The spectrum of silicon has been the subject of numerous investigations, of which a convenient summary to the year 1912 has been given by Kayser. The spectrum is of special interest on account of the remarkable series of variations which accompany changes in the experimental conditions. These changes were first systematically investigated by Lockyer, who further showed that the different groups of lines which were developed with increasing intensity of discharge appeared in the spectra of stars following each other in order of increasing temperature. Four distinct groups of lines were recognised by Lockyer, namely:— Group I. λ λ 3905·8, 4103·2 Group II. λ λ 3853·9, 3856·1, 3862·7, 4128·1, 4131·1, 5042, 5057. Group III λ λ 4552·8, 4568·0, 4574·9. Group IV. λ λ 4089·1, 4116·4.§ The lines of Group I correspond to the arc spectrum, and are most strongly developed in stars approximating to the solar or G type. The chief lines of Group II are produced in the ordinary spark spectrum, but the fainter lines are best obtained in vacuum tubes; they have their greatest relative importance in stars of types F and A. Lines of Groups III and IV are also most effectively observed in vacuum tubes, and are produced in turn as the energy of the discharge is increased; they occur in successively higher stages of the stellar sequence, those of Group IV not appearing until the hotter B stars are reached.