( 136 ) 



Fig. 1- 



Fig. 1 represents a section 

 of tlie burner, yl is a cop- 

 per trough, 80 cM. long, 

 8 cM. wide and 5 cM. deep, 

 tliickly coated with varnish 

 and having a broad flange. 

 The planed brass plate B 

 is firmly screwed upon 

 the flange and a leather 

 packing makes the joint 

 air-tight. On this cover, 

 which has a rectangular opening 75 c.M. long and 2 cM. wide, are 

 fixed two brass rulers, C and C", 75 cM. long. Thev are so adjusted 

 that at they form a slit, having an exactly uniform width of about 

 0,1 cM. over the whole length. The prismatic s})ace between C and 

 (y is closed at each end by a small triangular brass plate. The 

 trough is filled to a certain height with a saturated solution of soda, 

 and into the remaining space a mixture of illuminating gas and air 

 is conveyed by means of tubes, entering at both ends. These tubes 

 are fed fVom a mixing bottle in which the gas and the air are being 

 driven through two separate regulating taps. 



If now the flame were left to burn without any further precautions, 

 the slit would soon be closed in consequence of the onesided 

 heating of the rulers. It was therefore found necessary to place the 

 trough in a vessel with running water, reaching up to the burner. 

 In this way a uniform and steady flame was obtained. 



A few millimeters below the level of the salt solution a platinum 

 wire P is stretched over the whole length of the lamp. Its ends are 

 soldered to insulated copper wires, which pass through the walls of 

 the trough, and are connected to the negative pole of a storage 

 battery of 20 volts. From the positive pole two insulated wires 

 lead to the ends of a long strip of platinum P' , which rests on a 

 glass plate at the bottom of the trough. As soon as the circuit is 

 closed, innumerable minute particles of the fluid rise into the space 



1) The abnormal solar spectrum of Hale; the peculiar distribution of light in 

 several of the Fraunhofer lines, even in normal conditions ; tbe variations in the 

 average appearance of the spot spectrum accompanying the eleven year period, 

 all these phenomena have been easily explained from the considerations here 

 alluded to (See W. H. Julius, Proc. Roy. Acad. Amst. IV, p. 5SÜ— 602; 66iJ— 666; 

 V, p. 270-302). 



The present investigation is a continuation of the experiments with the long 

 sodium flame, a short account of which has already been given on those former 

 occasions in support of our theory. 



