Fog and Radio

Everything in How Lighthouses Work 1 to 7 assumed that the sailor could see the light. Fog takes that away. A lamp with a range of 20 nautical miles is no use when the visibility is a few hundred metres, and the rocks that the Stevensons built on do not go away because the weather has closed in. This chapter is about what lighthouse authorities did instead: they made noise, then they used radio, and in the end they took the people out of the towers altogether. We will look at fog bells, guns, horns and sirens, at how sound behaves in air, at the radar beacon called a racon, at lightvessels, and at the automation that ended the keepers.

Noise When the Light Fails

The foghorn page opens with the job: a device that uses sound to warn vehicles of navigational hazards in fog, for when "visual navigation aids such as lighthouses are obscured". Its history section says audible fog signals have been used "for hundreds of years", at first seashell horns, fog bells or gongs struck by hand. The fog bell page dates the first bell made for the purpose to 1766 at Nidingen on the Baltic Sea. Bells were rung by hand, which was a chore in a long fog, or worked by clockwork that had to be wound, as in the clockwork of How Lighthouses Work 5. An American, Andrew Morse Jr., made a "perpetual fog bell" used in 1839 at Penobscot Bay, driven by a long beam that rose and fell with the waves; the page adds that the float was quickly carried out to sea. The same page says there were 93 fog bells in the United States in 1878, and that bells had a short range, especially in wind. In England most mechanical fog bells were taken out of use around 1905.

Some places used guns. The foghorn page says a small cannon was fired at some lighthouses, but it was labour-intensive and dangerous, and that Trinity House, the English authority, did not use steam whistles and preferred an explosive signal. It later developed a system for firing a gun-cotton charge electrically, though the charge had to be replaced by hand each time. At Portland Bill, with a signal every five minutes, that meant lowering the horns, inserting two new charges and raising them again every five minutes in fog. In Scotland the Skerryvore page says a fog bell had been installed by the time of the keeper James Tomison, who joined in 1861, and calls it one of only two operating in Scotland at the time. The Bell Rock page records a much older legend: that a warning bell was fixed on the rock in the 14th century by the Abbot of Arbroath and stolen a year later by a pirate, a story told in Robert Southey's poem. I give it as legend because the page does.

Then the mechanics improved. The foghorn page says the first automated steam-powered foghorn was invented by Robert Foulis, a Scot who emigrated to Saint John, New Brunswick, and was built at Partridge Island in 1859 (as the "Vernon-Smith horn", because the plans were passed on under another man's name; Foulis was later credited). He is said to have noticed on a foggy night that low notes carried better than high ones, so the device made a low note and a coded pattern. The page adds that a Captain Newton also claimed the idea of loud, low-pitched signals, so the credit is disputed. In the United States Celadon Daboll built a coal-powered trumpet, which was not universally adopted. In Britain John Tyndall and Lord Rayleigh ran experiments, and Rayleigh's work for Trinity House led to a siren with a large trumpet installed at Trevose Head in 1913. The siren page gives an earlier line: sirens with slotted cylinders were developed in 1877 to 1880 by James Douglass and George Slight of Trinity House, and the final version was first installed in 1887 at the Ailsa Craig lighthouse. Ailsa Craig was also where Thomas Stevenson's last lighthouse (1886) was, in How Lighthouses Work 7, but the pages do not say the two installations are the same, so I only note the shared name. From the early twentieth century, the foghorn page says, the diaphone, an organ stop invented by Robert Hope-Jones and developed as a fog signal by John Northey of Toronto, became the standard apparatus for new installations; it ran on compressed air and made very powerful low notes.

Sound in Air

Light covers 20 miles in a fraction of a millisecond. Sound is a different thing altogether. The speed of sound page says that at 20 °C it is about 343 m/s in air, about 331 m/s in dry air at 0 °C, and 1,481 m/s in fresh water at 20 °C, a little over four times faster. In still air it depends mostly on temperature, and the page adds that humidity raises it by 0.1% to 0.6%. History is on the page too. Newton's 1687 figure of 979 ft/s is described as too low by about 15%, an error the page ascribes to treating the compression of air as isothermal instead of adiabatic, and the William Derham who in 1709 timed gunshots from a church tower wrongly concluded that rain and fog slow sound down, a conclusion accepted until Tyndall disproved it. That last point matters here: fog does not slow the signal. What the page does describe is refraction: because temperature falls with height, sound is bent upward, away from listeners on the ground, which leaves an acoustic shadow at some distance from the source.

