472 Mr. W Brennand. [Jan. 26, 



resulting in the production of several very interesting and very 

 beautiful metallurgical specimens. 



The great advantage of this process is, like that of the foregoing 

 separation from arsenic, its extreme simplicity, the low temperature 

 which renders it possible to work upon very large quantities at one 

 time, and the very small amount of time necessary for this separa- 

 tion in comparison with the process hitherto adopted, and the 

 absence of loss in the bismuth operated upon by volatilisation. It is 

 obvious that where metals can be so easily treated in large quantities, 

 the labour and skill hitherto necessary is very considerably reduced, 

 and there is the additional advantage that the loss attending large 

 operations is minimised. 



In this and in my previous papers upon this beautiful metal 

 bismuth, I have been able to point to simple dry processes for its 

 separation from gold, lead, copper, arsenic, and antimony, and all 

 these processes are available for treating with care large quantities at 

 one time. When it is remembered what is involved in having to 

 dissolve any quantity of bismuth in acids, and its subsequent pre- 

 cipitation from solution, it surely will be admitted that much of the 

 difficulty in purifying crude bismuth has been effectively removed, 

 as the methods given have been found possible in practice, and ad- 

 vantageous. 



I have introduced upon the diagram the points at which arsenic is 

 volatilised, and also the point at which antimony separates from 

 bismuth under the conditions described in this paper. 



III. < ; On the Three-Bar Motion of Watt." By WILLIAM BREN- 

 NAND. Communicated by C. B. CLARKE, F.R.S. Received 

 January 2, 1893. 



(Abstract.) 



The figure represents a simple form of " Watt's Parallel Motion." 

 OA = O'B = r are the arms that can turn freely about O,(y, fixed 

 centres, in one plane. The link AB = 2l is pivoted at A and B. 

 As the arms move, P the middle of the link, traces out a portion of 

 the curve, viz., from Q to Q', backwards and forwards, nearly in a 

 right line. 



00 = O'C = d. Of the three parameters d, r, Z, any one can be 

 taken as unit (in this paper I is taken 10 units) ; then d and r are 

 independent parameters. The problem Watt had to solve was to 

 discover numerical values of d and r that should give the tracin- 

 point I the smallest deviation from a right line. 



Watt gave a series of values for d and r which are employed by 

 gineers, with small thumb-rule ameliorations, to this day. They are 



