handwritten scrap log, the judge was satisfied that no such movement took place and declined to rely on the evidence of this ‘silent witness’.
Sir Henry Duke found that for the collision to have taken place as alleged by the American Merchant’s pilot, it would need to have occurred in an ‘extraordinary’ location beyond the buoyed channel. Furthermore, the ships could only have reached this location if Matatua had traversed the last four miles at over twice the speed of American Merchant. This absurdity ‘throws an obvious reflection on the whole of the evidence which he gives…’
Matatua’s pilot was found to be more credible. He was the only one familiar with the river, and the judge believed him, consistent with the advice of the nautical assessors. American Merchant was held 100% to blame.
The great 1920s Court of Appeal
An appeal was heard by Lord Justices Scrutton, Bankes and Atkin; a combination widely considered to comprise the greatest Court of Appeal of the 20th century. Allegedly, the Admiralty President had failed to adequately heed the gyrocompass evidence.
Lord Justice Bankes gave the leading judgment in what he termed ‘the “Man versus machine” case’.
American Merchant’s successful transatlantic crossing was considered to be strong prima facie evidence of the record card’s correctness, and the Royal Navy’s use of gyrocompasses was indicative of their reliability. However, the expert evidence acknowledged that experienced gyrocompass users habitually performed a check against a second gyro or magnetic compass, as it was not unknown for the devices to go awry.
The suggestion that the interlocked ships had drifted through 180 degrees in 10 minutes was found to be unrealistic, as such a manoeuvre would require full speed and helm action. It logically followed that American Merchant’s witnesses were ‘telling a story which is not true in order to make it fit in with the gyro’.
Those witnesses were also contradicting the gyro evidence for the period preceding the collision. The record card showed the vessel on a 258-degree heading only briefly, whereas American Merchant’s witnesses claimed that heading had been maintained for 12 minutes and 30 seconds.
There were accordingly no grounds for reversing the Admiralty President’s judgment, as the gyrocompass evidence was not only ‘inconsistent with the defendants’ case but it is quite obvious that there is something wrong with the gyro...’
Appeal to the House of Lords
A further appeal to the House of Lords was also dismissed. Lord Sumner was clear that:
‘the gyro-card is only one piece of evidence in this case, peculiar, indeed, in being mechanically produced and incapable of being interrogated, but in its own way untrustworthy and obscure at times. No superiority in kind can be claimed for it from the point of view of a judge, who has to test its value as against the evidence of human witnesses in the case.’
Lord Buckmaster acknowledged that it was difficult to test its value when the experts had not been adequately questioned on the causes ‘of the aberrations to which the compass is occasionally liable’.
Nevertheless, the record showed the locked-together vessels turning under ‘[t]he momentum of the collision, now increasing, now diminishing, now increasing again the rate of swing…’ This was ‘very odd’ and enough to render the gyrocompass evidence ‘valueless’. Lord Sumner would ‘not accept the view that the Matatua must fail unless her witnesses connect the wandering, of which the card contains such signal proof, with some physical explanation which would trace the aberrations in question to some particular source.’
Legacy
American Merchant may stand as authority for the onus of proving the fallibility of machine-made evidence. Provided a party can logically show the machine to be in error, they should not have to prove the technicalities of the malfunction.
As we embark on an age of artificial intelligence, including the creation of the very first synthetic biological intelligence, powered by cultivated human brain neurons, it is hoped members of the judiciary will not have cause to again refer to machinery becoming ‘upset’ or ‘deranged.’
Return to news headlines 
