# Popular Science Monthly/Volume 86/May 1915/Measuring Heat from Stars

 MEASURING HEAT FROM STARS
By Dr. W. W. COBLENTZ

IF the title of this paper had been chosen as, "Measuring Heat from Stars—Of What Practical Use Is It Anyway?" it would express the feeling of the average interviewer. The question is not meant to be contemptuous. It is the expression of the mind that figures everything on the basis of an immediate return upon his investment.

The interrogator may be one "engaged in writing an article for some magazine" and he must needs "tell the layman the practical side of the subject." Suppose I, in turn, ask my interrogator the question: "Of what prospective use was your layman when he was in the first stages of development?" Said interrogator shrugs his shoulders, smiles, and admits that perhaps the "layman" is not the only one to be considered; that some of the results of investigations must go into the great storehouse of knowledge, no one knowing what their ultimate use may be; that the great unknown can be explored only step by step; that each achievement may be only one more link in the chain of knowledge, perhaps to be disputed and refuted by some future investigator or perhaps be put to some practical use of some future "layman."

Another question raised is, "Why the Government should be measuring the heat of stars?" not realizing that such activities are incidental, conducted to assist, if possible, its citizens in every possible way. The narrow-gauge college professor may perhaps take it as an intrusion upon his field. To such an one the writer can but tell the lesson taught him some months ago, when one bright morning, on walking through the woods, a loud commotion was heard in the topmost branches of a tall oak tree. Two wood-peewees were quarreling for the possession of this tree—as a place to catch flies! There were hundreds of other trees close by, then why quarrel about this one. How like the scientific man, I thought. We quarrel for fields of research, just as though the heavens were not ablaze with objects for investigation.

A further question asked by the interrogator is, "Well, how do you measure the temperature of the stars?" This question may be easily dismissed by saying that we can not measure the temperature of the stars. The best we can do is to attempt to measure the rate at which they are losing heat. But until the past summer even this attempt was of little interest to the astronomer. Heretofore experimenters were glad to be able to record heat from a few of the brightest stars; let alone attempting to measure stars of the 6.7th magnitude which are about 1480th as bright. This is not saying that previous attempts were of no avail; for they were the stepping stones which aided in what small success was attained with the Crossley reflector on Mt. Hamilton, this past August. If, then, the "layman" with his question of the practical side will please wait a little while longer, and in the meantime consider that the present work is simply another stepping stone in the path of conquest of the secrets held in the firmament, it will greatly help the investigators who are not concerned with the immediate commercialization of everything in and under the heavens.

II. A Brief Summary of Previous Attempts at Measuring Stellar Radiation

The measurement of stellar radiation has been attempted by three methods: (1) by means of thermoelements, (2) by means of a Nichols radiometer, and (3) by means of a selenium cell.

Among the earliest attempts by means of thermoelements are the measurements of Huggins.[1] He used one or two pairs of elements of bismuth-antimony in the focus of a refractor having an aperture eight inches in diameter. He recorded positive deflections for Sirius, Pollux, Regulus, and Arcturus. The data given are very meager. It required from four to five minutes (fifteen minutes in one record) to obtain a reading.

Thermoelectric measurements of the radiation from Arcturus and Vega were made by Stone[2] who used a refractor 12.75 inches in diameter. In spite of the excessively long time (about ten minutes) required to obtain a reading he appears to have obtained fairly reliable results. His measurements show that Arcturus emits more radiation than does Vega; his numerical measurements for June 25, 1869, being Arcturus: Vega${\displaystyle =}$3:2. Considering the fact that the infrared radiations from Arcturus suffer greater absorption than those of Vega in passing through an air mass highly saturated with water vapor, and in passing through the glass lenses of the refractor this ratio (3/2) is in close agreement with subsequent measurements using a reflecting telescope.

Recent measurements of stellar radiation were made by Pfund,[3] using thermoelements in an evacuated receptacle. The receivers attached to the junctions of the bismuth alloys (BiSn—BiSb) were about 1.2 mm. in diameter. The sensitivity was such that the radiation from a Hefner lamp at a distance of 1 m. gave a deflection of 2400 mm. He used a reflecting telescope thirty inches in diameter, and made measurements on Vega (7.5 mm. deflection), Jupiter (part of disk; 3 mm.), and Altair (2.0 mm. deflection, sky hazy). The ratio of the radiations Vega: Altair${\displaystyle =}$3:7, which is at variance with the results obtained on Mt. Hamilton, and emphasizes the importance of making observations through an atmosphere free from water vapor. He concluded that with a more sensitive galvanometer and one of the largest reflectors it would be possible to observe stars to the fourth magnitude.

