360 CARNEGIE INSTITUTION OF WASHINGTON. 



the remainder were, for the most part, specially prepared for this research by 

 Mr. Tanaka. 



(2) Each dissolved metal was found to have its characteristic luminescence 

 spectrum, composed of one or more series (usually two or four series) of over- 

 lapping bands. 



(3) These bands, expressed in frequency units, are equally spaced. 



(4) The location and frequency interval of the bands are the same for a 

 given metal and are independent of the solvent. 



(5) The position of the crest of the envelope of the entire group of bands, 

 i. e., the position of maximum brightness in the spectrum, shifts toward the 

 violet with increasing molecular weight of the solvent. 



(6) The frequency interval is definitely and inversely related to the atomic 

 weight of the activating metal, and Mr. Tanaka has identified, by means of 

 this relation, the active element in numerous calcites, fluorites, and other 

 luminescent solids. 



Constant Frequency Intervals in Flame-spectra and in Incandescent Solids. 



The search for constant frequency intervals in flame-spectra (made by Miss 

 M. A. Ewer) has been completed. All the flame-spectra recorded by Eder and 

 Valenta^ have been studied, and it is found that the lines and bands recorded 

 by them as occurring in the spectra of some 23 elements are capable of arrange- 

 ment in groups or series having a constant interval, and that the intervals thus 

 found, as in the case of the kathodo-spectra measured by Tanaka, are defi- 

 nitely related to the atomic number of the element. The curves of these two investi- 

 gators, which are very similar, are not identical in their present form, and their 

 precise relation to each other is being investigated. 



In view of these facts it seemed desirable to study the continuous spectra of 

 flames in which such metals as calcium, magnesium, aluminum, etc., were in 

 process of combustion; also to investigate their oxides and other oxides which 

 show unusual types of incandescence. 



Photographs of these spectra^ were made by Dr. L. J. Boardman at my 

 request and were measured as to density in great detail upon his micro- 

 comparator with photo-electric cell and slit. Explored thus step by step, 

 several hundreds of measurements a small fraction of a millimeter apart 

 being taken, these continuous spectra show a definite structure hitherto un- 

 suspected because invisible. Like the kathodo-spectra of luminescent solids, 

 they are made up of several series of overlapping and submerged components, 

 equidistant as to frequency. The intervals correspond to those found by Miss 

 Ewer for the flame-spectra of the respective metals, and the location of the 

 crests is such as to fill out the missing places between the lines of the flame- 

 spectrum. 



Results obtained with certain photographs of incandescent tungsten (to 

 be mentioned at present as preliminary and with due reservations) indicate a 

 probability that the above structure is common to all incandescent solids. 



Photo-luminescence of Flames. 



The photo-luminescence of flames (see Year Book No. 21, p. 387) has been 

 further investigated during the current year, and it has been found that in 

 addition to the enhancement of the bands previously measured ' certain 



' Eder and Valenta, Atla3 Typischer Spectra (1911). 

 * Boardman, Phys. Rev. (2), vol. xx, p. 552 (1922). 

 » Nichob and Howes. Phys. Rev., xxii, p. 425 (1923). 



