380 On the CONTRACTION 



late, and continues to dilate, as the temperature falls, till con- 

 gelation actually commences, whether this occurs as foon as the 

 water reaches the 3 2°, or after it has defcended any number of 

 degrees below it. 



Supposing this peculiarity of water to be eftablifhed, it mud 

 appear, indeed, a very odd circumftance, that heat fhould pro- 

 duce contraction in this fluid, while it caufes expanfion in other 

 bodies ; and no lefs ftrange, that within one range of tempera- 

 ture it mould contract, and in another expand, the very fame 

 fubftance. Before a deviation from fo general a law fhould be 

 received as matter of fact:, the proofs of its exiftence ought to be 

 clear and indifputable. 



The experiments hitherto publifhed, from which this Angula- 

 rity has been deduced, have all of them been performed upon 

 water contained in inftruments fhaped like a thermometer glafs, 

 and confuting of a ball with a (lender ftem; and the expanfive 

 or contractive effects of heat and cold have been inferred, from 

 the afcent or defcent of the fluid in the ftem. 



To fuch experiments it has been objected, that the dimenfions 

 and capacity of the inftrument undergo fo much change, from 

 variation of temperature, that it is difficult, if not impoflible, to 

 determine how much of the apparent anomaly ought to be im- 

 puted to fuch changes, and that it is not improbable that the 

 whole of it may be afcribed to them. 



The object: of this paper, which I have now the honour to 

 read to the fociety, is to prove by a fet of experiments, conduct- 

 ed in a manner altogether different, that the common opinion is 

 founded in truth, and that water prefents itfelf as that ftrange 

 and unaccountable anomaly which I have already defcribed. 



It is worth while, before detailing my experiments, to give a 

 fhort account of thofe obfervations which led to the difcovery of 

 the fact, and which in fuccefhon have extended our knowledge 

 of it, as well as of thofe obfervations which have at different 

 periods been offered to difcredit, and to bring it into doubt. 



The 