An extensive series of measurements of the radiation from Arcturus, Vega, Jupiter and Saturn were made by Nichols[4] by means of his radiometer, which, like the thermopile, absorbs all the radiations of all wave-lengths falling upon it. The receivers were 2 mm. in diameter. A candle at a distance of 1 m. would have given a deflection of 724 mm. He used a two-foot reflector and observed deflections of 1 to 2 mm. The sensitivity of his radiometer was such that a deflection of 1 mm. would be caused by 1/68,750,000 of the heat received on a surface equal to the aperture of the concave mirror from a candle at 1 meter distant. Or, neglecting atmospheric absorption, the sensitivity was such that by using the two-foot mirror to focus an image of the flame upon the radiometer, he would have obtained a deflection of 1 mm. from the candle placed at a distance of 5 miles. He concluded that the thermal intensity was Vega: Arcturus: Jupiter: Saturn${\displaystyle =}$1:2.2:4.7:0.74. As for the possibility of further work he concluded that by using a five-foot reflector it would be possible to observe white stars down to the second magnitude and red stars possibly to the third magnitude.

The Boys[5] radiomicrometer has also been tried in measuring radiation from stars. The instrument was used with a sixteen-inch reflecting telescope. The slight deflections obtained on various planets and stars were regarded as of questionable origin.

The earliest measurements of the light from stars by means of a selenium cell were made by Minchin,[6] who used a 2-foot reflector. He examined about a dozen stars, some being as small as the third magnitude. Owing to the peculiar properties of the selenium cell, which is highly selective in its response to radiations of different wave-lengths, the data can not be used in comparing the radiation from different stars. The selenium cell can be applied, however, in the measurement of the maximum and minimum of light emission from a variable star which does not change in color. For this purpose it has been used by Stebbins[7] in connection with a 12-inch refractor.

III. A Brief Account of the Present Measurements of Stellar Radiation.

The telescope used in the present investigation was the well known Crossley Reflector which is part of the equipment of the Lick Observatory at Mt. Hamilton, California. The reflecting mirror is three feet in diameter. The altitude of the station is a little over 4,000 feet. The summer months being rainless; there being no fog or dew; the night temperature being only a few degrees lower than the day time—these were items which made it possible to have fairly uniform conditions on different nights.

The radiometers used in these measurements were minute thermocouples with receivers 0.3 to 0.4 millimeter in diameter; i. e., about as large as the punctuation mark at the end of this sentence. These thermocouples, the elements of which were bismuth and platinum, were mounted in glass receptacles, as shown in Fig. 1, from which the air could be evacuated. The vacuum-was then maintained by occasionally heating metallic calcium, Ca, contained in a quartz-glass tube shown in Fig. 1. Metallic calcium has the property of absorbing atmospheric

Fig. 1. Showing the Glass Receptacle which Contains the Thermocouples, E, and the Calcium Ca Used to Maintain a Vacuum.

gases when warmed to a low red heat. This glass receptacle was then mounted in a brass box as shown in Fig. 2, which was made especially to take the place of the plate-holder in the camera which is part of the equipment of the telescope. By removing the screws S, S, it was therefore a matter of only a few minutes to dismount this radiometric outfit and substitute the plate-holder. In this manner part of the night was spent in making radiometric measurements on stars, after which the telescope was surrendered to another observer who was photographing a newly discovered satellite of Jupiter.

