is a group of long lines of a sharp appearance at the less refrangible end of the spectrum. Rays more refrangible than Cd 171 are most numerous in the case of iron, then cobalt, and least so in nickel, and the dispersion and refrangibility of the rays increase in the same order.

The copper and silver spectra are very similar, consisting of fine but strong lines, and a scattered group of more refrangible short lines, stronger but less refrangible in the case of silver than in copper. Gold and platinum are very similar to these.

In the series magnesium, zinc and cadmium, similarity is shown in the number, shape and intensity of the lines. The refrangibility of the strong lines increases with atomic weight. There is also considerable similarity between tin and lead and the arsenic group. These spectra consist of pairs or groups of long lines with interspersed dots. The latter group-arsenic, etc.—are distinguishable by their nebulous short lines and dots.

The spectra of the alkalis are very simple. Lithium contains only 13 lines, of which 11 are long and well defined. In sodium there are 24 lines, of which 5 are long lines. In potassium we have 47 lines and only 3 long lines. The spectra of the aluminium group are characterised by dots.

Reference should be made to the original paper, where photographs will be found of the various spectra.

From a careful study of the spectrum of beryllium,

1 One of the reference lines in the cadmium spectrum.

H

Hartley was able to assign it to a place in the periodic table, which is the one it now occupies.1

Lecoq de Boisbaudran 2 found it possible, from a consideration of their spectra, to predict the atomic weights of certain elements. The method used will be illustrated in the case of germanium, for which it has received a most striking confirmation.

In the spark spectra of silicon, germanium and tin, there are two brilliant lines, and also in the spectra of aluminium, indium and gallium. The wave lengths (2) of these lines in millionths of a millimetre are as follows:

The atomic weights are Si=28, Ge=?, Sn=118, Al 27.5, Ga=699, In=113.5. Al=27.5,

Comparing the differences between the wave lengths and the differences between the atomic weights we get the following:

In the column "Variations" is stated the percentage of the first difference, which must be added to it 40.51 × 44.3

to give the second difference, thus 44·3+ 100 62.4. The following ratio gives a method for determining what the values of ▲ for the atomic weights of silicon and germanium and for germanium and tin should be:

Al-2Ga+In: Al-2Ga + In = Si - 2Ge+ Sn: Si - 2Ge+ Sn

(At. Wt.)
3.051

The number 3.051 is the percentage of the difference Ge- Si, by which this difference must be increased to make it equal to Sn-Ge, therefore first difference = = 44.32.

90 2.03051

Hence we get

Si= 28.044-32
Ge = 72.32>44:32
Sn=118-0045-68

The atomic weight as subsequently determined by Winkler was 72.3.

We thus see that there is a close relation between certain lines in the spectra of allied elements and their atomic weights. The subject, although of great interest, presents many difficulties, owing to the complexity of spectra, and has at present been little investigated.

The absorption and phosphorescence spectra of the rare earths have been employed by Sir William Crookes in his laborious and beautiful researches on

these substances. Nothing definite has, however, yet been arrived at, but it is to be hoped that it will be in the near future, when we shall be able to find the atomic weight of any element from an observation of its spectrum.

Magnetic Properties.

Faraday divided the elements into two great classes from a consideration of the manner in which they behaved when suspended between the poles of a large electro-magnet. Those which tended to set axially were called paramagnetic, whilst those which set equatorially were called diamagnetic elements. In this way he grouped the elements as follows:

Diamagnetic-H, Na, Cu, Ag, Au, Hg, Zn, Cd, Pb, Sn, P, As, Sb, Bi, S, Se, Cl, Br, I, Tl, Si.

Paramagnetic-N, O, Fe, Ni, Co, Mn, Pt, Os, Pd, Ir, Rh, Cr, Ti, Ce, C, K, U.

As thus classed, the elements do not exhibit any general relationships. We notice that similar elements, e.g., Cu, Ag, Au; P, As, Sb, Bi; Fe, Co, Ni, etc., have the same magnetic property. On the other hand, however, the closely allied elements, Na and K, O and S, N and P, Si and Ti, are of opposite properties.

In 1879 Carnelley showed that if the properties of the elements be compared with their position in Mendeléeff's Table, then those elements which are in the even series are always paramagnetic, whilst those in the odd series are always diamagnetic. This is true for every case as yet tested, and is at once obvious if the magnetic property 1 B. 12, p. 1958.

of an element be put against it in the periodic classification.

In Watt's Dictionary (1st Edit.) it is stated that elements of low atomic volume are mostly paramagnetic, whilst those of high atomic volume are mostly diamagnetic. There are, however, some very important exceptions to this, e.g., K is paramagnetic, whilst Ag, Cu, Au, Sn, Hg, Cd and Si are diamagnetic. In fact, the above statement cannot be regarded as true.

The following table gives a list of the few quantitative measurements made, with their observers. The sign+signifies paramagnetic and diamagnetic. This table gives the specific magnetic powers, that is to say the action exerted by a magnet on unit volume of the substance at unit distance, taking water = 1.