371 



Wedgewood had stated in his memoir that the supply of the 

 clay was inexhaustible ; but when the stock first made was 

 disposed of, he was unable to find the identical spot where 

 it had been obtained, and the contraction of the new speci- 

 men was different. Had he used it as he did the other, and 

 merely directed the employment of a number by which to 

 multiply the indications of the scale, no inconvenience would 

 have resulted ; but he thought he might bring it to an iden- 

 tity by adding ' earth of alum,' obtained by precipitating 

 alum by carbonate of potassa; a product which Apjohn or 

 Kane will tell you is very far indeed from being pure alumina. 

 This unhappily made the contraction irregular, and the clay 

 pieces much less capable of resisting a high heat. Its indi- 

 cations were found to differ from those of the first set, and 

 it fell into disuse, especially for two reasons. The first that 

 Wedgewood had assigned to his degrees, a value enormously 

 too large, so that he supposed the extreme heat of a furnace 

 to be about 30000 of Fahrenheit, when it is only 4000. 

 Many years since, in our Transactions, I had pointed out 

 this error, and corrected it, with tolerable success, as was 

 long afterwards confirmed by Daniell and Prinsep. The se- 

 cond, that a long exposure to a low heat produces the same 

 contraction as a short exposure to a high. This is said by 

 Sir James Hall to have been established by Dr. Kennedy, 

 whose experiments, however, are no where published ; and 

 I confess that I doubt the fact. Guyton De Morveau has 

 even made an observation which may account for the mistake. 

 He found that similar pieces exposed for half an hour in a 

 powerful furnace, one surrounded by siliceous sand, the 

 other by powdered charcoal, marked 90 and 60, in conse- 

 quence of the different conducting powers of these media. 

 Now it is possible that the Scotch philosopher may have 

 overlooked this influence, and not allowed time enough for 

 the higher temperature to be fully transmitted. The py. 

 rometers of Daniell and others which have since been con- 



VOL. II. I 