The ideal-gas formula for the speed of sound is the square root of (adiabatic index times gas constant times absolute temperature, divided by molar mass). The page gives the adiabatic index as about 1.4 for air. I supplied the gas constant and the molar mass of dry air from memory, so they are my assumptions. The code runs the model for five temperatures and then uses it to answer two practical questions: how long a signal takes to reach a ship, and how far apart a sound travelling through the air and one travelling through the water would arrive.

Exercise 2 puts Newton's 979 ft/s (298 m/s) 13.1% below the model's 20 °C value, where the page says about 15%; the page does not say what temperature that figure refers to, so I cannot say whether the difference matters. The model gives 331.3 m/s at 0 °C (the page says about 331) and 343.2 m/s at 20 °C (the page says about 343), so it agrees with the two values I could check. The ratio of the water speed to the air speed at 20 °C is 4.31, matching "a little over four times". A signal 5 km away takes about 14.8 s at 10 °C, and a ship 10 km away hears the horn almost 30 s after it started. The foghorn page says each horn was given its own timing characteristics, set by a clockwork coder, so that mariners could identify it, much as lights have their characters. Over 1 km at 20 °C, a sound in water would arrive 2.24 s before the same sound in air. That is the idea behind the submarine bells I describe below: a ship that can listen to both gets a time difference to turn into a distance. The page on fog bells says the bells were later replaced by electric transducers, and that the underwater work led, through Reginald Fessenden, to sonar. I have not checked that history beyond the page.

Lightvessels

Where a rock or sandbank is too deep or too soft for a tower, the light goes on a ship. The lightvessel page opens by saying the first modern lightship was invented by Robert Hamblin in 1734 and was off the Nore sandbank in the Thames estuary. Its section on British lightships says David Avery and Robert Hamblin put the earliest one there in 1731, a private venture that made a profit without the government enforcing payment. That is a conflict inside one page (1734 or 1731), and I cannot resolve it from the page. Other vessels followed off Norfolk (1736), Owers Bank (1788) and the Goodwin Sands (1793), and Trinity House later displaced the private operators. The first United States lightship was established in Chesapeake Bay in 1820.

The Scottish connection is here too. A converted fishing boat renamed Pharos, meaning lighthouse, entered service on 15 September 1807 "near Inchcape, Scotland". The Bell Rock page says that Bell Rock is also known as Inchcape, so I take this to be the rock Robert Stevenson was building on. That would put a lightvessel on station four years before the tower's light of 1811, although the pages do not say that the vessel was there for the building work. Its anchor, the first of the mushroom type on a lightvessel, weighed 1.5 tons, and the page gives a rule of thumb for the chain: 6 feet for every foot of water. Cast-iron chain from the 1820s made the mooring far more effective. Lightvessels also carried bells. In the early twentieth century, tests by Trinity House found that a bell lowered about 18 feet could be heard at 15 miles, but that the practical range in operating conditions was 1 to 3 miles. The code below checks the list of Trinity House hulls on the page against the count the same page states, and works out the chain and the bell ranges.

The page lists seven solar lightvessels (hull numbers 2, 5, 6, 7, 9, 10 and 17) and four of the larger "20 class" (19, 22, 23 and 25), which is 11 lightvessels, plus two light floats (LF2 and LF3). The page's own text says there are "currently eight unmanned lightvessels and two smaller light floats". The two counts disagree and I cannot tell which is out of date; the page does not say whether some of the numbered hulls are spare or retired. The chain rule gives 60 m of chain for 10 m of water and 180 m for 30 m. The practical bell range of 1 to 3 miles (1.6 to 4.8 km) is only 7% to 20% of the 15-mile test range, which is a reminder that a figure from a trial in good conditions says little about a bad night. The page also says the crews of lightvessels kept records of passing ships, observed the weather and sometimes carried out rescues, and that the 20 class, which relies on diesel-electric generators, is used where a main light with a range beyond 20 nautical miles is needed, because a solar lightvessel's main light reaches at most 19.

