248 SCIENCE PROGRESS 



are present, and which was the first to crystallise, while the 

 percentages of the several constituents in the slag can be calcu- 

 lated from its chemical analysis. More than a hundred and fifty 

 slags are studied in this way, and the conclusions to which 

 the investigation leads rest therefore upon a sufficient basis. 

 It is unmistakably established that, for slags composed of the 

 various minerals enumerated, crystallising under atmospheric 

 pressure without fluxes, the mineral which first begins to 

 separate out is that which is in excess as compared with eutectic 

 proportions. For example, the eutectic ratio for augite and 

 olivine is about 6S to 32. From a molten slag consisting 

 essentially of these two minerals in any other proportions than 

 these, augite or olivine crystallises first, according as there is 

 an excess of one or the other constituent. In like manner are 

 determined the eutectic ratios for other pairs of constituents : 

 plagioclase and diopside, 65 to 35 ; plagioclase and olivine, 70 to 

 30; melilite and anorthite, 65 to 35; melilite and olivine, 74 

 to 26 ; with more or less approximation in different cases. 



The second part of the memoir deals with the lowering of 

 freezing-point in mixed slags. It is shown, for example, that, 

 while the melting-point of augite is about 1225 and of olivine 

 about 1400 , and both minerals crystallise freely, a mixture of 

 the two will remain fluid down to much lower temperatures. 

 The lowest freezing-points are for mixtures in the neighbour- 

 hood of eutectic proportions, and the minimum point is about 

 120° below the melting-point of the augite, and 300 below that 

 of the olivine. This lowering of freezing-point is not established 

 by direct experiment, but from data of the total heat of fusion. 

 Akerman determined for a large number of slags the heat 

 required to raise a unit mass from o° to a state of fusion. This 

 consists of three parts, expended (a) in raising the crystalline 

 slag to the melting-point, (b) in melting it, and (c) in producing 

 a sufficient degree of superfusion to make fluidity evident. 

 This last term is relatively small, and may be presumed ap- 

 proximately the same for the different slags of the series. 

 Since augite and olivine are known to differ little as regards 

 specific heat and latent heat of fusion, the term b is approxi- 

 mately constant for a series of augite-olivine-slags, and the 

 term a depends principally upon the temperature of fusion. 

 The measurements of total heat of fusion therefore afford an 

 indirect means of estimating the lowering of freezing-point due 