Referring to Fig. 1, it may be added that the star light after reflection from the telescope mirror passes through a fluorite window, F, and is brought to focus upon the receiver, E, of the thermocouple where the rays are absorbed thus heating the thermojunction. This extremely minute amount of heat is sufficient to warm the thermojunction a few hundred-thousandths of a degree and thus generate an electric current which passes through the coils of a miniature tangent galvanometer, shown in Fig. 3. Unfortunately one sees nothing of these coils of wire which are imbedded in two blocks of Swedish iron. Because of its high magnetic permeability, iron is used to shield the suspended system of magnets from variations in the magnetic field, such as produced by passing street cars, etc. This galvanometer is very sensitive, so that it responds to a current of less than one ten-billionth

Fig. 2. Showing the Glass Receptacle of Fig. 1 Mounted by Means of the Screws S, S, in the Plateholder of the Crossley Reflector.

of an ampere, and in observing the heating effect produced by different stars, measurements could be made in four to five seconds.

This outfit consisting of an ironclad galvanometer and two receptacles containing the thermocouples was constructed in Washington, D. C, and carried to Mt. Hamilton, Calif., a distance of over 3,200 miles without serious mishap (one thermocouple was broken in climbing the mountain). A vacuum pump was shipped, but it was not unpacked. The slight vapors given off from the cement and stop-cock grease were removed from these receptables by occasionally heating the calcium by means of a small alcohol blast lamp. From this it is evident that one of the principal achievements was in demonstrating that with an equipment of several thermoelements, mounted in evacuated receptacles, one can go to the remotest station for radiation measurements, without taking an expensive or cumbersome vacuum pump.

After observing for several nights it was found that red stars emitted far more radiation than do the blue ones having the same photometric brightness, and attention was given mainly to the solution of this question. Accordingly stars were selected having the same visual magnitude, but differing in color; and which were close together in right ascension and zenith distance, in order to obtain the measurements of the ways traversing the same air mass. In passing through our atmosphere, the radiations from the red stars suffer a greater absorption than do the blue stars. For this reason it was desirable to eliminate, as much as was possible, the effect of air mass.

As the work progressed, it became a rather instinctive feeling to avoid the blue stars of less than the fourth magnitude, owing to the difficulty in measuring their radiations. However it was found possible to measure the heating effect of red stars down to the 6.7 magnitude.

Fig. 3. Ironclad Thomson Galvanometer Used in Measuring the Electric Current Generated by the Thermocouples. One of the iron shields Is shown to the right.

If it had been merely an attempt to show the possibilities of the instruments, then by selecting red stars, and by increasing the galvanometer sensitivity, positive indications could have been obtained of radiation from stars of the eighth to ninth magnitude. That, however, would have been simply a spectacular achievement, to awe the layman, and under the present conditions of observation, could not contribute much to science.

The aim was to do one thing thoroughly, rather than to attempt a varied program. This one thing was the establishment beyond all reasonable doubt, by two distinct methods, that the red stars as a class emit a far greater amount of total radiation than do the blue stars; that they have a higher emissivity, or, in other words, that they are cooling faster than the blue stars. (In parenthesis it may be added that it is doubtful whether the interior of a red star is cooler than a blue star. The whole mass is no doubt shrinking and the temperature may be actually rising. But the idea to be conveyed is that in the various stages of stellar evolution, the "red star stage" seems to be the one in which a star is "burning out" the fastest.) The first method, viz., measuring the total radiation from red and blue stars having the same photometric brightness, as already mentioned, is somewhat uncertain, because the radiation received from a star is a function of its size, distance, temperature and especially its emissive power. Some of the largest stars are no doubt the farthest from us. The second method of observation consisted in roughly separating the star's rays into a spectrum, by means of an absorption cell of water which absorbs most of the infra-red rays and transmits the visible rays. Hence, by measuring the total radiation from a star, and also the part which is transmitted by the absorption cell of water we obtain an estimate of the relative amounts of energy in these two parts of the spectrum. This measurement is a ratio of two quantities of energy; and hence is independent of the size, the distance and the temperature of the star. It gives us direct information of the emissivity of the different parts of the star's spectrum. It is true that it gives us information of only two parts of the spectrum; but, from our knowledge of the solar spectrum, and of the spectra of terrestrial substances, this information enables us to make important deductions as regards the distribution of energy in the spectra of stars.

