History Of The Incandescent Lamp - By John W. Howell And Henry Schroeder (1927)

You are reading: Chapter 8: Photometry



The incandescent electric lamp is responsible for the great development which has taken place in the art of photometry. Before Edison invented his lamp, photometry was a crude art and was little used. Laboratories and some gas plants had photometers, using as standards of light either candles or oil lamps, both of which were so variable that there were no really dependable standards. Even though the law in some states specified that a gas burner which consumed five cubic feet an hour, should give sixteen candle power, some gas companies that had photometers rarely used them.

The method then in use was to burn two standard candles at one end of the photometer as a standard of light. These candles were set in a balance which weighed the consumption of material per minute while the measurements were being made, and measurements were then corrected for variations from the specified rate of consumption of the candles. The light given by these candles varied with other conditions, such as the length of the wick, for example, but it was a rule not to trim the wicks during the measurements. In fact the public was not much concerned with candle power at that time.

When incandescent electric lamps came into use, their candle power and efficiency at once became matters of great importance and interest. Edison claimed to get eight 16-candle power lamps per horse power of electricity and constant tests were made to see that the lamps made each day met this condition. Candles were used as the standards of light, but their daily use in the Photometers was soon superseded by carefully standardized incandescent lamps. These standard lamps burned as long as the photometer was in use, which was usually all day, and were frequently checked and corrected by comparison with a number of other carefully prepared lamps used only for this purpose. This method of photometry originated in the laboratory at Menlo Park and has been the universal method ever since.

Every lamp made was measured to determine the voltage at which it gave 16 candle power. This voltage was measured by means of a reflecting electro-dynamometer made for the purpose.


This measures the mean horizontal candle power of lamps. At the right, the standard lamp is set, against which the lamp to be photometered (on the left) is balanced. The voltage and current taken by the lamp being photometered is measured at the same time.

The scale read to 150 volts and each volt near the hundred volt mark gave a deflection of about three-sixteenths of an inch on the scale. The measurements were made on a circuit of 150 volts, resistance being put in series with the lamp to reduce the line voltage to that required on the lamp, which was measured by the electro-dynamometer. The voltage of the line was held constant at 150 volts by manual regulation.

There were no ammeters at that time. The resistance in series with the lamp when it was being measured was in steps of of one ohm each. The voltage on the lamp was read on the electro dynamometer and the resistance in series with the lamp was noted. The difference between 150 volts and the voltage on the lamp was the voltage on the variable resistance and this voltage divided by the resistance noted was the current passing through the lamp. All lamps at that time were measured at 16 candles and curves drawn on cross section paper made it possible to determine the candles per horse power which the lamp gave with the known voltage on the lamp and the resistance in series with it on the 150-volt circuit.

Each lamp, being rated to consume one-eighth of an electrical horse power, was therefore designed for 93¼ watts (one-eighth of 746). The correctness of the power measurements was checked by burning a lamp in a calorimeter and measuring the rise in temperature of the water during a measured time. The calorimeter method was well worked out in its details and gave quite good results. This method of measuring lamp efficiencies continued until about 1890, when reliable voltmeters and ammeters were developed by Edward Weston. It was not necessary to measure the efficiency of every lamp made, but it was necessary to photometer every lamp manufactured to determine the voltage at which it gave its rated candle power.

The first test department was established at the laboratory in Menlo Park, in 1880. In March, 1881, a much more complete test department was set up in the Lamp Factory at Menlo Park by Dr. Edward L. Nichols. In April, 1882, the Lamp Works moved to Harrison and with it the test department. In 1887, the test department was moved to Edison's new laboratory at Orange, N.J., and in 1893 it was moved back to Harrison.

About 1886 J. T. Marshall, of the General Electric Company, invented a method of determining the voltage at which a lamp gives its rated candle power without requiring the use of any electrical measuring instrument. At one end of a photometer he placed a seasoned lamp of the type to be measured on the photometer and of a known voltage, which was the voltage for which the lamps to be measured were designed. The lamp to be measured was placed at the other end of the photometer and the two were connected in multiple so that both had the same voltage acting on them.


In this photometer lamps were measured for their voltage to give their rated candle power without the use of electrical measuring instruments. An empirical scale attached to the balancing screen (in the center) gave the voltage.

If the lamp being measured was of the same voltage as the standard lamp, they would both give the same candle power and the photometer spot would balance in the center of the scale, this point being marked as the voltage of the standard lamp. If the two lamps differed in voltage, they would also differ in candle power and the balancing position of the spot would indicate the voltage of the lamp being measured, by means of an empirical scale.

Voltage fluctuations on the line did not affect the results in this method as the two lamps varied similarly, so this photometry was done on a line which was not carefully regulated. This method allowed very fast operation and was a great boon to photometer work. This "sliding scale" or "same circuit" photometer was in regular use until the advent of the drawn tungsten wire lamp in 1911, when it was no longer necessary to photometer each drawn tungsten wire lamp. Every carbon, GEM, tantalum and pressed filament tungsten lamp had to be photometered individually to determine its voltage.

