A nautical mystery

About the size of a hockey puck, John Harrison’s H-4 is perhaps the most famous watch ever built. In the 18th century it represented a technological breakthrough as momentous as today’s global positioning system.

By providing mariners with an almost childishly simple means of calculating their longitude, Harrison’s H-4 ushered in the era of modern celestial navigation. It played a key role in opening world markets to British shipping; it served as the model for copies that Capt. Bligh would carry on H.M.S. Bounty and Capt. Cook on his second voyage of exploration to the South Seas.

We know a great deal about the watch itself. However, despite two centuries of comment and conjecture, there is one thing about H-4 that we do not know: how well or how poorly it performed on its first trailblazing test 236 years ago on a voyage from Portsmouth, England, to the West Indies.

The story of H-4 begins early in the 1700s when an English fleet sailed blindly onto the rocky shores of the Scilly Islands with the loss of 2,000 men. Stung by the disaster, Queen Anne’s government offered a £20,000 reward (millions of dollars in today’s money) to anyone who could come up with a feasible method of fixing longitude to within 30 nautical miles. Longitude, of course, is a function of time; it is time expressed east-west in degrees of arc. And on a two- to three-month Atlantic crossing the royal challenge meant calculating time to an error limit of about two minutesan unheard-of achievement. (The latitude issue had already been largely solved. Mariners had long understood that the sun passed over their local meridian once each day. They used this information to calculate latitude.)

John Harrison, an obscure clockmaker living in the north of England, was a self-taught mechanical wizard. He had already worked out principles of temperature compensation based on the unequal contraction of two metals, brass and steel. He had fashioned a gridiron pendulum whose length did not change with a rise or fall of the thermometer. Harrison’s clocks attained astonishing accuracy, losing or gaining no more than a second a month. In the 1730s he adapted his ideas to a trio of spring-driven but unwieldy and sometimes erratic sea clocks. Ultimately he incorporated the fruits of his tinkering into a watch that combined the twin virtues of reliability and portability. It was this watch, H-4 (now on permanent display along with its predecessors at the National Maritime Museum in Greenwich, England), that we identify as the first practical sea-going chronometer.

Set in a silver case, H-4 was a mechanism of extraordinary design. It ran on jeweled bearings and had an early example of a sweep second hand. It had maintaining power to prevent the stopping of the train when the watch was being wound. Most important, it had a bimetallic brake that checked the wildest fluctuations of timekeeping caused by temperature shifts.

On November 18, 1761, confined to the care of Harrison’s son, William, H-4 left Portsmouth on an 81-day journey to the English naval base at Port Royal, Jamaica. The question being tested was this: could a shipmaster successfully carry the time of a reference meridian with him on an extended voyage? The theory of longitude was no novelty. Its practical solution was another matter. Mariners knew that the Earth rotates on its axis at the rate of 15° an hour, or one degree every four minutes. To determine longitude they had to be able to compare ship’s time with the time at a reference location like their home port. They could observe, for example, the ship’s time of Jupiter’s eclipsing moons or of Mercury’s transit across the sun. And by consulting a table that predicted the time of the same event at, perhaps, London or Portsmouth, the 18th-century navigator could calculate longitude east or west of these reference cities by converting the difference in time to degrees and minutes of arc.

Practical difficulties

But the problem was obvious: how does one peer through a telescope from the deck of a heaving vessel? The path of Earth’s moon against a backdrop of stars also could be used as a celestial clock, but the math needed to work through the problem lay beyond the reach of the average seaman.

Thus, masters and mates did what they had always done: they judged east-west distance by estimates of course and speed, set and drift. Frequently, this was no better than guesswork. Until the navigator could carry his reference time as easily as he did his charts and sextant, the longitude problem would remain a closed book. Eventually H-4 would open it. On its first sea trial, however, it did nothing of the sort.

Most historians of celestial navigation contend that Harrison’s watch accumulated a negligible error at Port Royal. The figure usually cited is 5.1 seconds slow. There is no good evidence to support this claim. Indeed, other commentators have expressed doubts. In his classic work on timekeeping at sea, The Marine Chronometer (1923), Rupert T. Gould argued that H-4’s supposed error could only have been valid “if the situation”meaning the longitude”at Jamaica was correctly determined.” The 11th edition of Encyclopedia Britannica (1910-11) was less equivocal: it dismissed as fiction the Port Royal longitude used in the test.

Two main objections emerged from the Jamaica trial. One, as we can see, was the longitude. The mean time at a reference meridian can only be found from the local mean time by adding or subtracting the longitude in time. Thus to calculate H-4’s error (unknown) at Port Royal the problem had to be worked this way: local mean time at Port Royal added to a known Jamaica-Portsmouth longitude difference, read as time, would have tallied with the correct reference time at Portsmouth, where the watch had been set. The other difficulty was H-4’s assumed rate, or average daily error.