Table I

 Object Magni-tude Deflec-tion Type Object Magni-tude Deflec-tion Type β Orionis 0.34 2.50[8] B8 p. 19 Piscium 5.30 0.46 N α Orionis 0.92 22.4 Ma γ Aquarii 3.97 0.24 A θ2 Tauri 3.62 0.18 A5 λ Aquarii 3.84 1.02 Ma ε Tauri 3.63 0.35 K δ Capricorni 2.98 0.28 A5 ν Tauri 3.94 0.12 A β Aquarii 3.07 0.55 G γ Tauri 3.86 0.36 G β Ophiuchi 2.94 0.37 K δ Tauri 3.93 0.52 K δ Ophiuchi 3.03 1.37 Ma α Auriga 0.21 6.14 G α Coronæ Borealis 2.32 0.48 A α Tauri 1.06 6.78 K5 γ Draconis 2.42 1.59 K5 6.84 β Ursæ Majoris 2.44 0.37 A δ Ceti 4.04 0.08 B2 γ Draconis 2.42 1.58 K5 ν Ceti 4.18 0.31 Ma 1.67 φ Pegasi 5.23 0.22 Ma 1.64

Some of the data obtained by measuring the total radiation from blue and from red stars, having the same brightness, will now be discussed. It was possible to make quantitative measurements on stars down to the 5.3 magnitude. It is to be remembered, in considering this data, that a star of say the second magnitude is only 12.5 as bright as a star of the first magnitude, i. e., one magnitude differs from another by the factor 2.5 in brightness. However the total radiation may be entirely different as may be noticed in Table I. In this table the last column gives the spectral classification used by astronomers. The blue stars have the classification B, A. These stars pass into the yellow gradation F, G, K, the latter being yellowish-red. The red stars are Class M and the deep red ones are Class N. Keeping in mind this classification, it may be noticed that the red star α Orionis emits about eight times as much total radiation as the blue star, β Orionis, which is much brighter to the eye. A similar example is the yellow star α Auriga (Capella) and the beautiful red star, α Tauri (Aldebaran), both of which stars are familiar objects. The latter is only about one-half as bright to the eye; and yet it emits the same amount of total radiation as does the brighter star. Comparing stars of the same photometric brightness the 5.3-magnitude stars Φ Pegasi and 19 Piscium are interesting examples because of their smallness, and because the latter is of a deep red color. The latter follows the general rule that the redder the star the greater the amount of total radiation received. The number of these very red stars is very small and they were not conveniently situated for observation. However from the observations on numerous stars of Class M as compared with blue and yellow stars of the same photometric magnitude, it is to be expected that these very red stars, Class N, will be found to have, as a general rule, the highest emissivity of all. Among all the data collected, on 105 stars, there are no exceptions to the general classification, viz., the redder the star the greater the amount of total radiation emitted. To some, of course, this information is not unexpected. However, if the reader will pause for a moment and consider that some of these measurements were made on starlight which left its source more than 160 years ago, and that stellar distances are so inconceivably great that another 160 years must elapse before the arrival of starlight which is being emitted at the present moment, it will be evident that every measurement has a value of far greater importance than merely confirming our expectations which are based upon our preconceived notions of what is occurring on a star and in passing through interstellar space.

The second method of studying the quality of the radiations of red and of blue stars, by means of the absorption cell of water, is more limited in range, because of the weakness of the radiations received. Only a few stars could therefore be investigated by this method. From the first method of observation it is to be expected that the total radiation from a red star contains more infra-red rays than does the total radiation from a blue star, and hence the amount transmitted by the

Table II

Transmission of Stellar Radiaition Through a 1 cm. Layer of Water

 Object StellarClass Transmis-sion inPer Cent. Remarks Blue Stars α Lyræ A 58 (Vega) α Aquilæ. A5 69 (Altair) β Orionis. B8 p. 42 (Rigel) A low value for a blue star Yellow Stars α Auriga G 48 (Capella) α Boötis K 45 (Arcturus) α Tauri K5 35 (Aldebaran) γ Draconis K5 32 Red Stars β Pegasi Mb 29 α Orionis Ma 27 (Betelgeux) α Scorpii Ma p. 27 (Antares) α Herculis Mb 21 Jupiter 65 ⁠Receiver in center of disk including part of dark band. 66 Receiver covers upper dark band. Venus 59 Saturn 55 Receiver covers central disk. Moon 14.7

water cell will be less for the red star, Class M, than for the yellow star, Class F, G, K, and a blue star. Class B, A. This is the true condition of affairs, as may be noticed in Table II., which gives the percentage of the total energy falling upon the thermocouple, which is transmitted by the water-cell, for blue, yellow and red stars. From this table it may be seen that as much as 60 per cent, of all the radiations coming from a blue star lies in the spectral region to which the eye is sensitive, while only from 20 to 30 per cent, of the total radiation received from a red star affects the eye and the photographic plate. This brings out very clearly why it is that a red star of the same visual brightness as a blue star (causes a larger galvanometer deflection) emits from two to three times as much total radiation. It means that from 70 to 80 per cent, of the radiation from a red star lies in the infra-red—beyond the spectral region to which our eyes are sensitive.

