120 Proceedings of the British Association, 



mass would be kept uniform in one direction, that is, horizontally ; 

 while in the vertical direction, it would vary in some degree as the 

 distance from the cooling surface. These conditions being satisfied, 

 we should likewise have a mass which would be free from flaws, and, 

 when cool, would be free from sensible strain; nothing could be 

 easier than to accomplish this, approximately, in practice ; it would 

 be only necessary to make one surface of the mould (the lower one) 

 of iron of a good conducting material while the remainder was of dry 

 sand. On trial, this plan was perfectly successful ; there was, how- 

 ever, a new, though not a very serious defect, which was immediate- 

 ly apparent — the speculum metal was cooled so rapidly that air- 

 bubbles remained entangled between it and the iron surface ; but the 

 remedy immediately suggested itself, by making the iron surface 

 porous, so as to suffer the air to escape ; in fact, by forming it of 

 plates of iron placed vertically side by side, the defect was altogether 

 removed. It only then remained to secure the speculum from cool- 

 ing unequally, and for that purpose it was sufficient to place it in an 

 oven raised to a very low red heat, and there to leave it till cold, from 

 one to three or four weeks, or perhaps longer, according to its size. 



" The alloy which I consider the best, differs but little from that 

 employed by Mr. Edwards : I omit the brass and arsenic, em- 

 ploying merely tin and copper in the atomic proportions, namely, 

 one atom of tin to four atoms of copper, or, by weight, 58.9 

 to 126.4. As it was obviously impossible to cast large specula 

 in earthen crucibles, the reverberatory furnace was tried ; but 

 the tin oxidized so rapidly, that the proportions in the alloy 

 were uncertain ; and after some abortive trials with cast-iron cru- 

 cibles, it was found, that when the crucible is cast with the mouth 

 up it is free from the minute pores through which the speculum 

 metal would otherwise exude ; and therefore such crucibles fully 

 answered the purpose. It was very obvious that the published 

 processes for grinding and polishing specula, being in a great 

 measure dependent on manual dexterity, were uncertain and not 

 well suited to large specula ; accordingly, at an early period of these 

 experiments, in 1827, a machine was contrived for the purpose, 

 which has subsequently been improved, and by means of it a close 

 approximation to the parabolic figure can be obtained with certainty ; 





