Lamps and Fuel

How Lighthouses Work 2 ended on a gap: 24 candles on a tower whose horizon was over 14 nautical miles shone only about 3.5 nautical miles. The tower was good enough and the light was not. This chapter follows the light itself, from the open flame to the Argand lamp, through the oils and the paraffin burner, to the electric arc, the filament bulb and the LED. Each step is a bigger number of candela, or a smaller bill for fuel, or a longer-lived lamp, and each can be put through the range calculation from How Lighthouses Work 1.

The Problem: A Flame Is a Dim, Hungry Light

A flame makes light from burning fuel, and the numbers are poor. The Wikipedia article on luminous efficacy says a candle gives about 0.1 to 0.2 lumens for every watt of heat it burns, and a 2800 K tungsten filament about 15. A lumen is the light flowing out of a source, a candela is how much of it goes in one direction, and a source shining evenly in all directions gives 4π (12.57) lumens per candela. The Wikipedia article on the kerosene lamp uses the same conversion, 12.57 lumens per candlepower. That is where this chapter's first assumption comes from.

A better lamp therefore does one of two things: it packs more light into one flame, so that a lighthouse needs fewer burners, or it burns fuel more completely, so that the nightly bill falls. The history below is a series of attempts at one or the other.

The Argand Lamp: A Flame With Its Own Chimney

The Argand lamp was invented by Aimé Argand. The Wikipedia article on the lamp gives 1780; the article on lighthouses gives 1782, and says the lamp was first produced by Matthew Boulton with Argand in 1784. Those sources disagree, so I give the date as 1780 to 1782 and rely on neither. The design is the same in both: a wick shaped like a sleeve, so that air passes through the middle as well as around the outside, and a glass chimney that steadies the flame and improves the airflow. Fuel runs by gravity from a reservoir above the burner. The lighthouse article says it "became the standard for lighthouses for over a century."

The lamp article gives the output as 6 to 10 candela, and quotes Thomas Jefferson calling it "a light equal to six or eight candles". The fuels named are whale oil, seal oil, colza, olive oil or another vegetable oil. Colza is rapeseed oil: the article on rapeseed oil says it was used with the Argand burner because it was cheaper than whale oil. The sources I read give no prices, so I do not put a number on "cheaper". What does the better burner do for a lantern?

The columns are the source, its intensity in candela and its nominal range in nautical miles (10 nm visibility); the last line is the range of the 10 cd lamp divided by that of one candle. One Argand burner of 10 cd reaches about 2.6 nm against 1.0 nm for one candle: ten times the intensity gives only about 2.5 times the range, the inverse square law at work again, plus the loss in the air. A lantern that wants the 3.5 nm of 24 candles needs four Argands of 6 cd or three of 10 cd, in place of two dozen candles.

What does a lamp cost to feed?

Fuel use can be estimated from the efficacy and the energy in the oil. Both are assumptions, not measurements. In particular, I have assumed that a wick lamp has the top of the candle range, 0.2 lm/W, which is a guess for an Argand lamp: its steadier flame may do better. The 37 MJ per kg for oil is general knowledge, not read from a source. Treat the result as an order of magnitude.

The columns are the lamp, its lumens, the heat power in watts, and the oil burned in a 12-hour night. On these assumptions a single 8 cd Argand burns about 0.6 kg of oil a night and is a heat source of about 500 W, against 63 W for a candle. Since the model uses the same efficacy for both, the fuel simply follows the light: one Argand does the work of eight candles for about eight times the fuel (0.59 against 0.07 kg). In this model a better burner saves wicks and tending, not oil. If a real Argand is more efficient than a candle, as it may well be, the model understates the saving; I have no source for the true figure.

Oils and Paraffin

The Argand burner is a burner, and the fuel is a separate question. Whale oil, seal oil, colza and olive oil all burned in it (lighthouse article). Kerosene, also called paraffin, is the fuel that followed. The Wikipedia article on the kerosene lamp says Abraham Gesner invented a substitute for whale oil in 1846, distilled from coal, and that Ignacy Łukasiewicz built the most modern and popular versions of the paraffin lamp in Lemberg in 1853. The lighthouse article says kerosene became popular in the 1870s, and that electricity and acetylene gas began replacing it "around the turn of the 20th century". The kerosene lamp article also says a centre-draft round-wick lamp gives "three to four times the output" of a flat-wick lamp, which is the Argand idea once more.

