The Stevensons

Three towers in How Lighthouses Work 6 had one name running through them: Stevenson. Robert built the Bell Rock, his son Alan built Skerryvore, and his sons David and Thomas built Muckle Flugga. This chapter steps back from the stone to the people. Who were they, how did one family come to supply the engineers of a whole country's lights for most of a century, how did one of them try to measure what the sea actually does to a tower, and what did the writer in the family make of it all? It is a story about engineering, but also about a body, the Northern Lighthouse Board, that kept the work going from one generation to the next.

Four Engineers, One Board

The Northern Lighthouse Board was set up by the Erection of Lighthouses Act 1786, which the Board's page dates to royal assent on 27 June 1786 (its infobox says 1867, which conflicts with the Act, and I have taken the Act). It is the general lighthouse authority for Scotland and the Isle of Man; its headquarters is a Georgian building in George Street, Edinburgh, with technical operations in Oban, and as of 31 March 2024 it was responsible for 208 statutory lighthouses (65 with a range over 15 nautical miles and 143 under, which does add up). Its first four lights were Kinnaird Head, North Ronaldsay, Scalpay and the Mull of Kintyre. The same page calls Robert Stevenson the commissioners' most famous engineer and says his sons David, Alan and Thomas followed him and that the family built most of the northern lights. It does not give the end of the family's tenure.

The family's link with the Board starts one step before Robert. His father, a West Indies sugar merchant, died in 1774, and his mother's second husband was Thomas Smith, a tinsmith, mechanic and civil engineer who joined the Board in 1786. Robert trained under Smith, married Smith's daughter Jean (so his stepsister) in 1799, and was appointed engineer to the Board in 1797, succeeding Smith. The code below records the dates the pages give, works out what can be worked out, and then follows the family tree from three younger members back to Smith.

The first block gives the lifespans (78, 58, 71 and 69 years) and the lighthouse rates. Alan's 13 lights in the 11 years from 1843 to 1853 is 1.18 a year; David's 28 listed lights from 1854 to 1880 is 1.04; Thomas's 31 from 1854 to 1886 is 0.94. These are crude: David and Thomas designed many lights together, so the same lighthouse appears on both lists and the counts must not be added; the pages differ over their own totals (Alan's page says 13 but lists 14 entries, counting Hoy High and Hoy Low as one; David's lead says "over 30" but lists 28); and Robert's page lists about 22 lights, but the extract I was given did not show the full span, so I left his rate out. Robert's 1797 to 1842 is 45 years, though the page calls it "nearly fifty", a conflict inside one page. Alan's term as Engineer in Chief begins in 1843, so there is no gap after Robert's. From Robert's appointment in 1797 to Thomas's last lighthouse (Ailsa Craig, 1886) is 89 years. Whether the family held the engineering post for all of those years is not something I can say: the pages give Robert's and Alan's years, but only say that David moved to the Board in 1853 and that Thomas designed from 1854, without giving the length of either man's service. The last three lines show the family tree: Robert Louis Stevenson, David Alan and Katharine all trace to Robert and, through him, to Smith.

What Each of Them Did

Robert Stevenson's page credits him with the movable jib and balance crane, with the use of Fresnel lenses (see How Lighthouses Work 4), with reflector design, and with rotating and shuttering systems that gave lights their distinct characters (How Lighthouses Work 5); for the last he received a gold medal from William I of the Netherlands. He was elected a Fellow of the Royal Society of Edinburgh in 1815. His page records the Bell Rock dispute that How Lighthouses Work 6 mentioned: John Rennie was a consulting engineer and Samuel Smiles's account favoured Rennie, but the Board credited Stevenson, and the page says historians have followed the Board. I give the dispute as the page states it and have not looked at the underlying sources.

Alan, the eldest son, was elected a Fellow of the Royal Society of Edinburgh in 1838 and was given an honorary law degree by the University of Glasgow in 1840. His page says he was Engineer in Chief from 1843 to 1853 and built 13 lighthouses in that time; the Skerryvore page adds that Thomas supervised the interior fitting out in 1843. Neither page gives a reason for Alan leaving the post in 1853, and his own page says nothing about lenses; that is a gap in the page, not evidence that he did no optical work. David joined the Board in 1853 and, with Thomas, designed lighthouses from 1854 to 1880; his page also says he helped Richard Henry Brunton with lighthouses in Japan, including a method to withstand earthquakes, and that his son David Alan built nearly thirty more.

Thomas is the one whose range goes beyond lighthouses. His page calls him a civil engineer, lighthouse designer and meteorologist. He co-founded the Scottish Meteorological Society, devised the Stevenson screen for sheltering weather instruments (the page gives no date), published a wave-height paper in 1852 and The Design and Construction of Harbours in 1864, and in 1869 ran an electric light experiment at Granton Harbour, with a light on the Trinity Chain Pier controlled from half a mile away. He was president of the Royal Society of Edinburgh from 1884 to 1886. Harbours were not always kind to him: construction of the Wick breakwater began in April 1863 and, according to the page, was abandoned in 1877 after storm damage.

