of a grain, or thirty times that given for sodium. By the time the lithium line is established the red potassium line at 234 and occasionally the violet line at 135° appear, and the blue and green fields become divided into bands which are so rapidly resolved into bright and dark lines, that it is difficult to note the exact time of the appearance of each. The spectrum increases to a dazzling brightness, and extends itself in both directions until it reaches from 231° to 140°.

During the third period the spectrum becomes more brilliant, and the lines more distinct. Several new lines make their appearance in different parts of the spectrum, of which the ones at 5110, 57°, and 67° are well defined, while others are faint and not always visible; some of them appearing only toward the close of the last period. In viewing the lines in the most refracted part of the spectrum, it has been repeatedly observed both by myself and others, that these lines were more strongly marked when entering the eye at an angle than when viewed directly. That this was not imagination is proved by repeated identification of lines at the same point on the scale.

At the termination of the blow the lines are rapidly swept away, sometimes in the inverse order of their appearance, but more generally they disappear within the space of two or three seconds, leaving a continuous spectrum as at first, though somewhat brighter. Sometimes the sodium and lithium lines are swept away with the others, and at other times they remain visible. In either case the change is very decided, and does not generally occupy more than three seconds. In the course of my observations, thirty-three lines have been detected, as given in the table below.

Some of the lines given by Lielegg I have failed to find, but have detected others not given by him.

1st Period, 234, 35, 50, 135.

2d Period, 234, 35, 43, 44, 44, 45, 46, 47, 48, 50, 52, 53, 56, 56, 61, 62, 62, 63, 65, 66, 67, 70, 72, 120, 135. 3d Period, 234, 35, 43, 44, 44, 45, 46, 47, 48, 50, 51, 52, 53, 56, 56, 57, 61, 62, 62, 63, 65, 661, 67, 671, 70, 72, 100, 102, 103, 105, 108, 135.

Among the dark bands detected, the most intense occurred at 44-46, 51-55, 56–58, 62–64; others were found at 33-34), 36, 37, 38, 40, 68-72.

Many of the dark bands were crossed by bright lines.

I have repeatedly observed the dark band considered by Roscoe to be a hydrogen absorption line, but have not noticed that its intensity varied with the dampness of the weather. Whether it is an absorption band or not can be determined by a series of observations continued through wet and dry weather. If this proves to be a hydrogen line, the Bessemer spectrum will be found more complicated than is generally supposed. It has

been thought by some that the dark bands in the spectrum are absorption lines due to the cooling of the outer sheath of flame, but it is more probable, that although the pellets of iron and slag tend to produce a faint continuous spectrum; yet in contrast with the very brilliant lines it appears discontinuous, the dark bands being merely intervals between the bright ones. The iron spectrum has not been satisfactorily identified. It has been suggested that the brightness and size of the lines of the Bessemer spectrum do not allow the iron lines to appear. In comparing the Bessemer spectrum with Bunsen's spectra of nickel, cobalt and calcium, no coincidences were observed except two or three in the latter spectrum. brightest calcium line, however, was not visible in the Bessemer spectrum. The Bessemer spectrum contains yet many mysteries to be solved, among which is the cause of the non-appearance of the lines of the spectrum at the beginning and termination of the blow.

The

This was readily solved when the numerous lines of the spectrum were attributed to carbon, but in proving them to be caused principally by manganese, their disappearance is not so readily accounted for.

One theory to account for it is that the luminous power of the flame is too small at the beginning and end of the process to produce a spectrum. In regard to this it may readily be shown that the brilliancy of the spectra of incandescent metallic vapors does not depend upon the illuminating power of a flame but upon the heat of the flame into which they are introduced. For instance, the spectra are more distinct in the non-luminous flame of a Bunsen lamp than in the ordinary luminous gas-flame. If we take the theory as referring to the feebleness of light given off by those substances in the flame which produce the spectrum, it will resolve itself into the one of change of temperature, notwithstanding the fact that the illuminating power of flames of the same temperature varies with the composition of the gas, because there is evidently enough sodium in the flame to give its characteristic line; hence, whatever might be the illuminating power of the flame, if the heat is sufficiently intense the sodium line will show itself.

Dr. Wedding adopts the theory that the absence of the spectrum at the beginning and termination of the blow is because the absolute quantity of the bodies volatilized producing the spectrum is at these times too small. His reasons for holding this view are as follows:-"A trace of sodium will give its characteristic line, but, according to Simmler, a much larger quantity of manganese is needed to obtain a recognizable reaction than that which can be detected by the well known blow-pipe reaction with carbonate of soda. Consequently, spectrum analysis does not depend alone upon the presence of a body but also upon the

presence of a certain quantity. And although manganese is always left in the iron, it may not be left in sufficient quantity at the termination of the blow to produce the spectrum, and for this reason the lines disappear."