The big jump for lighthouses was the paraffin vapour burner. According to the lighthouse article it was invented in 1901 by Arthur Kitson and improved by David Hood at Trinity House, and gave "an output of over six times the luminosity of traditional oil lights". The article does not describe how the burner works, so I do not either. An 85 mm Chance Brothers unit made around 1914 for Sumburgh Head was used until 1976. How much does six times the luminosity buy?

The columns are the intensity factor, the nominal range and how many times the starting range that is. The starting light of 1,000 cd is an assumption. The result is the point of the whole chapter: a light six times as bright reaches only 1.4 times as far, and a light a hundred times as bright only about 2.1 times. A claim of "six times the luminosity" is a claim about candela, not about miles. Adding lamps has steeply diminishing returns, which is why How Lighthouses Work 4 turns to lenses, which gather the light that a lamp already makes and send it where the sailor is.

The Electric Arc

The carbon arc lamp makes light from an electric spark between two carbon rods. The Wikipedia article on arc lamps says Humphry Davy demonstrated it in the early 1800s, with sources citing 1802, 1805, 1807 or 1809, so the date is not settled. Arc lamps need an electrical supply, which for a lighthouse in the 1800s meant its own generator (the lighthouse articles below mention steam-driven ones).

The lighthouse sources disagree about the first use, and I list the claims side by side rather than choose between them:

The 1875 date in the lighthouse article may be the date of the illustration that the South Foreland article mentions, but I have not checked that. The 1858 trial, the 1862 Dungeness installation and the 1871 Souter lighting come from pages that agree with each other, so those are the ones I would quote. The Souter article gives the light as a carbon arc of "800,000 candle power", visible up to 26 miles (it does not say statute or nautical). That intensity is the source's claim; I have not found how it was measured, and it may include the effect of the lens. What range does it predict?

The first two lines are the luminous range at 10 and 20 nm visibility; the last three are the geographic range for three assumed light heights, with an eye 5 m up. At 10 nm visibility an 800,000 cd light has a luminous range of 25.1 nm, close to the 26 miles in the article, which is reassuring but not a proof, as the units are uncertain. The real lesson is the second half. A light this bright is limited by the horizon, not by the light: for a tower of 20 to 60 m (assumed) the geographic range is only 14.0 to 20.8 nm, however clear the night. Beyond a point, more candela does nothing and the next improvement must come from height. The candles of How Lighthouses Work 2 were limited by the light; the arc is limited by the earth.

Bulbs and LEDs

Filament bulbs followed. The Wikipedia article on the incandescent light bulb gives Swan's working demonstration on 17 January 1879 and Edison's first successful test on 22 October 1879. It gives typical efficacies of 16 lm/W for a 120 V bulb and 24 lm/W for quartz halogen, and a theoretical maximum of 52 lm/W for a tungsten filament. The luminous efficacy article gives about 15 lm/W for a typical 2800 K tungsten bulb, 17 to 20 for halogen, and about 102 lm/W for LED screw-base lamps (up to 172 for some tubes). The lighthouse article gives a filament lifetime of "about 1,000 hours" and an LED lifetime of "50,000 to 100,000 hours", says LEDs were widely used by 2020, and that a typical LED system costs about 20,000 euros according to a supplier. Here is the comparison laid out:

The halogen figures differ slightly between the two articles (17 to 20 against 16 or 24); I use 20 below. The bulb article's 1,000 to 2,000 hours is for general-service lamps, so the 1,000 hours I use is the cautious end, and I have applied the filament lifetime to halogen as well, which is an assumption. Now put those numbers in a lamp of 1,500 lumens running every night:

The columns are the source, the power it draws, the electricity used in a year and the number of lamp changes a year. For the same 1,500 lumens the filament draws 100 W and the LED about 15 W, and the energy falls from 438 to 64 kWh a year. The bigger change for a lighthouse is the last column: 4.4 filament changes a year against one LED change in about eleven years (0.09 a year). A light that needs a bulb changed every few months needs someone nearby; one that runs for years does not. That is one reason the unmanned light became practical, a story that How Lighthouses Work 8 tells.