The application by Prof. Bond of photography to stellar astronomy was confined to stars brighter than the seventh magnitude, and with the hope probably of finding some encouragement of obtaining photographs of much fainter stars, he has discussed very fully in the Astronomische Nachrichten, No. 1158 and 1159, the effect of increasing the times of exposure of the photographic plates. The result which he arrives at is that "deficiency of light can be more than compensated for by a proportionate increase in the time of exposure;" and concludes that an exposure of ten minutes with the Cambridge Equatorial would give a photographic image of a star of the ninth magnitude. This, however, appears to be a question that can be decided only by actual experiment.

The last essay referred to closes Prof. Bond's work in stellar photography. He intended to resume the subject and was confident of being able to make photographic images of stars down to the ninth magnitude, but failing health and a desire to finish his work on the great nebula in Orion prevented him from undertaking any further experiments. Having participated in the work of measuring the photographic images and in reducing the measurements, and having a vivid remembrance of the labor performed, I take the liberty of stating my own estimate of the two methods of observing double stars, viz: the photographic method and that of direct measurement with a filar micrometer or heliometer. In the case of double and triple stars I have no doubt that the better method is that of direct measurement. The labor of setting the circles and finding the star is common to both methods, but during the time required to adjust properly the clock work and make a series of photographic images a practised observer would make and reduce a series of direct measurements. It is thus possible by the direct method more. easily to repeat the observation under varied conditions of atmosphere, observer and instrument, and in this way to render the final result less liable to systematic errors. It is true that according to Prof. Bond's calculations, the photographic method is decidedly the more accurate, but some experience makes one, I think, distrustful of inferences drawn from the values of probable errors. The most dangerous errors of observation are those that are constant or systematic, and these the theory of probability does not and can not recognize. It is this class of errors which is the source of the large discordances frequently seen in different determinations of the same quantity while each single determination has apparently a very small probable error. It should therefore be the care of the observer to vary the circumstances of his observation in order to avoid systematic error, or to give it as much as possible the nature of those irregular errors which in the long run tend to eliminate themselves. Finally, it may be stated as a general

rule that, other things being equal, the simpler and more direct the method of observing, the better. In order to justify the interposition of any new process it must be shown that this process gives greater accuracy or greater rapidity of observing or both. Thus the chronographic method of observing transits is justified on the ground that it gives greater ease and rapidity of observing, since the gain in accuracy is scarcely sensible.

In the case of groups of stars like the Pleiades or Præsepe there would be a great advantage in using the photographic method provided the plates could be made sufficiently sensitive, so that images of stars of the ninth and tenth magnitudes could be obtained. Mr. Rutherfurd, who has done so much in astronomical photography, has made photographs of both the groups mentioned, and from his plates Dr. B. A. Gould has deduced positions of the stars. Dr. Gould's memoir on the Pleiades was presented to the National Academy five years ago, and it is to be regretted that it is not yet published.

The first application of photography to determine the times of contact in a solar eclipse was made by Mr. Warren De la Rue in the case of the eclipse of July, 1860. In the Philosophical Transactions, of the Royal Society of London, 1862, Mr. De la Rue has given a full account of his work, together with an engraving and description of the micrometer employed for measuring the photographs. The Kew heliograph was used in obtaining the photographs, an instrument designed by Mr. De la Rue for the purpose of making photographs of the Sun's disk. By means of a Huyghenian eye-piece the image of the sun was enlarged to a diameter of nearly four inches. Two position wires at right angles to each other were placed in the focus of the eye-piece at an angle of nearly 45° to a parallel of declination, and the final correction to the position of these wires was found by observing transits of the sun's limbs after the method proposed by Mr. Carrington. The wires being photographed on the plates their images furnished a means of finding the zero of position, and the plates were measured for the position angles of the cusps and their chords, also the radii of the sun and moon and the positions of their centers. By comparing the interval of time of two consecutive photographs with the measured position of the centers, the relative motion of the sun and moon was deduced and hence the times of contact which agree tolerably well with each other, but not so well perhaps as those of the ordinary direct observation. The value in arc of a single division of the measuring scale was found by measuring on the plates the radius of the sun in parts of the scale and then assuming a value of the sun's radius. A very doubtful correction of -4""1 is deduced for the sum of the semidiameters of the sun and moon.

Mr. De la Rue speaks of having made some observations to determine whether in the process of drying the images had undergone any distortion, and says: "The result, however, proved that there was no appreciable contraction, except in thickness, and that the collodion film did not become distorted, provided the rims of the glass plates had been well ground." This point being a fundamental one appears worthy of further investigation. There is some uncertainty and vagueness about Mr. De la Rue's methods which do not give much confidence in the accuracy of his results, but on the whole he appears to have shown that the photographic method promises good results which is all perhaps that a first trial could be expected to do.

