MR. CHARLES DE WATTEVILLE ON FLAME SPECTRA. 161 



and which is surrounded itself by the external portion of the flame. Here it may 

 undergo a breaking-up, or in any case a modification of its physical or chemical 

 constitution. In support of this view, one might mention the great changes produced 

 in all the propertied of iron, even in the solid state, by variations of temperature of 

 much smaller range than those which occur in the flame. The metallic vapour, thus 

 modified, passes in succession through zones of the flame of gradually decreasing 

 temperature, and in which it exists in other intermediate states which might l>e called 

 " states of less advanced dissociation." 



The following observation will serve to give a clearer idea of this method ol 

 considering the phenomena. If by means of a short-focus lens a small image of the 

 flame is thrown on to the slit of the spectroscope, so that the whole length of the 

 image is less than that of the slit, then it will be found, as for example in the case of 

 potassium, that the spectrum is divided longitudinally into three well-defined parallel 

 strips. In the lower strip, which corresponds to the blue cone of the flame, all the 

 lines will l)e found together with the band spectrum of carbon. At the upper edge 

 of the central band the five groups of four extremely close lines, which belong to the 

 t\\<> secondary series of potassium, will be seen to terminate; also the continuous 

 spectrum which accompanies the line sj>ectrum ends at this edge. Now it is precisely 

 the point in the flame corresponding to the upper edge of this strip which marks the 

 termination of what is considered to be the hottest zone of the flume, namely, the 

 part surrounding the blue cone and whicli appears to the eye to l>e more luminous 

 than the other portions. The third horizontal strip, corresponding to the highest 

 portions of the flame, contains only the strongest lines of the principal series of the 

 metal standing out from a completely dark taickground. 



In the case of other metals whose more complex spectra have not yet been divided 

 into series, considerable differences have l>eeu found, as the tables show, in the 

 intensities of one and the same line depending upon whether this line has IHJCII 

 produced in the neighbourhood of the base or in the upper regions of the flame. 

 C'/ertain of their lines even exist only in the neighbourhood of the blue cone. It is 

 very probable that an examination of the relative lengths of the lines and of their 

 appearance would result in the determination of new series. In fact, in the spectrum 

 of the alkaline metals and of those metals whose flame spectra do contain known 

 series, a complete analogy is seen in the appearances of those lines whose wave-lengths 

 are connected with one another by some numerical relation. I have elsewhere (p. 160) 

 pointed out in the flame spectrum of chromium the presence of two triplets which 

 are probably members of the same series. 



If we now compare the flame spectra thus produced with those of the arc and the 

 spark, it will be noticed that as a rule the lines which are found in the flame spectrum 

 are those which are the strongest lines in the arc spectrum. In certain cases, some of 

 the more intense arc lines are absent, whereas less intense arc lines are to be found in 

 the flame spectrum. On the other hand, none of the characteristic lines of the spark 



VOL. CC1V. A. Y 



