AN ACCOUNT OF AN EXPERIMENT MADE TO DETERMINE WHETHER GRAVITATION FORCE

VARIES WITH TEMPERATURE.

BY A. E. KENNELLY, Edison Physical Laboratory, Orange, New Jersey.

RESULT. Negative. No change was detected in the weight of a platinum wire between the temperatures of 20°C., and bright red heat, say 800°C., the limit of certain detection being 0.35 %, or one part in 300.

So far as our present limited knowledge of gravitation extends, this force varies only with the distance of the material particles considered, and is independent of every other condition of matter, space or time. This singular degree of constancy, so remote from the conditions in which we find other physical forces, suggests either a discrepancy in our data, or a lack of evidence bearing upon this vast subject, whose very cause is still so obscure.

The experiment of weighing a body, at different temperatures, to ascertain if any variation in weight takes place depending upon thermal conditions, has probably been repeated more than once since the days of the Phlogiston controversy, and of VOLTAIRE. The absence, however, of general information in text-books upon the limit to which any such research has been carried, prompted the experiments about to be detailed, whose result, although negative, may either save others the time necessary to ask Nature the same question, or induce them to seek their answer within a closer degree of accuracy than that here stated.

An incandescent electric lamp was placed on the pan of a delicate chemical balance, and its terminals placed in communication with a dynamo by means of a pair of fine copper wires, just heavy enough to carry the lamp current without fusing, and dipping vertically into mercury cups suitably placed beside the pan. This arrangement, of course, materially reduced the sensibility of the balance, not only on account of the mechanical damping of vibration, and the weight of the copper wires varying with the depth of mercury immersion, but also on account of the capillary force brought into play at the contact surfaces. By carefully observing the extremity of the pointer through a lens, however, a very fair degree of accuracy could be attained. The lamp was first balanced at the nor

mal temperature, and then, with the balance true, the electric current was forced through it, bringing it to incandescence. The connections and wires were easy to arrange in such a manner that the vertical component of the electro-magnetic forces brought into play would be insignificantly small, except the repulsion between the fine copper electrodes and the mercury into which they dipped, an influence that did not seem to be capable of elimination, and whose effects upon the observations could fortunately be differentiated and allowed for.

The first lamp tried was an Edison bamboo carbon lamp of special construction, having a comparatively heavy and spiral filament in ten spires 0.75 cm. in diameter, the filament itself being of rectangular cross-section 0.028 X 0.048 cm., or 0.00133 sq. cm. area. The length of the filament was about 30 cms., its surface 4.5 sq. and its weight 0.0405 gramme.

cms.,

The weight of lamp and connecting wires together, (with an inverted glass beaker placed over all as a cover to check convection air currents) was 124 grammes, and with this load the balance readily indicated too of one milligramme, the arms being 22 cm. long, the pointer 30.5, and the period of one double vibration 40 seconds. The immersion of the fine copper electrodes in the cups reduced the limit of appreciation to 2 milligrammes, and the period of double vibration to 3 seconds. The electrical pressure brought to bear upon the lamp was 120 volts, the current being almost 2 amperes. This brought the lamp to vivid incandescence within one second. The duration of current application varied in different trials between 5 and 30 seconds. No change was ever noticed within five seconds of closing the lamp circuit. After that time a slight indication of lessening weight in the lamp pan would show itself, increasing steadily with the duration of current flow. In most cases this apparent loss of weight tended to a maximum (equivalent to about 15 milligrammes), some 150 seconds after the current was cut off, when the effect slowly diminished, and a tendency to restore the original balance set in. This behavior, coupled with the fact that no change in the above order of events could be detected with the current through the lamp in the opposite direction, pointed to convection as the source of variation. Under the circumstances, a change in weight due to any effect in temperature upon the weight of filament; would have been detected probably to the extent of ± 2 milligrammes, and almost certainly had it amounted to 3 mgms. The experiment was, however, very inconclusive, since the total weight of filament being 40.5 mgms., the limit of certain detection was only 7.5% of

all, and carbon, owing to its lightness, is a material really unsuited for the experiment.

A special platinum wire incandescent lamp was then tried. This wire was 51.3 cms. long, 0.032 cm. in diameter, and was coiled into a helix of 18 spires. Its weight was 863 milligrammes. The glass globe, connections and general arrangements were all similar to those preceding. The lamp and beaker were balanced by 131.3 grammes. The fine copper electrodes were 0.0065 cm. in diameter, and were separated by a distance of 0.5 cm. With careful immersion in the mercury tubes, the limit of appreciation of balance was. I mgm. The current forced through the lamp was slowly increased in successive trials from 1.5 to 2.9 amperes, which raised it to bright redness. It was found impracticable to raise the temperature beyond this stage, since the rigidity of the platinum threatened to depart, rendering collapse of the heavy spiral imminent. The results obtained in this case are, therefore, limited to the stage of temperature corresponding to bright redness, assumed, in the absence of more precise knowledge, to be about 800°C.

