IRON CYLINDER BURST BY ELECTRICITY. £8£ 



tliat they contain no sulphate of soda*. For example, to 

 decide whether the Scotch salt contains an alkaline sul- 

 phate, or not, I dissolved 1500 grains in a pint of boiling 

 water, and evaporated till fourteen drachm measures only 

 remained, the common salt being removed as soon as it was 

 formed. The residuary liquid was divided into two equal 

 portions, one of which gave 18£ grains of sulphate of barytes, 

 and the other, 14 grains of ammoniaco-magnesian phos- 

 phate. The proportion between these numbers is so nearly 

 that which has been already assigned, (viz. 112 to 90,) that 

 we may safely infer the total absence of sulphate of soda- 

 This salt, indeed, is considered as incompatible with muriate 

 of magnesia; but after digesting, for two or three days, 100 

 grains of the former, with 20 of the latter, evaporating to 

 dryness, and washing the residuum with repeated effusion* 

 of alcohol, I found, that two grains of the muriate of mag- 

 nesia had escaped decomposition. 



IV. 



On the Action of the Electric Fluid, by which an Iron Cy- 

 linder an Inch and half thick was torn asunder: in a Lei" 

 ' ter from Mr.*** to J. C. DELAMETHERiEf. 



SIR, 



JL Have already communicated to yon, in different letters, Metallic ey- 

 my experiments on various metallic cylinders, that were / n Jj 1 * J? 

 torn by electric discharges. I half tilled with water the cy- shock*. 

 linder, Plate VTI, fig. 1, which was open only at one end. 

 In this water I immersed a small leaden rod, which I sur- 

 rounded with wax at the place where it entered the cylinder, 

 in order to insulate it, as at fig. 3. I then formed a com- 

 munication between one of the surfaces of a powerful elec- 



* I employed move attention in investigating the presence of sulphate 

 of soda in the products of sea-water; because this sali: is stated to be 

 one of its ingredients by the Bishop of Llandaff, (Chemistry vol. 

 ii. p. 62,) and by other chemical writers. 



t Journ. de Physique, vol. LXVI1I, p. 411. 



trie 