The absorption cell tells us nothing of the size or the distance of the star. It indicates that the shape of the spectral energy curve of the star is such that only about one fourth of the, total energy emitted by a red star lies in the visible and in the ultraviolet part of the spectrum. However, it should dispel all doubt as to the quality of the radiations emitted by red and by blue stars. The absorption does more than merely tell us the region of the spectrum in which the most energy is distributed. It may prove useful in detecting dark companions of star systems. There are no doubt many blue and yellow stars having companions which have become so cool and nonluminous that their presence can not be detected photographically. To the eye the star will appear to be blue or yellow. A good illustration is the ordinary paraffin candle. To the eye it looks yellow, because the red-hot wick contributes but little to the total luminous output. But the red-hot wick is very rich in infra-red radiations, as compared with the luminous flame, and when measured with a thermopile (or some other similar radiometer), and an absorption cell, it is found that the red-hot wick contributes materially to the total energy radiated from the flame. In a similar manner a red-hot star will contribute materially to the total radiation from a star system which appears to be blue. The conspicuous star β Orionis (Rigel) is an excellent illustration. Astronomers have classified it "B 8 p" the letter "p" meaning that it has a peculiar spectrum. Whether this peculiarity is sufficiently definite to indicate that there is a companion star, the writer does not know. However, the writer's radiometric classification, Table II., would place it with the yellowish-red stars having a transmission of some 40 per cent, through the absorption cell, instead of with the blue stars. No doubt this star has a dark companion which has thus far escaped detection.

An excellent example which the writer desired to study, but was prevented owing to the fact that at that time (August, 1914) the star does not rise before dawn, is Sirius. This star has a companion which has become so cool that it is rated as a tenth-magnitude star. Although the presence of a companion star was suspected, the light coming from it is so weak that in the presence of the bright star (Sirius), for some years, it defied detection. This companion star is of enormous size, being one half as massive as the bright component; but it sends out only 1/30,000 as much light. It would have been very interesting to determine what amount of radiation from this star system is transmitted through the water cell.

Two stars having 5th magnitude companions were studied, viz., β Pegasi and α Herculis. Both stars appeared to have an excessive amount of total radiation in comparison with their photometric brightness. In the case of α Herculis the companion star caused a deflection of almost a centimeter on the galvanometer scale. In Table II. it will be noticed that this star has the lowest transmission through the water cell. No doubt this is attributable to the large amount of infra-red radiations contributed by the companion star, which caused galvanometer deflections about 15 those obtained from the bright component.

Attention has already been called to the fact that the receiver, which was attached to the thermojunction, was very small. It was therefore possible to measure the radiations from different parts of the surface of a planet. One of the most interesting series of measurements was made on the light reflected from the bright and the dark bands of Jupiter. Both gave practically the same transmission through the water cell, showing that whatever may be the cause of these dark bands, the diminution in brightness is quite non-selective as regards the infrared. Interesting measurements were made on Saturn and its rings. Measurements on a planetary nebula showed no positive indications of radiations from it. However, from the observations on blue and red stars, it was not expected that definite indications would be obtained of radiations from a nebula.

In marked contrast with Venus, Jupiter and Saturn, only about 15 per cent, of the light reflected from the Moon is transmitted by the water cell. This is attributable to the fact that the Moon having no atmosphere, the surface becomes warm from exposure to the Sun's rays, and in turn radiates heat waves, which are not transmitted by the absorption cell of water.

In view of the fact that, heretofore, observers were glad to obtain any indication of the radiation from stars and planets, it is of interest to record that in observing the radiation from Venus it was necessary to place a resistance of 50 ohms in series with the galvanometer in order to reduce the sensitivity and thus keep the galvanometer deflection (which amounted to 127 cm.) upon the scale.

