This paper records experiments undertaken to extend to liquids the accuracy of quantitative analysis recently attained by the spectrography of alloys. How to produce from a solution a spectrum which shall truly represent the solution, and what effect may be expected from the presence of other metals than the one under determination, are among the questions dealt with, for they must come into consideration in applying any method of quantitative spectrography of solutions. The Production of the Spark . ( a ) The Sparking Vessel for Liquids .—At the outset of the investigation it became apparent that the older forms of sparking apparatus, used by Hartley, Pollok and Leonard, and others, were unsuitable for quantitative work, for it was found that, owing to incrustation of the electrodes and to decomposition of the solution around them, the spark soon became unrepresentative of the bulk of the solution.