The reason why it was not necessary to photometer drawn tungsten wire lamps was that the filament could be made to such an exact diameter and length that each lamp could be manufactured to extreme closeness of candle power and efficiency. Sample lamps were photometered to check the manufacture, and the lamps usually came out so close to their designed rating that if the photometric measurements were found to differ from the designed rating, the chances were that there was an error in the photometric readings.

Carbon lamps were all measured for mean horizontal candle power, because they varied in candle power in different parts of the horizontal plane. It was the practice at first to select an average position and measure the lamps in this position. Later the lamps were rotated about their vertical axis while being measured in the photometer, their average horizontal candle power being obtained in this way.

The relation of the horizontal candle power of each type of lamp to its spherical candle power was known. The spherical candle power is the average candle power in all directions. The relation between the two, known as the reduction factor, was 0.825 for the oval anchored carbon lamp, so that its spherical candle power could be determined by taking 82½ per cent of its horizontal candle power measurement.

When tungsten filament lamps were developed in many forms of filament shapes, their ratios of horizontal to spherical candle power varied a great deal. It therefore became advisable to change all ratings to the basis of spherical candle power, which eventually became standard practice. The spherical candle power can readily be measured by burning the lamp in a hollow sphere which has a matt-white inside surface.


With this photometer the spherical candle power of a lamp, the average candle power it gives in all directions, can be obtained by but one measurement.

The cross reflections inside the sphere, coming from the light thrown out in all directions by the lamp being measured, fall on a diffusing glass test plate located on the periphery of the sphere, the direct light from the lamp being screened from the test plate. The light from the test plate is balanced by a standard lamp and therefore gives, in one reading, the spherical candle power of the lamp being measured.

In the early days the efficiency of lamps was measured in candles per horse power. Later the name WATT was given to the unit of electric power and after that lamp efficiencies were stated in watts per candle. The candle in both cases was the horizontal candle power. With the measurement of lamps in spherical candle power, the lamp efficiency was stated in watts per spherical candle. The two terms, watts per candle (WPC) and watts per spherical candle (WPSC), often led to confusion, so that efficiency came to be designated by the term lumens per watt (LPW). This term LPW has an advantage in that the higher the efficiency, the higher the LPW becomes numerically, whereas the reverse obtains with the terms WPC and WPSC.

The lumen is the unit of light flux, and the efficiency of all lamps is now measured in LPW. The lumens delivered by any lamp are 12.57 times its spherical candle power. A lumen is the light flux which a point source of one candle power of light will throw upon a surface of one square foot, every point of which is located one foot distant from the point light source. So if a light source of one spherical candle is placed at the center of a sphere of one foot radius it will yield as many lumens as there are square feet on the inside surface of this sphere, or 12.57 lumens.

In all photometric work, the light of the lamp under test is compared with the light given by a working standard lamp, the correctness of this working standard lamp being frequently determined by comparing its light with that given by a number of photometric standard lamps kept for the purpose. Hitherto all practical photometric measurements have been made by a visual comparison of the lamp to be measured with the standard lamp and, therefore, have depended on the judgment of the eye.

Photometry is not the measurement of an external or objective dimension, but of a sensation, and it is difficult to make a quantitative measurement of our sensations. The attempt to apply measurement to the sensation of smell has not met with success. In spite of the delicacy with which different sensations of taste may be discriminated, it has been impossible to measure taste, particularly as there seem to be physiological reasons for a rapid approach to a saturated condition of the sensation. A similar difficulty arises in the action of light on the eye.

Many attempts have been made to develop a practical method of photometry which did not depend for its accuracy on the human element. Among them may be mentioned the Thermopile, and the Bolometer, which have both been used to measure the whole radiant energy given out by a lamp. This was done by means of electrical apparatus, the dark heat rays being filtered out from the luminous rays by a process of selection. The proportion of energy in the luminous rays is so small compared with the thermal or heat energy rays that it has been impossible to arrive at any precise measurement of light alone. The electrical properties of selenium have given some promise of a quantitative indication of the intensity of light. Photographic methods have been suggested and tried by exposing strips of sensitized paper for a definite time and comparing them with the shades obtained from known illuminations. None of these schemes has been able to compete in a practical way with an ordinary visual photometer.

Recently, however, Charles Deshler, of the General Electric Company, has developed a photometer which substitutes a "mechanical" eye for the human eye. The comparison of light sources, the photometer spot and the working standard lamp have been entirely eliminated. The lamp whose lumens are to be measured is placed as usual in the sphere of a spherical photometer. The integrated light of the lamp passes through a suitable color filter, or test plate and filter, and impinges on a photo-electric cell. A suitable potential is placed across the cell, and the current flowing under these circumstances is proportional to the lumens given by the lamp. This current is measured by a microammeter or galvanometer which thus becomes a "lumen-meter". The photometer is extremely accurate, eliminates the varying human element of the eye, and is much more rapid than any visual photometer.

The principle of the photo-electric cell is based on an electrical property of alkali metals when subjected to light. When the surface of such alkali metals as potassium, barium, strontium, sodium, etc., is exposed to light, it liberates electrons like the heated filament in a radio tube. The cell usually consists of a glass bulb with two terminals. One, the positive terminal, is at the center of the bulb and is equivalent to the plate of a radio tube.


