by Vaclav Smil
FIGURE 2.2. Drawing attached to Edison’s fundamental U.S. Patent 223,898 for the incandescent light shows a tightly coiled filament, but carbonized materials used by Edison in 1879–1880 would have made that impossible, and the first lamps had simple loop filaments. The coiled metallic filament was patented only in 1913 by Irving Lang-muir (see figure 1.8). Reproduced from Edison (1880a).
But more than a century later none of these names is known to the general public, and even the historians of lighting mention some of them primarily because the men were involved in complex litigations with Edison’s company during the 1880s (Pope 1894; Dyer and Martin 1929; Covington 2002). Their names are forgotten because many of their inventions entered the cul-de-sacs of subdivided lighting. They sought materials with very high melting points because the proportion of light to heat radiated by a filament rises with the increasing temperature, and the most efficient light will be produced by the highest filament temperatures. Carbon, with the highest melting point of all elements (3,650°C), is an excellent incandescing material, and they used it as solid rods or charcoal tubes but could not produce sustained light. Platinum has a relatively low melting point of (1,772°C), which can be raised by alloying. In addition, unsuccessful early lamp designs had imperfect vacuum, or their imperfectly sealed glass bulbs were filled with nitrogen.
Edison’s R&D Dash
Edison (figure 2.3) began his lighting experiments—in August 1878, after a memorable trip to the Rockies and California, when he rode most of the way from Omaha to Sacramento on a cushioned cowcatcher of a locomotive (Jo-sephson 1959)—by retracing many unproductive steps that preceded him. Much of his early effort was devoted to alloying platinum with iridium and testing various coatings, including oxides of titanium and zirconium, that further improved the filament’s incandescence. Edison’s first patent application for the incandescent light was made on October 5, 1878 (U.S. Patent 214,636, granted on April 22, 1879): it had a platinum-iridium alloy filament.
Ten days later, on October 15, 1878—after the initial capital was raised from investors by Edison’s long-time friend, patent attorney Grosvenor P. Lowrey—an agreement was reached to set up Edison Electric Light Co. with the objective “to own, manufacture, operate and license the use of various apparatus in producing light, heat or power by electricity” (Josephson 1959:189). In December 1878 Edison boasted to a New York Sun reporter that “I am all right on my lamp. I don’t care anything more about it” (quoted in Friedel and Israel 1986:42), but in reality, there was little progress during the winter months of 1878–1879 as improved versions of alloyed filaments were tested and as entirely new, but unpromising, designs (including luminous zircon and carbon rod resting on platinum) were tried.
FIGURE 2.3. Thomas A. Edison in 1880 at the time of his work on incandescent lights and electric systems. Library of Congress photograph by Emil P. Spahn (LC-USZ62-98067).
At that time, Edison was only one of many inventors (and a late starter) racing to develop practical incandescent light: Moses Farmer, William Sawyer (financed by Albon Man), and Hiram Maxim (1840–1916, much more famous for his 1885 invention of machine gun) were his main competitors in the United States, and St. George Lane-Fox and Joseph Wilson Swan in the United Kingdom. What eventually set Edison apart was not primarily his legendary perseverance in pursuing a technical solution, or the financing he received from some of the richest men of his time, but the combination of a winning conceptualization and a fairly rapid realization of an entire practical commercial system of electric lighting (Friedel and Israel 1986; Israel 1998).
In that sense, he deservedly ranks ahead of even Joseph Wilson Swan (1828–1914), an English chemist and physicist who remains credited in the United Kingdom as the inventor of lightbulb. Swan rejected platinum as the best incandescing substance already during the 1850s and concentrated instead on producing suitable carbon filaments. By 1860 he made a lamp that contained the key ingredient of Edison’s early promising models of 1879: Swan’s carbon filament was made by packing pieces of paper or cardboard with charcoal powder and heating them in a crucible. The carbonized paper strip was then mounted in an (imperfectly) evacuated glass vessel and connected to a battery. Swan eventually abandoned these short-lived lights that could reach only red glow, but resumed his experiments during the late 1870s when better vacuum pumps became available.