IV. The Absolute Value of the Total Radiation from the Stars

It is of interest to obtain a rough estimate of the total amount of heat received from a star as compared with the heat received from the sun, which is of the order of 1.9 gram calories per square centimeter per minute. This was accomplished by standardizing the thermocouples and galvanometer in terms of radiant power. In this way it was determined that the amount of starlight which caused a deflection of 1 mm. = 34 X 10−14 gram calorie per sq. cm. per minute. Or it would take 100,000,000,000,000/34 minutes, i. e., six million years to raise the temperature of 1 gram of water 1° C. The star Polaris is an excellent example. It produced a galvanometer deflection of 6 mm. Hence it would require only one sixth as long to raise the temperature of 1 gram of water 1° C. In other words, assuming that, in the meantime, all the incoming radiations are absorbed and that no heat is lost by conduction, convection or radiation, then it will require the radiations from Polaris to fall upon 1 square centimeter continuously for one million years in order to raise the temperature of 1 gram of water 1° C. In marked contrast with this value, the radiation from the sun which is transmitted by our atmosphere and falls upon an area of 1 sq. cm. of the earth's surface is sufficient to raise the temperature of 1 gram of water 1° C. in about one minute. Moreover, the total radiation from all the stars which at any moment can fall upon 1 sq. cm. of the earth's surface is so small that it would have to be absorbed and conserved continuously for a period of 100 to 200 years in order to raise the temperature of 1 gram of water 1 degree centigrade. Evidently the incoming stellar radiation can contribute but little in retarding the cooling of the earth. For the measurements of nocturnal radiation, which is usually a loss of terrestrial radiation into space, indicate that for a lampblack surface the outgoing radiation may be as high as 0.1 the incoming solar radiation. The emissivity of the materials forming the earth's surface may be much lower than this value. Nevertheless, it is much greater than can be compensated for by a continuous incoming of stellar radiation. The temperature of the earth must therefore ultimately tend towards the absolute zero of temperature.

V. Some Astroradiometric Problems that Await Solution

A complete statement can not be made of the problems in stellar radiation that require investigation. They stare one in the face whichever way one turns. Realizing the inadequacy of the radiometric apparatus now available one must sit with eyes closed in order not to become impatient with existing conditions, both as regards the production of radiometric apparatus sufficiently sensitive to make the measurements and as regards financial assistance which is necessary to carry on the work. It is not a work that can be "cleaned up" in a season. It will require years of painful, nerve-racking toil in order to accomplish anything of worth.

While it is hoped that the present investigation will make available to the astronomer one more instrument for the investigation of celestial objects, it is desirable to emphasize that, from the insensitive nature of the instrument, the astronomical application can not be very wide as compared with the spectrograph. However, its physical properties are such that, in a limited field, it can be employed in attempting the solution of some of the most fundamental questions in astrophysics. Take for example the question of the emissivity of blue stars as compared with red stars. The general conclusion appears to be that blue stars are at a higher temperature than red stars, and that the emissivity of the red stars is higher than that of the blue stars. The higher emissivity of the red stars would be attributable to a marked change in the distribution of energy in the spectrum, brought about by a change in the physical condition of the stellar surface.

With the rather insensitive radiometric outfit used in the present investigation it +was shown that the total radiation received from a red star is two to three times that of a blue star of the same photometric brightness. These observations should be extended. Another field of astroradiometric research is the measurement of the radiation from variable stars, especially those which undergo a change in color. A general radiometric survey of the stars is desirable; especially of star systems which may have companions which are too dark to detect photographically. The bright components of these stars would give an excess of total radiation as compared with other stars of the same color and having the same photometric brightness. Two stars apparently giving such an excess of total radiation were found in this preliminary survey and no doubt many other examples will be found.

