01Executive summary
Six milestones in which electricity acquired its first mass market, and in which the decisive engineering turned out to be system design rather than lamp design.
Davy drew an electric arc in 1801 and it waited sixty years for a power source worth using. Gramme made the generator commercial from 1871. Brush's arc system lit Cleveland in 1879. Edison produced the first incandescent lamp suitable for quantity manufacture in the same year, and opened Pearl Street Station in 1882. Tungsten displaced carbon from 1902, and the fluorescent lamp arrived in 1938 at roughly forty times the efficacy of the earliest carbon filament.
02The arc: a good light with no power behind it
Humphry Davy found in 1801 that a brilliant arc could be drawn between two slightly separated carbon rods in a battery circuit. His battery could not sustain a stable continuous arc, and there the matter rested for six decades. A trial installation went into the Dungeness lighthouse in 1862 and the first practical application followed at Le Havre in 1863, powered by inefficient generators of an early type.
Arc lighting was not commercially feasible until roughly a decade after Le Havre, and what changed was not the lamp. It was that electricity became available far more cheaply than batteries or early generators could supply it. A device waiting on a cheaper input is a recurring situation and it is frequently misdiagnosed as a problem with the device. The correct question when something works and does not sell is which input is too expensive, not which component is inadequate.
Once cheap power existed the arc developed quickly. Gramme designed an alternating-current generator for Jablochkoff's "candle" of 1876 — two carbon rods side by side separated by kaolin that vaporised as they burned — specifically because one rod of a direct-current arc burns faster than the other. The lights appeared on Paris streets in 1878 and were adopted across Europe for their brilliance against gas.
Brush's system, installed in Cleveland in 1879 and New York in 1880, beat it on two engineering points rather than on light output. His lamps burned about twice as long before the carbons needed replacing, and he designed a generator delivering constant current with voltage varying under load, plus an automatic clutch holding the burning rod ends a constant distance apart. Consumables and regulation, not brightness, decided the market.
03Edison: a lamp designed backwards from the system
Edison began work on an incandescent lamp in 1877. He was neither the first to try nor the first to succeed — such lamps had been made as early as 1820, and scores of people including Joseph Swan in England had produced working designs over the intervening half century. What Edison first produced was a design suitable for quantity manufacture and use, which is a different achievement and should be described as such.
His starting point was a market observation rather than a technical one: enough was known about arc lights to make it obvious they were too brilliant for domestic interiors. The target was therefore a softer, less intense light — and the whole design follows from constraints he then chose deliberately.
- Parallel, not seriesHis first 1878 lamps were platinum filaments at 10 volts in series, and proved unreliable. Series is acceptable for street lighting; for general use each lamp must be separately controlled, which requires parallel circuits.
- Higher distribution voltageHe chose 110 volts rather than lower, to reduce transmission losses — a decision with consequences he could not have foreseen, since it fixed a standard still in use.
- The three-wire systemDistributing at 220 volts with each lamp seeing 110, and any unbalanced current returning through a neutral, halves the copper needed for the same delivered power.
- High-resistance filamentHigher voltage across a lamp demands a high-resistance filament. Metal filaments would not serve, so he searched for a material that carbonised well — carbonised thread lasted two days, and split bamboo lasted.
Edison did not invent a lamp and then find a use. He chose a distribution voltage to limit losses, which forced a high-resistance filament, which forced the materials search that is the part everyone remembers. The famous filament hunt is a downstream consequence of a network decision. Designing the system first and letting it dictate component requirements is why his lamp became a product while better lamps did not.
The demonstration of 500 lamps at Menlo Park in 1880 drew crowds large enough that the railway ran special trains. Practical installations came first where a self-contained plant made sense: the arctic expedition steamer Jeannette in 1879, which ran dependably for two years until the vessel was crushed in the ice, and the steamship Columbia in 1880 with a five-candlepower lamp in each stateroom. By 1882 there were over 150 isolated plants in residences, hotels, mills, offices, stores and ships.
04The central station: electricity becomes a utility
An individual steam-driven generator in a cellar is workable and has obvious limitations — somebody must run a boiler in the basement. Generating centrally and distributing by wire is more satisfactory, more economical, and makes electricity available to customers who would never operate a plant.
The first central stations were the small experimental plant of three Brush generators opened by the California Electric Light Company in San Francisco in September 1879, selling electricity for arc lighting, and its larger successor built within a year as demand proved the concept. The Holborn Viaduct station in London followed in January 1882, supplying 3,000 incandescent lamps from Edison dynamos driven by reciprocating engines.
Pearl Street Station in New York opened on 4 September 1882 with six direct-current generators totalling about 900 horsepower — enough for 7,200 lamps at 110 volts — driven by reciprocating engines on Babcock and Wilcox boilers. It served around sixty customers and nearly 1,300 lamps at opening, through underground conduits carrying half-round copper bars in iron tubes, separated by cardboard washers and filled with asphaltum compound for insulation. It burned down on 2 January 1890, by which time the model had spread widely.
The British Electric Lighting Act of 1882, aimed at protecting the gas-lighting monopoly, forbade the construction of large generating stations. It measurably impeded electric lighting in Britain at precisely the moment the central-station model was being established elsewhere. This belongs in an engineering history rather than a political one because the effect was technical: the country that had run one of the first central stations in the world spent the following period unable to build the thing that made them economic. Regulation shapes what gets engineered, and a rule written to protect an incumbent can hold a whole industry back for years.
The limitation that set up the next part
Direct-current stations had a real advantage — storage batteries could provide standby service — and one decisive weakness. Distribution voltage had to be low, because it was the same voltage the lamps used, and there was no way to change it. Low voltage means high current for a given power, and losses rise with the square of current, so a direct-current station can only serve customers within a short radius. That constraint is what the following part is about.
05Fifty years of lamp efficacy
Luminous efficacy, lumens per watt
The step from filament to fluorescent is a change of mechanism rather than an improvement of one. A filament produces light as a by-product of being hot, so most of the input necessarily leaves as heat. A fluorescent lamp excites a coating with ultraviolet from an arc, which does not require the emitter to be incandescent at all. Roughly a factor of forty across sixty years, with about two-thirds of it coming from abandoning the original mechanism.
It is worth noting that gas lighting was still an important source of illumination at the start of the twentieth century and was itself improving in the same period. Drummond's limelight of 1825 and von Welsbach's incandescent mantle of 1885 both work by heating a solid to incandescence in a gas flame — the same principle as a filament, without electricity. The incumbent technology was not static while its replacement developed, which is normally the case and normally forgotten.
06Takeaways for current practice
- When something works and does not sell, find the expensive input. The arc lamp waited sixty years on the cost of electricity, not on the lamp.
- Design the system first and let it specify the component. Edison's filament search was forced by a distribution-voltage decision.
- Consumables and regulation decide markets more often than performance. Brush beat Jablochkoff on carbon life and arc control.
- A change of mechanism beats refinement of one. Two-thirds of the efficacy gain came from not heating anything.
- The incumbent improves too. Gas lighting was advancing throughout the period electric lighting was displacing it.