from the pole of the superposed magnet. Thus we see how the same atoms endowed with forces of the same strength, may take different relative positions, and thus produce very different crystal-forms in the same matter. We may take 5a for an illustration of the atomic arrangement in the diamond, while 56 may stand for graphite. But there is always a change of density accompanying the different forms in allotropy, and this fact is also illustrated by configurations 5a and 5b. In bodies formed of the same kind of elementary atoms, as in allotropy, it is evident that their relative densities will be directly as the number of atoms contained in the unit of volume. As our configurations illustrating allotropy contain the same number of magnets, it follows that the relative densities of these configurations are inversely as their areas. Now the area of 5a (measured on the original prints) is 818 square millimeters, and the area of 5b is 992 square millimeters, hence the density of 5a is to the density of 56 as 992 is to 818. Thus we see how the arrangement of magnets in 5a may stand for the molecular structure in the diamond while 5b may stand for that in graphite.

Numerous instances exist in chemistry of the same elements combined in the same porportions, yet producing bodies crys tallizing in different forms, and having different densities, color, transparency, hardness, etc. As examples of this phenomenon of isomerism we may cite calcium carbonate, which crystallizes in two forms differing in density, viz: as calc spar of a specific gravity of 2.72, and as aragonite of a specific gravity of 2.93. Configuration 6a may stand for the molecular structure of calc spar, while 66 may stand for that of aragonite. The relative densities of these two configurations are as 208 to 247.

A striking example of isomerism is given in titanic acid, which crystallizes in three distinct forms: as anatase, specific gravity 382; as brookite, specific gravity 402; and as rutile, specific gravity 4.25. These three isomers may be illustrated by 8c, 86 and 8a, which have respectively the densities of 382, 364 and 360.

It will, of course, be understood that the above paralellisms are given merely as illustrations of how our experiments may serve to explain and illustrate the phenomena on the assumption of the atomic hypothesis and on the supposition that the actions which, in the experiments, take place in a plane, may similarly take place among repelling and attracting points situate in space of three dimensions.

Other forms of the Experiments.-Instead of floating the magnets they may be suspended by fine silk fibers. In this method of experimenting the attractive action of the superposed magnet is replaced by the action of gravity, which draws the mutually repellant needles toward the vertical.

An advantage of this form of the experiment is that the configuration can be transported, and may thus serve in illustration of a moving molecule as is set forth in the kinetic theory of gases. It is interesting to watch the mutual actions of two or more approaching configurations, and to observe the motions in the exterior and in the contour of a suspended configuration on its impact against a resisting or a yielding surface.

Professor O. N. Rood suggested to me to replace the sus pended magnets by gilded pith balls, hung by silk fibers and similarly electrified.

Professor Frederick Guthrie, of London, under date of May 21, writes: "If the corks are made somewhat wider than in your larger needles, the needles move and arrange themselves very quickly if they are turned over and floated on perfectly pure and freshly filtered mercury. Those which reach the edge incline with their corks in the capillary trough."

1.

B

M

R

Method of projecting the magnified images of the experiments on a screen. To exhibit these experiments before a large audience it is best to use short magnets made as follows: Magnetize rather large sewing needles with their points all of the same polarity, then take each needle between the flat jaws of a pair pliers and with a pair of cutting-pliers snap off the needle close to the jaws of the other pliers. Thus form a series of magnets about 4 inch in length. Run each of these through a thin section of a small cork and then coat both needle and cork with shellac varnish. Float these magnets in a glass tank placed over the condensing lens of a vertical-lantern, or you may even float them directly on the condenser itself if this is

made of an inverted glass shade filled with water. This form of condenser was first used by Dr. R. M. Ferguson, of Edinburgh.

Fig. 1 shows the arrangement of the experiment. The rays of light, R, from a heliostat, or from an oxyhydrogen light, fall on an inclined mirror, A. placed under the water condenser C. The needles float on the surface of the water in this condenser. The rays which have passed through the lens, L, are reflected by the swinging mirror, B, to the distant screen, where they form the images of the floating magnets. The magnet is held over the needles at M, by means of a wire which is wrapped round the magnet to serve as a handle. If a long magnet be used it will work well if its pole is brought over the needles* by inclining it.

These experiments with floating magnets give forcible presentations of the reign of law. It is indeed quite impressive to see order being evolved out of chaos as we hold a magnet over a number of needles, carelessly thrown on water, and witness them approaching and, one after the other, entering into the structure of that geometric figure which conforms to the number of magnets composing it.

ART. XXVIII.- On the presence of Dark Lines in the Solar Spectrum, which correspond closely to the lines of the Spectrum of Oxygen; by JOHN CHRISTOPHER DRAPER, M.D., LL.D., Professor of Natural History in the College of the City of New York.