The next and last memoir published on this subject is that by Messrs. De la Rue, Balfour Stewart and Benjamin Loewy, on Solar Physics, Philosophical Transactions, 1869, giving an account of the positions and areas of the sun-spots observed with the Kew photoheliograph during the years 1862 and 1863. In order to determine the heliographical latitude and longitude of a spot two elements are required, viz: its angle of position and its distance from the sun's center at the time of observation. These elements were obtained by measuring the photographic plates, the position wires being similar to those used by Mr. Le la Rue in the solar eclipse of 1860, and their zero being determined in the same manner. This memoir contains a large amount of interesting information concerning the position and areas of the solar spots, but the determinations of position are vitiated in some degree by the optical distortion of the instrument. The observers at Kew have made experiments to determine the amount of distortion, but no definitive result has yet been reached. They say that the following facts may be regarded as established: "1st, that the image of any object photographically depicted is liable to a distortion, which varies at different distances from the center of the field, and the amount of which may be determined for every instrument by methods similar to that employed by ourselves; 2d, that in our case the image of an object is larger when formed near the edge of the field than at the center, and that the amount of elongation of a unit of length at the center increases with its distance from the center." Their conclusion is that the inquiry is not sufficiently far advanced to justify any corrections of the positions of the spots on account of the effect of distortion, but they express the hope that at last they will be able to give thoroughly satisfactory constants for the effect of displacement in their instrument. The photographs of the solar eclipse of August, 1869, have not yet been measured and discussed so far as I know, but it is pretty certain that all the images are affected by distortion which will in a measure render the results dependent and untrustworthy. It is to be hoped, however, that these pho

tographic plates will be subjected to a careful examination in order that some estimate may be made of the extent of error to which they are liable. In the case of a solar eclipse, or of a transit of a planet over the sun's disk the photographic method has very great advantages over the observations of contact in many respects, and the errors to which it is subject are worthy of the most thorough investigation. The observation of a contact is uncertain on account of irradiation, and it is momentary also, so that if lost by a cloud, or in any way, the observer is compelled to view for several hours the phenomenon without being able to observe it. On the other hand when the sky is clear a photographic image can be obtained in an instant, and even if all the contacts be lost, valuable results might be secured if the readings of the photographic plates can be correctly reduced. Just here then is the point for experiment, investigation and invention, since it is most desirable that no doubt should remain as to the possibility of correctly measuring and reducing the photographic observations of the transit of Venus.

Mar. 25, 1871.

ART. VI.-On Ralstonite, a new Fluoride from Arksut-Fiord; by GEO. J. BRUSH.

THE recent exploitation of the Greenland cryolite has not only led to the discovery of crystallized cryolite, but has given to mineralogical science several new fluorides, two of which, thomsenolite and pachnolite, are found in beautiful crystallized forms.

I now call attention to another fluoride observed, a few months since, by Rev. J. Grier Ralston of Norristown, Pa. Mr. Ralston found a mineral in minute octahedrons associated with thomsenolite, and being unable to identify it, he sent it to Prof. Dana, by whom the specimens were passed over to me for examination.

The crystals of the new mineral are octahedral; and in some cases they are very minute, but occasionally one to one-and-ahalf millimeters in diameter. They are often implanted on the thomsenolite crystals, and also apparently intercrystallized with this species, making it extremely difficult to separate the new mineral sufficiently pure for analysis. The planes of the octahedron are often tinged slightly yellow, and many of them are dull and iridiscent, owing to an excessively thin film of oxide of iron, and hence exact measurement of the inclinations of the faces cannot be made. But they appear to be symmetrical with equilateral faces, and, in some cases, have all the solid angles replaced by a minute plane. With so regular a habit and the

planes alike in lustre I cannot doubt that they are isometric octahedrons, and hence that the small plane on the angles is a cubic plane.

The mineral is colorless to white, with a vitreous luster, and has a hardness greater than fluorite, equal to about 45. Specific gravity, taken on 25 milligrams, gave 24.

In the closed tube the mineral whitens, yields water at first, then gives off fumes and a copious white sublimate, while the walls of the tube are etched. The vapor in the tube, as well as the water, reacts acid. Even at a very elevated temperature the heated fragment retains its original form and does not fuse. In the open tube the mineral gave similar reactions.

In

B.B. on charcoal a faint white sublimate was observed. the platinum forceps the mineral whitens without fusion and colors the flame a soda-yellow; moistened with sulphuric acid this coloration was unchanged. Heated with cobalt solution in the platinum forceps the mineral assumed a bright aluminablue. In salt of phosphorus the substance was readily dissolved, giving a colorless bead in both O. and R.F. In a soda bead it dissolved with effervescence.

The mineral was found on close examination with a magnifier to be so intimately associated with thomsenolite, that it was deemed impracticable to separate a sufficient quantity for any thing more than a preliminary qualitative examination in the wet way. A small portion, some 30 milligrams, selected with great care to ensure purity, was, decomposed in a platinum capsule by sulphuric acid. It gave off fluohydric acid, etching a glass plate, and on solution it gave with reagents a precipitate of alumina, with evidences of the presence of lime and soda. With the spectroscope, the pure mineral alone gave only the soda line, but when even a very minute speck of thomsenolite was associated with it, the lime line was very marked. The solution from which alumina was separated afforded on evapora tion to dryness a minute residue, which, when examined with the spectroscope, gave both soda and lime lines.

It therefore appears that the mineral under examination is essentially a hydrous fluoride of aluminum with probably a small amount of calcium and sodium. Its isometric form, and its infusibility, distinguishes it from all other fluorides yet described as occurring at the cryolite locality in Greenland.

The only mineral which in general chemical characters approaches it is the fluellite from Stenna-Gwyn in Cornwall. This rare mineral, according to Wollaston, is a fluoride of aluminum, but it occurs in elongated orthorhombic octahedrons. These facts are, I think, sufficient to warrant the conclusion that the Greenland octahedral mineral is a new fluoride, and I propose for it the name Ralstonite, after its discoverer.