On closing the lamp circuit a change representing loss of weight in the lamp pan would generally be detected within two seconds. This appeared to remain stationary a few seconds longer, (at about 1 mgm. equivalent), and then a further steady diminution set in, progressing with the duration of current application, and generally finding a maximum (equivalent to 10 or 20 mgms.), after the current's cessation. The duration of the flow varied from 10 to 40 seconds. With the lower strength of current first adopted, 10 seconds was required to bring the platinum wire to red heat, while, in the latter. trials, the stronger currents produced visible red heat in four seconds. The facts that :

(1.) No change of weight was observed coincidently with the rise or fall of temperature in the wire,

(2.) The change of weight apparent took place before any high degree of temperature was attained, and lingered long after the reduction of that temperature,

(3.) The absence of any apparent relation between these phenomena and the direction of current in the wire,

all suggested that the belief that the changes in balance observed were due in the first stage to a steady electro-magnetic repulsion between mercury cups and the immersed electrodes, and, in the later stage, to the expansion of air round the lamp by heat generated within it.

The limit of probable detection in the weight of the lamp would have been 1 mgm., while a change of 3 mgms. would hardly have escaped observation, a percentage of 0.35, or nearly one part in 300.

It would, of course, be quite possible to reduce considerably this outstanding limit of uncertainty, by making the necessary weight in the lamp wire, and with increased care in mounting the same.

Appendix. The electro-magnetic repulsion acting in these experiments can be approximately computed from a formula due to CLERK-MAXWELL.

where f is the force of repulsion in dynes, c is the current flowing through the circuit in absolute measure, d is the distance separating the axes of the wires, in cms., and r is the radius of the wires in cms. In this case c is 2.9 amperes, or 0.29 units, d is 0.5 and r is 0.0033. Whence f is 0.84 dyne, equivalent to 0.86 milligramme weight, or of the same order as the first effect noticed. EDISON LABORATORY, ORANGE, N. J., 29th Oct., 1890.

CORONAL EXTENSION.

By C. M. CHARROPPIN, S. J.*

The most distant stretch of the streamers of future eclipses, in all probability, will never be recorded on a sensitive plate; because, since the coronal rays diminish in brightness as they recede from the sun, a limit must be reached, when the faintest beams will be of equal brightness with the illuminated air: then, and only then, will they fail to impress themselves on the photographic plate: for the haloid salts of silver deal only with lights and shadows. They will delineate the most delicate pencils of light: they will record the very line where the faintest ray is immersed in the tiniest shadow: but when contrast ceases to exist, they at once become dumb. The faintest beam of nebulous matter, which the large eye of the most powerful telescope refuses to reveal, leaves its impression on the sensitive plate but the same silvered film will often fail to notice the brightest of Jupiter's satellites in transit, when projected on the planet's disc. Why, then, should the almost beamless nebula be * Professor of Astronomy, University of St. Louis, Missouri.

photographed, and a bright satellite leave no impression? Because, in the first instance, there was contrast between the faintest pencil of light and a dark background; in the second place, this contrast was wanting; the two discs reflecting nearly the same amount of light.

Thus far I have advanced no new theory. I have stated a fact which Prof. HOLDEN has alluded to in his article on the eclipse of December 21st, a fact well understood by the scientific fraternity at large. If I have insisted upon this point, it is because I consider it the key to the solution of the problem of photographing the extension of the outer corona.

Much has been said and written concerning the instrument best adapted to the photographing of the corona-whether the reflector will give better results than the refractor-whether large objectives are preferable to smaller ones, etc. Again, the question of diaphragms has been much discussed; and, as to the limit of proper exposures, the conflicting opinions are, "Sine fine dicentes." But how little has been said about the proper development of the photographic plate after exposure, and yet success depends principally upon this trying operation!

It is one thing to make a proper exposure, and another to secure, in the development of the latent image, its most delicate details. It requires judgment and a practised eye to know when to retard and when to accelerate the developer. If the negative is under-developed, the niceties of the faintest shadows are not brought out. Let the negative be over-developed, then those minute details and slightest contrasts are all lost in one common blur. Do you desire to bring out a cloud effect in a landscape, then your exposure must be very short; nay, instantaneous; but you are apt to lose the finest details of the woodland scene. If, on the contrary, you are endeavoring to bring out the softest contrast of the tinted leaves of that autumnal forest, your exposure will be longer, your developer will be retarded by Potassium bromide, but, meanwhile, your clouds will be lost in a perfectly black sky. There lies precisely the difficulty of photographing a total eclipse of the sun. You aim at capturing the full extent of coronal streamers, and, at the same time, you expect the polar filaments to be well defined on your cliché. In the one field a perfect negative is to be obtained of four different light-giving objects, differently illuminated, requiring four different exposures, viz: the actinic protuberances; the bright inner corona; the polar filaments; and the extremely delicate shadings of the extended

See Publ. A. S. P. vol. II, pages 94 et seq.