ON THE REFRACTIVE INDICES OF GASEOUS POTASSIUM, ETC. 137 



the length of the tubes, or of the thickness of the ends, or to unequal heating of the 

 air, which had given trouble previously, were avoided. 



In the case of zinc, the temperature necessary was so high that the glass 

 diaphragms used to close the outer ends of the furnace began to soften and buckle. 

 This difficulty was overcome by substituting worked plates of fused silica. 



In the course of the work it was found that the dispersion would be considerable. 

 Arrangements were therefore made for determining the index for more than one wave- 

 length. A mercury vapour lamp of BASTIAN'S pattern was found to give an excellent 

 green (X = 4460), and red light (X = 6562) was obtained from a hydrogen vacuum 

 tube. But these methods proved unsatisfactory, and eventually it was found possible 

 to work with approximately monochromatic light sifted out with a slit from the 

 spectrum of the white light of a Nernst filament dispersed through four glass prisms. 

 The light thus obtained was sufficiently monochromatic to give twenty or thirty sharp 

 black interference Iwinds, and it was used in all the later experiments. 



Potassium. 



Of the elements whose indices still remain to be measured in the gaseous state, the 

 most important group is that of the alkalis. It was, therefore, decided to begin with 

 potassium, which seemed likely to prove the easiest. 



So far as we are aware, only one attempt has hitherto been made to measure the 

 index of a member of this group. In the course of his brilliant research on the 

 optical properties of sodium vapour, Professor R. W. WOOD* measured the retardation 

 of light in passing through a column of dense vapour, and compared its value at 

 different points of the spectrum with that produced at the wave-length of the yellow 

 helium line. By the application of the Sellmeier dispersion formula to his results he 

 deduces an index of T0000275 for infinitely long waves for sodium vapour saturated 

 at 644 C. Unfortunately the density of the vapour corresponding to this tempera- 

 ture is yet unknown, so that it is impossible from the data available to calculate the 

 al)solute atomic refractivity of sodium. 



The relative values for different wave-lengths are, of course, independent of the 

 density, and in themselves constitute a most interesting exemplification of the 

 variation of the index in the neighbourhood of an absorption band. Thus, the index 

 increases from the infra-red to the line D,, where it Incomes very large. On the blue 

 side of D 2 the index increases rapidly from very small values, but remains less than 

 unity even as far as X = 2260. 



In view of these facts the investigation of the index of potassium seemed likely to 

 yield results of similar interest. But, unfortunately, the chemical difficulties proved 

 so great that, in spite of numerous attempts, it was found impossible to obtain 

 absolute values with the apparatus employed. At the temperature at which 



* ' PhiL Mag.' September, 1904. 

 VOL. OCVII. A. T 



