298 Messrs. Roberts and Wrightsori on the 



lation, the results are remarkably concordant and their mean 

 agrees very closelv with that obtained by Mallet a method. 

 We venture to think, therefore, that the density of bismuth in 

 the solid and the fluid state may now be considered to be 



definitely settled. . T , 



Fig. I is the oncosimeter diagram of experiment JNo. I (see 

 table), with a calculation of the fluid specific gravity annexed. 



Fig. 1. 



Oncosimeter Diagram of Bismuth Ball, 2-25 inches diameter, immersed in 



Fluid Bismuth (see table, Exp. 2). 



Weight of ball and immersed part of stalk 3 ?i5- OZ ' 



Specific gravity of do. do. »<- 



First floating effect 



34-76x9-755 = 10 . n 



Specific gravity of fluid = 33-46 



When first immersed, the floating effect is 1'3 troy ounce, 

 which (with the weight and density of the ball known) is all 

 we require to determine the fluid-density. Bismuth has a low 

 heat-conducting power ; and therefore the mass of the ball is 

 reduced by surface-melting before much heat can penetrate to 

 the centre. Hence the diagram does not accurately show, 

 as in metals of high conducting-power, the change of volume, 

 the effect being compounded with that produced by loss ol 



mass. . , . , . -, . -, 



In the case of iron, the conducting-power of which is high, 

 diagrams taken with the oncosimeter show correctly the ex- 

 pansion of the ball until it is in a uniformly plastic state. * ig. I 

 is a diagram of a cast-iron ball, 4 inches diameter, immersed 

 in fluid iron of the same quality (No. 4 Cleveland). In this 

 the solid iron is shown to be of less volume than the liquid 

 for about 25 seconds, then to rise gradually in volume until 

 in 4 minutes it becomes plastic (this having been proved by 

 taking balls out in this stage of temperature, when an iron 

 pin could be run through and through the metal as though it 

 were a piece of putty). The slight fall of the line for about 

 2 minutes towards the left of the diagram probably shows a 

 slight loss of mass owing to surface-melting ; aud then the 

 whole ball melts with great rapidity, and as it joins the liquid 

 metal of the bath the line of volume shoots rapidly down to 