To this theory there are same some strong objections. 1st. If we take manganese in sufficient quantity and hold it in a flame. the spectrum will increase in brightness until a uniform temperature is attained; but when the amount of manganese vaporized begins to diminish, its spectrum will gradually decrease in brightness until it disappears. Now, if the disappearance of the manganese lines in the Bessemer spectrum is owing to the diminution of the quantity of manganese, we should infer that these lines would gradually grow more indistinct and then fade away; but on the contrary, the manganese spectrum increases in brilliancy from its first appearance, and is more intense just before being swept away than at any other time. The analysis of the smoke, which appears when the flame ceases, proves that a considerable quantity is still volatilized, and it is notable that in manganiferous iron this quantity increases towards the close of the blow. 2nd. It would be more difficult to account by this theory for the non-appearance of the sodium line at the beginning of the blow, as sodium then in all probability exists in the issuing gas in sufficient quantity to produce its spectrum at a high temperature, as it is only by special precaution that we can keep it out from any flame. 3rd. A still greater difficulty would arise in applying this theory to the spectra of sodium and lithium at the close of the blow. As has before been stated, these lines sometimes disappear at the moment of complete decarbonization, and sometimes remain. In the former case, to say that our friend sodium had given out would be doing great injustice to that element, as it has never given us reason for bringing so grave a charge against it. Dr. Wedding in attempting to demonstrate that the non-appearance of the manganese lines is owing to the lack of sufficient quantity volatilized to produce its spectrum, makes the following statements:

From analyses made by Brunner we find that the manganese contained in the iron falls from 3:460 per cent in the raw material, to 1645, 0·429, and finally to 0.113 per cent in the decarbonized product; and that the protoxyd of manganese in the slag first increases from 3700 per cent to 37.90 per cent, and then sinks to 32-23 per cent, and furthermore, that a certain quantity of manganese is to be found in the smoke. How much manganese is really lost by volatilization cannot be determined, since data are wanting as to the absolute quantity of slag and iron, consequently we cannot determine how much manganese has been lost by means of the eruptions.

But since the manganese contained in the pig-iron decreases

constantly, and that contained in the slag after the termination of the boiling period also decreases, a considerable volatilization of this body is probable just at the time when the spectrum is best developed. Comparing with this the experiments that can be made in the laboratory we arrive at the hypothesis, that the oxydized manganese which has entered into the slag is not volatilized but is retained by the slag; it can, therefore, get into the flame only in the shape of solid or fluid combinations.

In the above statements the results of the analysis prove that some of the manganese in the slag is volatilized. We cannot consider the manganese spectrum during the entire process as due wholly to the volatilization of the manganese directly from the iron, for while the amount eliminated from the iron grows continually less, the manganese spectrum grows brighter. Owing to the intimate mixture by the blast of the iron and slag, the manganese oxyd contained in the latter, is brought in contact with the melted iron and vaporized. This mixing of the slag and iron would cease at the termination of the process, and this would account for the sudden diminution of smoke.

If there was a sufficient carbonic oxyd flame to render the escaping gases glowing it is evident they would not issue from the converter as dark smoke, but as incandescent vapor having its characteristic spectrum. The lack of sufficient flame may, therefore, account for the disappearance of the manganese spectrum. The Bessemer flame presents other problems, and opens an intensely interesting field for scientific investigation; and by the use of more delicate instruments than have yet been employed for this purpose, discoveries may be made which will throw new light upon the subject of spectrum analysis.

ART. XXIX.-On a simple method of measuring Electrical Conductivities by means of two equal and opposed magneto-electric currents or waves; by ALFRED M. MAYER, Ph.D.

[Read before the Troy meeting of the American Association for the Advancement of Science.]

1. General description of the Method.

A MAGNET is firmly supported in a horizontal position with a portion of its length projecting beyond a fixed stop (see fig. 2); over this free end of the magnet, and resting against the stop, are placed two similar flat spirals, formed of the same quality of copper wire, and having the turns of one spiral in a direction the reverse of those of the other. The spirals are clamped together and their four terminal wires are carried vertically downward into four separate cavities containing mercury; these mercury-cups are so connected with a reflecting-galvanometer

that, when the spirals are together slid off the magnet, the two equal electric currents, thus generated, simultaneously tend to traverse the galvanometer in opposite directions, and therefore its needle remains stationary. If we now introduce into the circuit of one of the spirals a resistance equal to that introduced into the circuit of the other, the needle will still remain at rest when the spirals are slipped off the magnet; but, if the resistance placed in one circuit is greater or less than that placed in the other, there will be a deflection of the galvanometer needles when the spirals are removed. Thus, by introducing wires of different metals into the circuits we can readily determine their relative conductivities, by making them of such length that their resistances are equal; which condition is attained when, on sliding off the spirals, the needle remains absolutely at rest. If, in the latter case, the wires have equal diameters then their conductivities are directly and their resistances are inversely as their lengths.

A modification of the above method is discussed in the conclusion of this paper; in which the magnet is replaced by the terrestrial magnetic force and the spirals and the wires by two similar coils, from two to three feet in diameter, formed of the two wires whose conductivities are to be compared. These coils contain equal lengths of the same sized wires and the same number of turns; the direction of the turns being opposed in the two coils. The coils having been bound together are placed in a plane at right angles to the line of "the dip," and the four terminal wires are so connected with the reflecting-galvanometer that the two induced currents tend to traverse it in opposite directions. The coils are now quickly rotated through 180°, around an axis at right angles to the line of the dip, and if the wires present equal resistances the needle remains at rest; if it is deflected, the direction and the amount of the deflection shows which coil has the lesser resistance and affords a means of estimating the same.

After this general description of the method I will present, in order, a description of the apparatus used, and of the actions which take place in it; the degree of precision of the method; examples of the determinations of electrical-conductivities, and experiments on the modification of the method.

2. Description of the Apparatus.

The magnet was formed of a combination of three steel bars, separated from each other by slips of wood 2 in. thick. The middle bar was 104 in. long and its ends projected 25 in. beyond the two side magnets. Each bar was 27 in. thick and 9 in. wide. About three months before this investigation was undertaken they had been magnetized to saturation by the