in Hot-blast Furnaces for making Cast-iron, 33 1 



of ammonia, or by carbonate of potash — showing the absence 

 of all metallic bases, except such as are alkaline. 



Potassium was proved to be present, and sodium too, by 

 crystallizing a portion of the same solution, examining the 

 successive crops of crystals that were formed, and persevering 

 in the crystallization until there did not remain a drop of the so- 

 lution. At first the well-known crystals of nitre were obtained. 

 Towards the end, however, when the solution had become small 

 enough to be transferred to a watch-glass, little crystals of ni- 

 trate of soda, in their well-known rhomboidal form, appeared. 

 This test for the presence of sodium is much more delicate 

 than might be imagined. But another and a readier test 

 showed the presence of sodium as decisively. A platinum wire, 

 scrupulously clean, is, by way of precaution, placed either 

 before the tip of the inner blue flame of the blowpipe, or so 

 as to touch the circumference of the blue flame of alcohol. In 

 either case, the colour of the flame is unaffected by the wire, 

 if sufficiently clean. Dipped, however, into a strong solution 

 of any salt containing potassium, and dried above a flame, the 

 wire will, before the blowpipe, show a violet flame, beyond 

 the wire, in continuation of the blue one, but short and spread ; 

 and, in the flame of alcohol, it will tint with a like violet co- 

 lour all above the wire. But a similar wire, dipped into a 

 rather strong solution of a salt containing sodium, and treated 

 in like manner, will give, before the blowpipe, an intense 

 greenish-yellow light, shaped so as to seem a prolongation of 

 the blowpipe's blue flame ; and, when placed in the flame of 

 alcohol, will imbue so much of the flame as is above the wire 

 with a similar colour. A sodium salt, although intermixed 

 with one of potassium in a smaller proportion than 1 to 100, 

 will give, with sufficient distinctness, a like indication of its pre- 

 sence. Any common gas-light, lowered till it burns blue, will 

 answer for the detection of sodium in this manner ; but, to ex- 

 hibit this test on the lecture-table, the flame of alcohol answers 

 best. By this test, applied in all these modes, the salt exuded 

 from the blast-furnaces gave distinct indications of the pre- 

 sence of sodium. 



The proportion of nitrate of soda obtained by crystallizing 

 the mixed nitrates of the two alkalis was manifestly small ; but, 

 as the two were evidently in a state of mixture — not of com- 

 bination — it did not seem worth while to ascertain the propor- 

 tions of each. Nevertheless, I made an approximative expe- 

 riment, in order to form some idea of the relative proportion 

 of the intermixed sodium salt. Equal weights of pure chlo- 

 ride of potassium and of mixed chlorides, formed by treating 

 the salt under examination bv pure muriatic acid, were se^m- 



2 U2 



