OF FLAMES EESULTING FEOM BESSEMEE PEOCESSES. 
491 
Western Railway during the month of January, 1895, from which the details of two 
spectra are taken for comparison of the changes of intensity in the lines of iron, as 
given in the preceding table, show the gradual progress of these changes in a very 
marked degree. The solar spectrum lines are represented by E,owland’s figures for 
the comparative intensities, and lines of the other spectra are compared directly with 
these as photographed by us with the instrument which was used for the Bessemer 
spectra. 
There can be but little doubt, if any, that the arc and solar spectra are more nearly 
alike in the intensity of their lines than any other two ; next to the arc, we have the 
oxyhydrogen flame spectra and Bessemer spectra. The change in intensity is much 
more striking in the spark spectrum of iron. In this the lines of the first group are 
either absent or present only as very weak lines, while those of the second group are 
very strong lines. 
We have clearly identified all the lines under discussion by making more accurate 
observations of the lines in the flame spectrum of iron. These latter measurements 
were made in the Chemical Laboratory of the University of Cambridge with a 
Rowland’s plane grating and spectrometer belonging to Professor Liveing. 
Oxide of iron heated in the carbon monoxide and oxygen flame gives a spectrum 
intermediate between the Bessemer and the oxyliydrogen flame spectra. 
The next line beyond this group of iron lines is one of wave-length 4555’1. It is 
best seen on the third sj)ectrum on the Plate 15 taken at Middlesbrough, but occurs 
also in other spectra. It is seen best in the earlier spectra of a blow, and is invisible 
in those which come later. It is, therefore, caused by a small quantity of a volatile 
element, unless, perhaps, the stronger continuous spectrum of the later periods of a 
blow masks the weaker lines. In this event, the weak lines of iron should also 
disappear, but they do not ; so the first inference is probably correct. The strongest 
caesium line has, according to Kayser and Runge, the wave-length 4555'44 in the 
arc spectrum, and the line in question is doubtless this line, which we know to be by 
far the strongest in the oxyhydrogen flame spectrum of cmsium. Rubidium and 
other alkali metals are present, and the presence of caesium may therefore be 
expected. 
On a New Line in the Si^ectrum of Potassium. 
In many of the spectra a sharp line occurs near wave-length 4642. This line also 
becomes w^eaker, and on some plates disappears, during the later periods of the blow 
It was not until after much careful study that it was finally traced to potassium. It 
is not recorded on any hitherto j'^ublished arc, spark, or flame spectra of this 
element. 
It occurs in the oxyhydrogen spectrum when potassium compounds are used, but it 
does not appear with the same intensity as in the Bessemer flame. Lonqiarative 
3 R 2 
