Transactions of section a. 629 



necting lines are then drawn tlirougli homologous spectral lines. The spectra 

 studied by the author in this way are chiefly those emitted by the metals in the 

 oxyhydrogen and oxycoal gas flames. These spectra are much simpler than those 

 of the same metals in the electric arc or spark, and may be regarded as the funda- 

 mental spectra of the metals. They are therefore the most suitable spectra for 

 comparison. 



As the flame spectra of the metals have not been fully investigated some linea 

 have been selected, to make the diagrams more complete, from arc and spark 

 spectra. In these cases the selection was made after a study of the character ot 

 the lines in these spectra. Later experimental work on flame spectra has con- 

 firmed the selection of some of these lines, and the work on the Zeeman efl'ect, of 

 Preston on magnesium, zinc, and cadmium, and of Lord Blythswood and Dr. 

 Marchant on mercury, confirms it in the spectrum of the latter metal. The 

 formulae and work of Eydberg and of Kayser and Kunge lead to the selection of 

 the same lines in all cases, and with these formulte as guides it is possible to 

 e.'stend the work to other lines and spectra. This has been done, but only to a 

 limited extent at present. The diagrams exhibited were drawn — (1) from atomic 

 weights and oscillation frequencies, and (2) from the squares of the atomic weights 

 and oscillation frequencies. 



The diagrams show very clearly that the spectra of similar elements are very 

 closely related to one another. That the spectra of potassium, rubidium, and 

 Cffisium are more closely related to one another than to those of lithium and sodium, 

 and that there is also a break between the spectrum of magnesium and those of 

 zinc, cadmium, and mercury, and between that of aluminium and those of gallium, 

 indium, and thallium. 



The connecting lines of the difl'use subordinate series of potassium, rubidium, 

 and caesium approach in the more refrangible lines measured to straight lines, 

 while those of the principal series are nearly straight lines in the second diagram. 



The lines joining the homologous lines of doublets and triplets approach one 

 another as the atomic weight decreases, and, in the second diagram, intersect in 

 points near the line of zero atomic weight. These curves give exact information 

 regarding the function of the atomic weight which determines the differences, in 

 oscillation i'requencies, between the lines in doublets and triplets. 



Equations are given, after the form of Rydberg's, for the principal series of 

 lithium and sodium and of potassium, rubidium, and caesium, and the calculated 

 numbers are in close agreement with the observed numbers. 



4. lieport on Radiation in a Magnetic Field. — See Reports, p. 52. 



5. An Experiment on Siniultaneoits Contrast, 

 By George J. Burch, M.A.^ F.R.S. 



It is well known that white objects seen against a red background look greenish- 

 blue, and orange against a blue background. 



This phenomenon is shown in a striking manner in the following experiment 

 due to Hering : — A small white diss is viewed with the left eye against a red 

 background, and another similar disc is viewed against a blue background with the 

 right eye. The discs are so placed as to occupy different positions in the field or 

 view. The result, when the light has been properly adjusted, is that the observer 

 sees an amethyst-blue disc and a topaz-yellow disc against a pale purple ground. 



The reason of this is demonstrated by the author in the following experiment : — 

 Two pieces of glass, one red and the other blue, are inserted in a stereoscope in 

 place of the usual slide, each glass having two small squares of black paper on it. 

 Viewed binocularly the four squares appear as two. In front of the instrument, 

 but out of the direct line of sight, are two adjustable slits, and over the eye-lenses 

 of the stereoscope are two diffraction gratings. The position of the slits is so 

 arranged that the spectrum of the first order of the left-hand grating falls on the 