THE measurement of the wave lengths of the dark lines of the solar spectrum obtained by photographs, and the construction of a chart of the same, has for many years occupied my leisure time. As a result of the investigations connected with this work, I have arrived at the belief that oxygen as well as other non-metallic gaseous elements are represented in the solar spectrum by dark lines, in the same manner as metallic substances. The lines in the case of oxygen are however very faint, when compared with those produced by metals in the vaporous state.

The apparatus employed in these investigations may be briefly described as follows: 1st, a spectroscope for photographing the normal solar spectrum. As my purpose was to obtain photographs in which the positions of the lines should be as true as possible, I resorted entirely to the process by reflection, and at no time did the solar rays pass through glass;

* The magnetic needles in the experiments may be replaced by pieces of soft iron wire, which will be magnetized by the induction of the superposed magnet.

all error that might arise during refraction was thus avoided. The mirrors of the heliostat were of flat glass silvered, the silver surface being polished served as the reflector. The surface of the concave mirror employed to bring the image of the slit to a focus, was also silvered and polished. Gratings of 4800 and 9600 lines to the English inch, ruled on glass by a machine constructed by myself and my assistant Mr. Sickels, and also an admirable one of 17,280 lines to the inch, for which I am indebted to Mr. Rutherfurd, were used. These were silvered with a thin coating, and the unpolished silver surface employed to give spectra by reflection. With the 4800 line gratings the photographs were in the 1st and 3d orders; with those of 9600 lines in the 3d order, and with 17,280 in the 1st and 2d orders. The accuracy of the gratings was tested with satisfactory results by taking photographs in equivalent orders of spectra on each side of the normal. The photographs for the determination of the wave lengths of the solar spectrum were in sections of eighty to one hundred and fifty wave lengths. The gratings were adjusted to the line of no deviation for the center of each section of the spectrum, as it was photographed.

The wave lengths of the lines of the spectrum were carefully measured on the original photographs, by projecting them upon a scale of wave lengths, each wave length being five millimeters in extent. The scale was drawn upon slips of paper, which had been glued to strips of well-seasoned pine wood cut with the grain. The lantern used for projection was that described in this Journal for April, 1878. The distance of the lantern from the scale, and the consequent magnifying power, was so adjusted as to make the leading lines of the photograph coincide with the same lines of Ångström, drawn in their proper position below the scale as is shown in the diagram given later on. Thus the positions of the lines in each section of one hundred or more wave lengths were all made visible at once, and the errors in Ångström's chart corrected. From 4045 to O in the ultra-violet the leading lines of Cornu were employed. Among the advantages presented by this method of studying and measuring the lines of the spectrum we may mention the opportunity offered for several persons to inspect at the same time the details of the section under examination, and submit them to intelligent discussion. To this we may add the facilities offered for comparing many photographs with each other by marking below the scale the peculiarities of one, and then projecting the others in order upon the marks made. In this way the effects of duration of exposure and manner of development of the image, together with the variation in the size of the slit and focal distance may be investi

gated, and their action on the details of the picture determined. Pictures may even be placed face to face, one a little above the other, and examined in that position by projection. From the measures thus obtained a chart of the spectrum was constructed, which extended from E in the green to P in the ultra-violet. The values assigned to the wave lengths in this chart are those of Angström, and it is my purpose to present the positions and characters of certain of these lines in this communication.

The great increase in the number of lines in the chart made from photographs by Mr. Rutherfurd's grating, compared with that of Ångström led me to collect all the measurements of spectrum lines of elements that I could find, for the purpose of determining the character of the newly measured lines. On comparing the lines of the spectra of oxygen, nitrogen and air, as given in Watt's index of spectra, from the researches of Thalén, Huggins and Plücker, I was struck with the number of approximate coincidences between the wave lengths of oxygen lines and those of dark lines in my map. Attempting to make a close comparison of the oxygen with the solar lines I was confronted by the following difficulties, viz: The measurements of Thalén, Huggins and Plücker were given in wave lengths only; fractions being omitted altogether. Error amounting to half a wave length could therefore exist in the position of a line, according as it fell on one side or the other of a figure on the scale expressing a wave length. In the values given to the air lines by Ångström in his chart, this difficulty did not exist; I therefore attempted the use of Ångström's values, employing the work of Huggins and Plücker, to separate as far as possible the oxygen from the nitrogen lines. This operation was, however, quickly discarded; because of the great differences existing between these authorities regarding the wave lengths of a number of oxygen and air lines. To obviate this trouble, I made photographic measurements of the lines of the electric spectrum of oxygen by the method

detailed below.

The apparatus employed consisted of a spectroscope with two flint glass prisms of 60°, adjusted to the minimum deviation of D'. Collimator and telescope objectives, achromatics of ten inches focus. This was used to make photographs of the spectra given by the condensed electric spark in oxygen, in air and in nitrogen. When so employed the eye-piece of the telescope was removed, and a camera placed in its stead. The slit was sometimes made as narrow as was possible. The induction coil was one of Ritchie's, giving a ten-inch spark, and having a hammer current-breaker driven by clock work. The battery was three two-gallon bichromate cells, the elements