Measuring the Sea: the Wave Dynamometer

In How Lighthouses Work 6 the Skerryvore page's figure of 6,000 lb per square foot raised a question: where did a number like that come from, and does the sea really push that hard? The answer is an instrument. Alan's 1848 account says that Thomas had "lately contrived an instrument for the purpose" of registering the force of wave impulses and had applied it at various parts of the coast; the reading tool reported that the account gives his paper in the Transactions of the Royal Society of Edinburgh as 20 January 1845. The Trinity House article calls it the marine dynamometer, built to record impact pressures, deployed on the Bell Rock and Skerryvore, and notes that in 1848 Alan wrote that no systematic attempt had been made to measure the force of a wave. The same article calls such measurements rare. It does not describe the instrument's mechanism, and neither did the part of the account I read, so I will not describe one. A search summary said Thomas collected the first systematic wave-impact measurements over some 23 months with three of the instruments, but I did not find that in the pages I read, so treat it as unconfirmed.

The readings I did find, all in pounds per square foot, are in the account: a best result at Skerryvore of 4,335 "as yet", a Bell Rock best of 3,013 (described as "one-fourth part" less), and average readings of 611 for the summer months and 2,086 for the winter months of 1843 and 1844. The account says the experiments "have not been continued long enough" to be fully useful. The Skerryvore page says the readings implied the tower had to resist 6,000 pounds per square foot (about 300 kPa), which is more than the 4,335 in the 1848 text. A search result quoted a peak of 6,083 from a travel-company article that I could not check. So I have three possible numbers for the Skerryvore peak (4,335, 6,000 and 6,083) and I do not know whether the larger ones come from later readings or from a design allowance; the sources I read do not say. The code converts them all.

The 6,000 figure is 287.3 kPa, the same as How Lighthouses Work 6 found, and corresponds to a head of 28.6 m of seawater. The best Skerryvore reading of 4,335 is 207.6 kPa (20.6 m of head), the Bell Rock reading is 144.3 kPa, and the winter average of 2,086 is about 99.9 kPa, close to one atmosphere (101.3 kPa), while the summer average of 611 is only 29.3 kPa. So the winter average was 3.41 times the summer one. Two checks on the text: 3,013 is 30.5 per cent below 4,335, not the 25 per cent that "one-fourth less" suggests (a quarter off 4,335 would be about 3,251), which may be rounding, a different pair of readings or a misprint; I cannot tell which. And the 6,000 figure is 1.38 times the best 1848 reading, which would be a sensible margin if it was a design allowance (my reading, not stated), and 2.88 times the winter average. If 6,083 is a later reading, it exceeds 6,000 by only 1.38 per cent, so the two are consistent with one another.

Waves and the Wave-Height Rule

Thomas's 1852 paper concerned how big waves get. His page gives the rule H = 1.5 times the square root of F, where H is the wave height in feet and F the fetch (the distance of open water the wind blows over) in miles, and says his observations were for fetches under 100 kilometres. Elsewhere the same page says "around 100 metres", which looks like a slip for kilometres (my reading). I give the rule as the page does; the full published formula may include other terms, and I did not check it. The code tries the rule at several fetches and runs it backwards: how long a fetch would give an 8 or 10 m wave, which is the size the page quotes for the Wick estimate? I did not confirm how that estimate was reached.

A fetch of 1 mile gives 1.5 ft (0.5 m), 10 miles 4.7 ft, 50 miles 10.6 ft and 100 miles 15 ft (4.6 m). Because of the square root, quadrupling the fetch only doubles the height; the last lines show that, with a ratio of 2.0 each time. Inside the range of the observations, 100 km (62.1 miles) gives 11.8 ft, about 3.6 m. To reach 8 m the rule needs a fetch of 306 miles (493 km), and 10 m needs 478 miles (770 km), far outside the 100 km. So either the Wick estimate came from a different method, or it extends the rule beyond what it was fitted to; I do not know which, and I would not trust the rule that far. Note also that this rule is about the height of the waves, not the impact pressure of a breaking wave on a tower, which is what the dynamometer was built to record.

The Family Link to Robert Louis Stevenson

The most famous Stevenson is not an engineer. Robert Louis Stevenson was Thomas's son and Robert's grandson, and his page names Alan and David as his uncles. He named his Bournemouth house Skerryvore after the lighthouse that his uncle Alan built. The next table, with ages worked out from the page's remark that he qualified for the bar in July 1875, aged 24, shows how little time he spent in the family trade.

His page says he got more material for writing from the lighthouse trips than interest in the engineering, that his father was disappointed, and that his mother said he was "wonderfully resigned". In 1887, the year his father died, his collection Underwoods reflects on turning away from the family profession. Exercise 1 works out the ages. The page does not mention the wave dynamometer, and I found no source that links the writer's work to the instrument, so I will not suggest a link.

Where the Pages Disagree

The family is well documented, but the pages I read contradict themselves in several places. These are the ones I noticed: the Board's formation (an Act of 1786, against 1867 in its infobox); Robert's service (1797 to 1842 is 45 years, called "nearly fifty"); Bell Rock built 1807 to 1810 but listed as 1811, the year it was first lit; Dubh Artach listed on Robert's page as a design of 1872, 22 years after his death; the number of Alan's lighthouses (13 against 14 entries) and David's ("over 30" against 28); a son of Alan dated 1827 although the marriage was in 1844; the height of Muckle Flugga; and the Skerryvore peak pressure (4,335, 6,000 or 6,083). I have no way to resolve these from the pages alone.