My own interest in the test grew out of a casual examination one day of the Jamaica meridians inscribed on an antique chart that I own. Printed in 1762 for the London mapmaker John Bowles, the chart underscores Port Royal as a position whose longitude had actually been “observed.” But this was not the case. The chart puts Port Royal 76° 30′ west of London (St. Paul’s Cathedral). Adjusted for the zero, or prime, meridian (Greenwich), a few miles east of St. Paul’s, the reckoning becomes 76° 36′ W. The modern determination of Port Royal is 76° 51′ W (Bowditch). The Bowles chart errs by 15′, or 60 seconds of time. A reading of 18th-century seamen’s manuals turns up similar discrepancies. John Robertson’s Elements of Navigation (1764) shifts Port Royal’s longitudeadjusted for Greenwicheast by 32 seconds of time. Andrew Wakely’s The Mariner’s Compass Rectified (1761) and Nathaniel Colson’s The Mariner’s New Calendar (1762) both stray by 52 seconds of time.

The sole mention of actual longitude now thought to have been employed by William Harrison can be found in a 46-page tract published anonymously in London a year after the Jamaica trial. The author almost certainly was the astronomer James Short, a supporter of John Harrison. Only a handful of copies are known to exist: the one I used I unearthed in the Boston Athenaeum Library. The pamphlet’s orotund titletypical of the ageis “An Account of the Proceedings in Order to the Discovery of the Longitude.” In it, Short fixed Port Royal’s longitude from a transit of Mercury observed simultaneously at London and Jamaica. His determination, adjusted for Greenwich, is 76° 45.5′ W, off by 22 seconds of time.

Longitude difference

Of course, in the context of the Jamaica trial the important figure was the longitude difference between Portsmouth (west of Greenwich) and Port Royal, since H-4 carried Portsmouth time, not GMT. Short reckoned the longitude difference as 75° 42.8′. The modern determination of this difference is 75° 45′. Short’s Jamaica longitude errs to the east by nine seconds of time. This is additive. Thus, on its face, the historical record of 5.1 seconds slow for H-4 cannot stand. It moves at best to 14.1 seconds slow. I say “at best,” for now the issue of rate comes into play. And here Short stands on even flimsier ground. The rate he applies to the watch does not bear close scrutiny.

As Short notes, William Harrison synchronized H-4 to local mean time at Portsmouth by the astronomical method known as equal altitudes. According to Derek Howse, former curator of navigation at the National Maritime Museum in Greenwich, a portable astronomical quadrant would have been used to measure identical altitudes of the sun before and after noon. Half the elapsed time (corrected for declination change and for the equation of time) added to the time of the first observation would have given the time of mean noon at the Portsmouth meridian. The younger Harrison and an astronomer named John Robison arrived with the watch at Port Royal aboard H.M.S. Deptford on January 19, 1762. A few days later, the two men ran another set of equal altitudes to establish their localPort Royalcomparison time.

Short, in the “Account,” reports that at Portsmouth the watch lost 24 seconds between sets of equal altitudes taken nine days apart. The arithmetic is straightforward: it gives an assumed daily rate of 2.66 seconds. But the concept if rate was imperfectly understood in 18th-century England. For one thing, there was no workup of the mean deviation from daily rate, a key ingredient of any chronometer test. For another, both the Harrisons, father and son, failed to declare any rate to officials overseeing the Jamaica trial. H-4’s purported rate was computed after the fact and made public for the first time in the Account of the Proceedings. By Short’s own estimate, the watch showed a total error of greater than three minutes at Port Royal. To win the longitude prize it would have needed to perform in the two-minute range (30 nautical miles at Jamaica’s latitude). It was only after applying the assumed rate of 2.66 seconds a day over 81 days that Short arrived at an acceptable error for the watch of 5.1 seconds slow (discounting the skewed longitude). The trial overseers rejected the claim.

In hindsight, there is small likelihood that an ungimballed watch, subject to the shocks and temperature changes of an ocean voyage, could have sustained a perfect, or unchanging, rate of loss or gain. A hundred years after Port Royal, an English astronomer called Hartnup worked out an elaborate scheme of temperature corrections for marine chronometers. In our own century, the U.S. Navy provided its ships with a form for predicting temperature error. Shortly after the Jamaica trial, John Harrison himself calculated a sliding scale of temperature corrections that he applied, successfully, to the second test of H-4, in 1764. This test resulted in a longitude reckoning that came within 10 nautical miles of the target figure at Barbadosthree times better than the accuracy needed to qualify the watch for the £20,000 prize.

Still, that was 1764. Jonathan Betts, curator of horology at the National Maritime Museum, notes, with justice, the “remarkable achievement” represented by H-4. But in 1762 a rate for the watch had not been established. And for want of a rate, he adds, the Jamaica trial was “next to useless.”

By Ocean Navigator