convenient spectroscope. The collimator and observing telescope may be fixed at an angle of 90°. When the prism is rotated the spectral lines will always be at the position of minimum deviation at the moment of crossing the reticule in the telescope. They will thus always be of maximum sharpness.

ment.

The operation of setting for minimum deviation of a line is accomplished by a single motion and the fixed position of the telescope simplifies the construction and the use of the instruIt is only necessary to have a good achromatic objective in order to be able to focus the photographic camera, for example, on the D line, and then, by simply rotating the prism, to bring a given region of the ultra-violet at minimum deviation to the point previously occupied by the D line.

It is evident that the prism just described causes the ray which has traversed it at minimum deviation to turn through an angle of 90° to the right. We will call this prism right-handed. If the same prism worked in the opposite direction, receiving the light on its large face and at a suitable angle, it would be lefthanded. Or such a prism may be considered to be reversed top for bottom, in which case the new prism is right-handed when it receives the light on its large face and left-handed in the inverse direction.

The condition of fixed deviation may be met with the greatest simplicity in the construction of a spectroscope having several prisms. It is evident that if we wish to obtain dispersion we must join in pairs prisms which are opposite in character.

Each of these prisms will produce a deviation of 90°, either in one direction or the other, depending upon whether it is made to work in this or that direction. Let us consider the possible solutions for various numbers of prisms.

For a one prism spectroscope either position of the prism may be employed. With two prisms we may obtain either direct vision with displaced path (Fig. 2), or parallelism of the collimator and telescope, as shown in Figs. 3 and 4. With this arrangement the observer is distant only some 20 cm from the slit and source of light. When a telescope of longer focus than

the collimator is employed, the light source and slit may be placed at any desired distance. In both cases the motions of the prisms. are effected by causing them to turn about fixed centers lying upon a perpendicular to the direction of the telescope, by means of two rods moved by a single fixed screw.

If three prisms are used, the collimator and telescope will lie at right angles, either to the right or left. The three prism spectroscope may be of the form (2), (3) or (4). Form (4)

is most convenient.

Finally, four prisms may be used, in which case we may have either direct vision (5), or an arrangement like that shown in Fig. 3, with the slit close at hand. For each of these forms there are two solutions, shown in (5), (6), (7), (8).

In all of these solutions, in order to keep the prisms in the required position, it is only necessary to arrange them in such a manner that alternate prisms will rotate in opposite directions. This is easily accomplished by means of a tangent screw having right and left threads, or, in the case where the slit and eyepiece closely adjoin, two right and two left threads. It is evident that the faces of the prisms are always completely utilized.

It is further evident that by setting the telescope in position determined in advance by the aid of a scale, it is possible to observe with (p—1) prisms, if that of order p is removed.

In this instrument a spectral line is defined by the position of the prisms at which it crosses the reticle. The angle through which the spectrum is moved by a rotation a of the prisms is 8a. The amount of this rotation may be determined with a precision equal to the sensitiveness by means of the image of a micrometer reflected once on each of the refracting prisms. We have realized this by means of total reflection prisms placed above each of the first three refracting prisms. A double total reflection prism brings down the ray so as to utilize as a reflecting surface the last refracting face of the last prism. Thus the observing telescope may be used to measure the rotation. If, moreover, there is any error in the motion of the prism, the readings will be affected in exactly the same way as the displacement of the

spectrum, and a line will always be defined by that division of the micrometer with which it coincides at the desired point.

In arrangements (5) or (7) we may have a motion of the micrometer in the direction of the spectrum, if the rectangular prism giving double total reflection is fixed. If it partakes of the motion of the fourth dispersing prism the motion of the micrometer will be in the opposite direction to that of the spectrum. The pointing of the micrometer wire will not be affected by this; but the system having direct motion is preferable.

Of these instruments we have constructed type (5), which performs perfectly. With the instrument it is possible to make a general examination of a region of the spectrum with a single prism and then to study it in detail with the four prisms. In order to realize the dispersion of eight prisms it is only necessary to employ the principle of auto-collimation, by placing a mirror in the position previously occupied by the collimator.

These instruments seem to be well suited for the use of chemists, as they will permit the use of greater dispersions than are now employed, on account of the possibility of determining positions with a precision of the same order as the dispersion.

RESEARCHES ON THE ARC-SPECTRA OF THE

METALS.

V. SPECTRUM OF VANADIUM.'

By B. HASSELBERG.

INTRODUCTION.

In my previous memoirs under the above title I have sought to bring to a temporary conclusion the spectroscopy of the metals nearest to iron. In connection with the spectroscopic researches published in the meantime by Kayser and Runge, and by Rowland, these will satisfy the most pressing necessity in this department of spectroscopy. But more than this they will not do. Only the first step has so far been taken toward the final goal, the complete knowledge of the spectra of the metals, since in spite of all our efforts surely no one of us can cherish the hope of having disentangled more than the principal features of these complicated phenomena. Many peculiarities of a doubtful character in the preceding results will first find their true explanation when similar investigations shall have been carried out for the whole series of the so-called rare metals, which are often widely disseminated, although in limited quantities. An important part of this problem is indeed accomplished in the recent large catalogue of solar spectrum wave-lengths by Rowland; in view of the fact, however, that many of these metals are only meagerly represented in the general solar spectrum, it will not appear unreasonable to undertake this investigation again and to continue it as far as possible.

I was led to give my attention first to vanadium among these metals by the circumstance that Baron Nordenskiold had received a good-sized piece of this metal from Moissan, who had prepared it in the electric furnace, and which Baron Nordenskiold had

1 Kongl. Svenska Vetenskaps-Akademiens Handlingar, Bandet 32, No. 2. Stockholm, 1899.