It is possible that measurements of the total radiation from stars may be of assistance in answering the question whether light is absorbed in traversing interstellar space. When one considers that the measurements, made last August, on the radiations from Polaris (the pole star) were vibrations which were emitted forty-seven years ago, and that the radiations from the Orion group of stars started on their journey through space 160 years ago, the distances involved are so inconceivable that one naturally wonders how it can be possible that there is not sufficient "cosmic dust" in interstellar space to scatter and thus diminish the visible radiations to a greater extent than the invisible radiations; and yet the spectrographic evidence seems to be against this sought-for absorption of light in space. Another question awaiting solution is whether there is a "dispersion" of light in space; i. e., whether there is a retardation of say the violet rays as compared with the infra-red rays, so that the infra-red rays get here quicker than do the violet rays. This can be determined by measuring the radiation from an eclipsing variable star. If there is a retardation of some of the rays then the maximum and minimum of light emission should be different for different parts of the spectrum. This, however, opens up a new question of infra-red radiation from a dark comparison star, which may cause the eclipse, and it does not seem desirable to prolong this speculation.

It is an easy matter to indicate the problems demanding solution. It is quite a different matter to produce the instruments for their solution and in leaving now the discussion of the results obtained in the present investigation it is desirable to emphasize once more that the advance made thus far in developing astroradiometric instruments is very small in comparison with what will be required in order to make real progress in the work. For example, in the pioneering work of Nichols about fifteen years ago, the radiation sensitivity of his instruments was such that a deflection of 1 mm. would have been produced on his observing scale by a candle removed to a distance of five miles. The sensitivity of the radiometric apparatus used in the present work was more than 100 times as great, so that 1 millimeter deflection would have been produced by a candle removed to a distance of fifty-three miles. However, the real knowledge will not be gained by measuring the total radiation from stars, but by dispersing the starlight and measuring the distribution of energy in its spectrum. This is the dream of the experimenter, and it is the goal toward which he has turned his efforts. But, in order to accomplish useful results, the present investigation shows that the radiation sensitivity must be still further increased by a hundred-fold. In other words, the radiometric outfit (including the reflecting mirror) must be sufficiently sensitive to detect the radiation from a candle at a distance of more than 500 miles. The instrument must be 10,000 times more sensitive than the one used by Nichols fifteen years ago, and perhaps from 100,000 to 500,000 times more sensitive than those used in the earliest attempts by Huggins and others, almost half a century ago. This shows how insignificant has been the gain in sensitivity in comparison with what will be required in order to accomplish much on the radiation from stars. It will be a nerve-racking investigation; but it is not so appalling as the above figures may indicate. Large reflectors are becoming more common every year. However, a very large reflecting mirror may not be desirable. The gain in light-gathering power in a very large reflecting telescope is not at all proportionate to the cost of manufacture and convenience of operation. The writer used a three-foot reflector, which, although operated by hand, could be set very quickly. If a six-foot reflector had been at his disposal the sensitivity would have been increased by only four-fold. This shows how little is contributed by the mirror, and how much of the burden as regards gain in sensitivity falls upon the radiometer.

The aforementioned extra hundred-fold gain in sensitivity required may be attained by the use of a reflecting telescope having a mirror seven feet in diameter, and by increasing the radiometer sensitivity twenty times. If the sensitivity of the thermocouples used in the present work can be increased two-fold, this will leave the galvanometer sensitivity to be increased ten-fold. By using a special pier for the galvanometer, and by using a light galvanometer suspension in a vacuum, it will not be a very difficult task to increase the galvanometer sensitivity ten-to twenty-fold. It looks, then, as though everything were in our grasp—everything except a six-to seven-foot mirror, set apart primarily for astroradiometric work. When one thinks of all the money wasted in idle pleasure, and in the wars of nations, it is pathetic to realize that but for a few hundred thousand dollars the aforementioned "layman" of this generation might live to see something "practical" forthcoming from the investigation of the radiation from stars.

1. Huggins, Proc. Roy. Soc., 17, p. 309, 1868–9.
2. Stone, Proc. Roy. Soc., 18, 159, 1869–70.
3. Pfund, Publ. Allegheny Obs., 3, p. 43, 1913.
4. Nichols, Astrophys. Jr., 13, p. 101, 1901.
5. Boys, Proc. Roy. Soc., 47, p. 480, 1890.
6. Minchin, Proc. Roy. Soc., 58, p. 142, 1895; 59, p. 231, 1896.
7. Stebbins, Astrophys. Jour., 32, 185, 1910; 33, 385, 1911.
8. Galv. Sensitivity ${\displaystyle i=1\times 10^{-10}}$ Amp.