Radar and the Racon

Radar made a new kind of fog signal possible. The racon, or radar beacon, is defined by the International Telecommunication Union as a transmitter-receiver on a fixed navigational mark which, when triggered by a radar, automatically returns a distinctive signal that appears on the radar display and gives range, bearing and identification. The racon page says it answers on the same frequency with a Morse character, drawn as a short line of dots and dashes radiating away from the beacon on the screen, usually a few nautical miles long. Racons mostly work in the X band (9320 to 9500 MHz) and many also in the S band (2920 to 3100 MHz). A modern one is frequency-agile: it detects the incoming pulse, tunes its transmitter and replies with a 25 microsecond signal within 700 nanoseconds. Older ones sweep slowly across the band over a minute or two and answer only about 5% of the time. The page says the large majority are on buoys rather than lighthouses, and that they mark things such as bridge spans and offshore platforms. The foghorn page reports an older version of the idea: underwater fog bells ringing a coded number such as 23, which it calls "an earlier precursor to RACON".

A racon must not hide real targets behind it, so it is silent for part of the time. The page gives the UK duty cycle as about 30%, usually 20 s on and 40 s off, or sometimes 9 s on and 21 s off (as on the Sevenstones lightship). The code turns the speed of light into timing, to show what these microsecond numbers mean on the screen. I use the vacuum speed of light, which is very slightly above the speed in air, so the numbers are close but not exact.

One nautical mile of range costs 12.4 microseconds of echo delay, there and back, so a target at 12 nm returns after 148.3 microseconds. A reply that lasts 25 microseconds therefore smears over about 3,747 m, or 2.02 nm, of range on the display, which fits the page's "few nautical miles". The 700 ns response time is worth about 105 m, so even a quick racon puts its mark slightly beyond the true position. The duty cycles are 33.3% for 20 s on, 40 s off, a little over the page's "about 30%", and exactly 30.0% for 9 s on, 21 s off; the American battery buoy cycle of 20 s on, 20 s off is 50%. I did not check the 33.3% against the page's "about"; it is simply the arithmetic of the figures.

Automation and the End of the Keepers

The foghorn page says that once automation became common in the 1960s and 1970s most older foghorns were removed rather than keep their complex machinery running, and replaced by electric diaphragm or compressed-air horns. Activation is automatic: a laser or photo beam is sent out to sea, and if it is reflected back because of fog or rain, a sensor starts the horn. The page adds that modern aids including GPS have in many cases made large long-range foghorns unnecessary, according to the International Association of Lighthouse Authorities. The lighthouse page makes the same point for the keepers: electrification and automatic lamp changers "began to make lighthouse keepers obsolete". It says that for years lighthouses were still staffed partly because keepers could serve as a rescue service, and that GPS and better navigation brought the phasing out of staffed lights. The lighthouse keeper page lists the other pieces: remote monitoring and automatic bulb changing from the 1960s, and it says that keepers still working today do building maintenance and repair buoys and technical equipment. The lighthouse page also describes how an automated light often replaces the rotating lens (How Lighthouses Work 4 and 5) with a high-intensity flasher that concentrates the light in time rather than direction.

For Scotland the pages give these dates. The Bell Rock page says the light has been automated since 24 October 1988 and that the Northern Lighthouse Board, which has been at 84 George Street in Edinburgh since 1832, monitors it remotely. The Skerryvore page says the lighthouse was automated in 1994, and the Muckle Flugga page says that the lighthouse was automated in 1995, when the last residents moved out. The lighthouse keeper page says that Britain's automation began in the late 1960s but that most lights kept their keepers until the 1980s and 1990s, and that the keepers left the Fair Isle South lighthouse on 31 March 1998, when the last staffed Scottish light became automated. The Northern Lighthouse Board page says it maintained foghorns at a number of locations between 1876 and 2005 and that the last, at Skerryvore, sounded for the final time on 4 October 2005. The two dates are for different things, so they do not conflict: the keepers left Skerryvore in 1994, 11 years before its horn was silenced. Why the horn was kept going so long is something the pages do not explain, so I will not guess.

From first lighting to automation the three towers served for 177 years (Bell Rock, 1811 to 1988), 150 years (Skerryvore, 1844 to 1994) and 137 years (Muckle Flugga, 1858 to 1995). These are span lengths, not records of continuous staffing: Skerryvore, for instance, was dark for five years after a fire in 1954 (that is from the Skerryvore page). The period from the Board's act of 1786 to the last Scottish keepers in 1998 is 212 years, and the last foghorn sounded 2,744 days, a little over seven years and six months, after the last keepers left. Nothing here tells us anything about the people who lived in those years, which is a different story from the dates.