1)0 MESSRS. A. SMITHKI.l.s. II. M. DAWSON, ANM) II. A. \V1 I.SOX: ELECTRICAL 



suspended by stretched platinum wires, according to the plan of ARRHENIUS. We 

 found them unsatisfactory owing to the slackening of the wires after heating. We 

 also tried thicker platinum plates attached to thin platinum rods and supported by 

 porcelain tubes. We l)elieve, ho\\cver, that our final arrangement was preferable in 

 point of rigidity and in including a symmetrical zone of gases for measurement. 



A vertical millimetre scale, H (fig. 1), fixed behind the glass cylinder, enabled the 

 height of the electrodes and of the inner cone of flame, K, to be read by means of a 

 cathetometer telescope. 



The source of electricity used by us consisted usually of three accumulators. A 

 German silver wire, 1'5 millims. in diameter and 20 metres long, carefully calibrated, 

 was stretched four times along a bench over a millimetre scale, E. By making 

 contact with two heavy three-legged contact-pieces at two points, any E.M.F. up to 

 57 volts could be taken off this wire. 



The current in the circuit through the flame was measured by a Kelvin astatic 

 reflecting galvanometer, K, of 5600 ohms resistance, provided with a shunt box, X. 

 In every case in taking a reading the current through the galvanometer was reversed 

 and the mean deflection taken. The sensibility of the galvanometer, as we used it, 

 was about 2*85 X 10~ 9 ampere for one scale division. All connecting wires were 

 supported on glass rods fixed in paraffin blocks. 



The E.M.F. given by the stretched wire was compared on all occasions with that 

 given by a standard Clark element, and the sensibility of the galvanometer was 

 repeatedly determined. 



In some earlier experiments higher E.M.F.s were obtained by means of Leclanche 

 cells up to thirty in number. 



In the following pages the measurement of current strength is always given, except 

 where otherwise specified, in terms of 10~ 7 ampere unit. 



Method of making an Experiment. 



In beginning an experiment the tubulated cylinder, T (fig. 1), was removed, and, 

 together with the sprayer, washed well with distilled water and dried by a current 

 of air. 



The cylinder was then replaced and the reservoir, E, filled with the solution to be 

 investigated. 



Any salt which in a previous experiment had deposited on the flame tube was 

 removed, and the cylinder, G (fig. 2), and mica plate washed and dried. The gas was 

 then turned on and, after an interval, lighted above the mica plate. Air was next 

 supplied until the sprayer came into action, when the level in the reservoir, R, was 

 adjusted. When the flame had become non-luminous the mica plate was removed, so 

 that the whole flame descended and burnt at the mouth of the flame tube. As soon 

 as' the air had reached the right pressure the mica plate was replaced, whereupon the 

 outer cone, U (fig. 3), of the Bunsen flame rose and burned above the mica. 