A working lamp—demonstrated for the first time at a meeting of the Newcastle-on-Tyne Chemical Society on December 18, 1878—had platinum lead wires and a carbon filament, the same components as Edison’s first longer lasting lamp revealed 10 months later (Bowers 1998). But Swan’s lights operated, as did those of all other inventors before 1879, with filaments whose resistance was very low, ranging from less than 1 Ω to no more than 4–5 Ω. Their goal was stable, long-lasting incandescence, but any mass deployment of such lights would have required very low voltages and hence impracticably high currents, resulting in a ruinously large mass of transmission wires. Moreover, pre-Edisonian lamps were connected in series and supplied with a constant current from a dynamo, making it impossible to switch on the lights individually and forcing a shutdown of the entire system because of a single interruption.
Edison’s key insight was that any commercially viable lighting system must minimize electricity consumption and hence must use high-resistance filaments with lights connected in parallel across a constant-voltage system. This concept, so contrary to the prevailing wisdom of the time, was at first questioned even by Francis R. Upton (1852–1921), the only highly trained scientist (mathematics and physics, at Princeton and Berlin University) in Edison’s laboratory, who was hired specifically to formulate and to buttress the inventor’s ideas in rigorous scientific terms. Forty years later Upton recalled how eminent electricians of that time maintained the subdivision of the electric current was commercially impossible, and how muddled was their understanding of the very concept (Jehl 1937). A few simple calculations illustrate the difference between the two approaches.
An incandescent lamp of 100 W whose filament had resistance of just 2 Ω and operated at 36 V (common pre-Edisonian ratings) would have required the current of 18 A (V/Ω). In contrast, voltage of 110 V and a lightbulb whose filament had resistance of 140 Ω—these being parameters chosen by Edison and Upton in order to reduce the cost of copper conduits and to make filaments stable (Martin 1922)—needed just 0.79 A for a 100-W light. This means that for the same transmission distance a system composed of the low-resistance lamps would have required 70 times the mass of identical conducting wires to distribute electricity than would the high-resistance arrangement. The former choice was obviously a nonstarter from the economic point of view. As Edison noted in his specification for the first patent concerning high-resistance lamps (U.S. Patent 227,229 submitted on April 12, 1879), “By the use of such high-resistant lamps I am enabled to place a great number in multiple arc without bringing the total resilience of all the lamps to such a low point as to require a large main conductor; but, on the contrary, I am enabled to use a main conductor of very moderate dimensions” (Edison 1880b:i).
This was a winning concept, but the patented lamp itself was only an improved, but impractical, version of his older platinum lights. The lamp had an alloyed platinum wire wound into a bobbin mounted inside a small vacuum tube (his technicians succeeded in creating a vacuum of nearly one millionth of the atmospheric pressure) that was placed inside a glass cover. A flexible metallic aneroid chamber underneath the vacuum tube expanded when the bobbin got very hot and interrupted briefly the electrical circuit and prevented the filament from overheating without affecting the uniformity of the light. This was an ingenious design, but the complex arrangement of connections, wires, and magnets and the placement of the vacuum bulb within another glass container were not obviously a basis for a sturdy and practical lamp.
Edison’s Success and Competitors
Once Edison abandoned the idea of a thermostatic regulator, he turned not only to achieving a nearly perf
ect, and long-lasting, vacuum inside the lamp and to searching for filaments other than platinum, but also to designing a new telephone and building his first dynamo, the largest electricity generator of that time. He returned to his search for a better filament only in August 1879. Many different substances, including bones and sugar syrup, were subsequently carbonized in a small furnace. Edison’s first success with carbonized cotton sewing thread in October 1879 was followed by intense activity to improve the design before its first public presentation. A filament made of carbonized cardboard (Bristol board) proved to be a much better choice than cotton thread, and new lamps lasted easily more than 100 hours.