This consists of a glass bulb, coated on the inside with an alkali metal compound, which emits electrons when subjected to light, so that the strength of the current flowing from this coating to a positively charged terminal in the bulb is a measure of the candle power of the light.

The other, the negative terminal, consists of an alkali metal film deposited on the inner surface of the bulb, and is equivalent to the filament in a radio tube. This metal film covers the entire inner surface of the glass bulb except for one clear spot, called the "window", through which the light to be measured can enter the interior of the bulb. When an electrical circuit outside the cell is established through a battery, with the positive of the battery connected to the center (positive) terminal and the negative battery terminal connected through a galvanometer to the metal film (negative) terminal of the cell, the electrons emitted from the metal film will be attracted to the positive terminal inside the bulb, as in a radio tube, and then will flow through the outside circuit back to the negative metal film.


The light from the lamp to be photometered falls on the photo-electric cells mounted on the outside equator of the sphere. The current flowing through the cells passes through a galvanometer which deflects a ray of light on a scale and thus indicates the candle power of the lamp being photometered.

This flow of electrons is the modern theory of the flow of electric current, and as the number of electrons emitted by the metal film depends upon the intensity of the light thrown on it, the strength of the electric current in the outside circuit becomes a measure of the intensity of the light. While this current is minute, of the order of a few millionths of an ampere, it can be measured by a microammeter or by the deflection of a galvanometer needle.

Potassium hydride, an alkali metal compound, is now generally used as the metal film on account of its relatively high melting point and sensitivity. The bulb is highly evacuated and filled to low pressure with an inert gas such as argon, helium, etc. The introduction of these gases produces ionization by collision of the electrons with the molecules of gas in the bulb so that a given intensity of light thereby greatly increases the strength of the current through the cell.

There are two essential difficulties which had to be overcome before the cell could be used satisfactorily for photometric purposes. The first is color sensitivity; that is, the cell responds to certain colors of the spectrum to a greater extent than does the human eye. This was the main difficulty which previously precluded the use of the cell, but it was overcome by the use of a color filter of the proper color. Lamps of equal candle power to the human eye, but which are different in efficiency, have different proportions of the various colors making up the light which they give and so the cell, without a proper filter, would indicate different candle powers. This would mean that, for example, a 100-watt MAZDA C lamp, which is about twice as efficient as a 10-watt MAZDA B lamp, but whose light is much whiter than that of the latter, would be indicated by the cell as giving more than twenty times the difference in candle power between the two lamps as seen by the human eye.

The second difficulty, the minuteness of the current, has been overcome by using a high sensitivity galvanometer or microammeter. With lamps of very low candle power more than one cell can, if necessary, be used in multiple to increase the amount of current.

Life Testing

After the invention of the lamp by Edison, two great questions had to be answered: how much power is required to operate the lamp and how long will the lamp last? In those days the power required to operate a lamp was expressed by the number of candles produced per horse power of electricity consumed. It was immediately observed that the candles per horse power became greater as the temperature of the filament was raised, and it was also observed that as the temperature was raised the life of the filament became shorter. Edison concluded that a lamp to be satisfactory must last about 600 hours, and tests were started to determine the candles per horse power at which the lamps would last 600 hours- So a life test department was created in 1880 at the laboratory at Menlo Park. Early in 1881, a much better one was set up in the lamp factory at Menlo Park.


This photograph shows part of the equipment used in life testing lamps at the Edison Lamp Works.

The life test department made it possible to rate lamps as improvements were made so that they would last 600 hours, and to determine the worth of experimental lamps, so its importance and value were recognized from the beginning. Life testing lamps at their normal rating took a long time, so, as early as 1880, tests at higher than normal rating were regularly made. Lamps were life tested at three times their normal candle power, 16-candle lamps being tested at 48 candles. As lamps improved in quality, this was changed to 64 candles and then to 80 candles.

For this testing a special generator was used, which was held at 150 volts, being regulated by hand. A resistance was placed in circuit with each lamp, which could be adjusted in steps of one ohm up to 100 ohms, so that any lamp could be burned at practically any desired voltage up to 150 volts. From the results of these tests of lamps, J. W. Howell determined in 1885 the relative lives of lamps at different initial candle powers. He found that the lives of lamps varied inversely as the 3.65ths power of their initial candle power. This exponent has been redetermined and checked several times since then by different people. Later, lamps were tested, not at fixed candle powers, but at fixed watts per candle, and recently at lumens per watt, this being now accepted as the measure of the efficiency of the lamps.

The necessity of life testing is just as great now as it was in the early days. Samples from the regular production of every factory are frequently and regularly tested to keep the makers informed of the quality of lamps made, and many experimental lamps from the development and research laboratories are constantly being life tested as an ultimate test to determine their success or value. Lamps are also tested for filament strength, brittleness, ductility, sagging, etc., and each test necessitates the destruction of the lamps tested in order to determine their ultimate characteristics. These tests have to be made with the greatest accuracy and care, the maintenance of the life test department costing a great deal of money and its work destroying a great many lamps.