jot MKSSRS. A. SCHUSTER AND G. HH MSAI.KCH ON 



edge of the most refrangible component, and there would be a tendency to s. i tin^ the 

 microscope on the strongest part of the line rather than on the edge of it. It is 

 easily seen that an error would l>e liable to arise, giving too great a displacement or 

 too small .1 vrloritv. We are not at all certain that if this source of error were 

 eliminated the double line would not show higher rather than lower values for the 

 velocity, and hope to decide this question by using greater dispersion. 



Comparing different capacities with each other, we find that for the spark gap of 

 5 millims. the velocities are greater for small than for large capacities; we offer :it 

 present no explanation of this unexpected result, which requires confirmation and 

 further investigation. When the sparking distance is increased to 1'5 centim., the 

 course of the sparks becomes so erratic that not much importance can be attached to 

 the figures. A query has been attached to those that are specially doubtful. On the 

 whole, there seems a tendency towards greater velocities in this case. 



Our normal spark gap of 1 centim. does not show any decided difference due to 

 capacity, and with our normal capacity of six jars the spark gap does not seem to 

 affect the result. Hence we are justified in thinking that under the conditions 

 named a reliable comparison may be made between the velocities obtained for 

 different metals. 



We have collected in Table III. the velocities V and V as measured on our photo- 

 graphs. The results are nearly always the mean of two sets of measurements from 

 different photographs. In some cases, such as magnesium, iron, manganese and silver, 

 no satisfactory measurements could be made. The spectrum, as it appears on our 

 films, does not always give the same distribution of intensity among the lines as is 

 shown by the spectra of sparks taken with an induction coil and jars of small capacity, 

 and it differs also, of course, from the arc spectrum. A discussion of the peculiarities 

 of our spectra lies outside the range of this paper, and must be reserved for another 

 occasion, but in the following account some of the chief features of our spectra are 

 shortly pointed out. The metals are arranged in order of atomic weights. 