But before the demonstration, scheduled for December 31, 1879, took place, the New York Herald scooped the event by publishing a full-page article in its December 21 issue. Authored by Marshall Fox, who spent two weeks in the Menlo Park laboratory preparing the piece, the lengthy article was informative, accurate, and well written (it is reproduced in Jehl 1937). Fox conveyed well not only the incredulity of the event (electric light being produced from “a tiny strip of paper that a breath would blow away”) and the glory of its final product (“the result is a bright, beautiful light, like the mellow sunset of an Italian autumn”) but also the basic scientific preconditions and challenges of Edison’s quest. Not surprisingly, it was later judged by Edison as “the most accurate story of the time concerning the invention” (Jehl 1937:381).
The first public demonstration of incandescent lighting took place, as planned, at Menlo Park on December 31, 1879. Edison used 100 cardboard filament lamps—long-stemmed and topped by an onion-shaped dome, each consuming 100 W and producing about 200 lm—to light the nearby streets, the laboratory buildings, and the railway station. In so many ways this was not an end but merely a frantic beginning of the quest for incandescent light. Other inventors rapidly switched to high-resistance designs, and new patents were filed at a dizzying rate. In 1880 Edison alone filed nearly 30 patents dealing with electric lamps and their components (TAEP 2002). Swan’s 1880 British patents, based on the key components of his previous designs, incorporated high resistance and the parallel arrangement, and after they were upheld in British courts Edison decided not to challenge them. Instead, a joint Edison & Swan United Electric Light Co. was set up in 1883 to manufacture Swan’s lamps in the United Kingdom.
In the United States, Edison had to embark on many-sided defense of his patents while also defending himself against numerous claims of patent infringement. Between 1880 and 1896 more than $2 million was spent in prosecuting more than 100 lawsuits of the latter kind (Dyer and Martin 1929). Edison’s greatest loss came on October 8, 1883, when the U.S. Patent Office decided that his lightbulb patents rest on the previous work of William Sawyer. The U.S. Circuit Court of Appeals made its final decision in favor of the Edison lamp patent only on October 4, 1892, more than 12 years after the filing of the patent itself, leaving the inventor to complain that he had never enjoyed any benefits from his lamp patents.
But protracted litigations did not stop the search for better filaments. Edison decided to ransack the world for the most suitable materials, and his laboratory tested more than 6,000 specimens of any available plant fibers, ranging from rare woods to common grasses. Edison finally settled on a Japanese bamboo (Madake variety), whose core had a perfect cellular structure and yielded a strong and highly resistant carbonized filament. These bamboo filaments of the early 1880s lasted about 600 hours, a great improvement on 150 hours obtained with the cardboard lamps of December 1879. But cellulose filaments, introduced by Swan in 1881, proved to be the most popular carbon choice, and their improved versions became the industry standard by the early 1890s.
Different filament configurations were designed to direct more light downward for the lamps mounted with their base up, while other designs for standard mountings tried to diffuse the light more evenly in all directions. As for the bulbs, they were all initially free-blown from 2.5-cm tubing, first in short and later in longer pear shapes, and their glass widened gently after it emerged from the base collar. In contrast, mold-blown bulbs, first introduced by the General Electric Co. during the early 1890s, had a short but almost perpendicular step just above the collar, and only then they widened into a rounded shape. Early bulbs came in a profusion of shapes, ranging from spheres to ellipsoids and from elongated teardrops to lozenges, with straight or curving sides, with or without necks, all of them with a short tip as the glass was nipped off while mouth-blowing the bulb, and Edison himself designed many of these shapes (figure 2.4). The first electric decade also saw the first colored, ground glass, opal and etched lamps, and ornamental lamps of various shapes.
FIGURE 2.4. Different kinds of Edison’s light bulbs from the 1880s. Reproduced from Figuier (1888).
The earliest bases were wooden or, between 1881 and 1899, plaster of Paris. Porcelain was introduced in 1900, and glass insulation, standard in today’s bulbs, was introduced a year later. Platinum was used for lead wires, and carbon filaments were connected first with clamps and after 1886 with carbon paste. Edison’s bamboo filaments had first a rectangular and later a square cross section and a hairpin shape. Squirted cellulose filaments, which dominated the market after 1893, soon substituted a loop for a hairpin in a round cross section. Edison’s bamboo filament was abandoned after Edison General Electric merged with Thomson-Houston in 1892 to form the still prosperous, and now highly diversified, General Electric Co. But modern lamps incorporate more than the great inventor’s high-resistance, parallel-connection design; they also perpetuate his patented mounting.
Swan’s lamps had a characteristic side-pin twist-lock base that was invented by his brother Alfred (G. B. Patent 9,185, June 19, 1884; U.S. Patent 313,965, March 17, 1885). This was a direct predecessor of the bayonet base and socket that is still used in the United Kingdom as well as by automakers for tail and parking lights of cars and trucks worldwide. Lamps made by Westinghouse, Sawyer and Man, and Thomson-Houston had a clip base, and more than a dozen different base styles still coexisted during the mid-1890s. But it was a simple screw base—whose initial inspiration came to Edison early in 1880 as he unscrewed the cover of a kerosene can, and whose design was done by two of Edison’s long-time associates, Edward H. Johnson and John Ott, during the fall of 1880 (Friedel and Israel 1986)—that was borne by about 70% of all lamps sold in 1900 and that became an industry standard by 1902.
Nothing demonstrated better the state of incandescent lamps after more than a dozen years of their rapid evolution than did the lighting at Chicago’s Columbian Exposition of 1893, the first such event solely illuminated by electricity (Bancroft 1893). Arc lights were still used for some outdoor locations, but the Westinghouse Electric & Manufacturing Co. was the principal lighting contractor, and it used 90,000 “Stopper” lamps operating at 105 V. These were basically Sawyer-Man lamps with a ground glass stopper as their base, nitrogen-filled bulb, and iron lead wires that were designed to circumvent various Edison’s patents before their expiry. Inside the Electrical Building at the exposition were more than 25,000 incandescent lights, including 10,000 General Electric lamps as well as designs by Westinghouse, Western Electric, Brush Electric, Siemens & Halske Co., and five smaller companies.
But all of these lamps shared one undesirable property: very low efficiency of converting electric energy into visible light. The earliest designs produced just over one lm/W, which means (converting with the average of 1.46 mW/lm) that they turned no more than 0.15% of electric energy into light (figure 2.5). Although this was a very low value on the absolute scale, it was an order of magnitude better performance than for paraffin candles, whose combustion converted a mere 0.01% of energy in the solid hydrocarbon into light, and a nearly four times the rate for gas lights: using typical performance data from Paton (1890), I calculated that gas jets converted less than 0.04% of the manufactured fuel to visible light. Improved filaments raised the typical performance to 2.5 lm/W (0.37%) by the mid-1890s (figure 2.5), and it was
obvious that the ongoing large-scale diffusion of electric lighting would greatly benefit from replacing inefficient, and fragile, bits of carbonized cellulose by more luminous, as well as sturdier, metallic filaments.
FIGURE 2.5. Efficacy of electric lights rose by nearly an order of magnitude between 1879 (Edison’s first long-lasting carbon filament) and 1912 (Langmuir’s tungsten filament in inert gases). Calculated and plotted from data in a variety of contemporary and retrospective publications.
Metallic Filaments
Standard efficacy of 3.5 lm/W (0.51%) was reached in 1898 by the deposition of carbon in the pores of cellulose filaments heated in the presence of petroleum vapor. In that year Carl Auer von Welsbach (1858–1929)—who in 1885 patented the incandescent mantle, a delicate gauze cylinder impregnated with thorium and cerium oxides whose greater luminosity delayed the demise of gas lighting (Welsbach 1902)—introduced the first working metal filament made of osmium. This metal’s melting point (2,700°C) is 1,000°C above that of platinum, and osmium lights had efficacy of 5.5 lm/W, but the rarity, and hence the high cost of the element, prevented their commercial use.