When fwimming in ft h A S N of OIL, 8cc. iG'^ 



prelTed upwards by a force fufficient to raife part of it above the 

 general level. But tliis heated portion of oil, in its endeavour 

 thus to rife up, will meet with a refiflance equal to the weight 

 of the incumbent lamp, which will determine it, in feeking a 

 vent, to Aide out from tmder the bafe in a thin fuperficialjlreain ; 

 and it feetns to follow^ with equal certainty, that this conftant 

 ftream will flow moft readily and moll cbpioufly towards that 

 lide of the bafe of the lamp where the refiflance is leaft, or where 

 it has the ftiortefl way to prefs forward ; that is, from xinder 

 the wick or flame, to the edge of the bafe which is the neareft, 

 according to what we have feen to be agreeable to the pheno- 

 mena. But, from the laws of motion, it is certain, that the re- 

 action of this ftream of rarified oil, thus iffuing moft rapidly 

 and moft copioufly from a particular fide of the bafe, muft im- 

 pel the lamp in the contrary diredlion, and make it fail in the 

 manner we have feen. It may further be remarked, that the 

 heated oil, fo retreating from the flame, and endeavouring to rife 

 fomewhat above the general level, in confeqvience of its dimii- 

 nilhed fpecific gravity, may more or lefs lift up that fide of the 

 bafe neareft the wick, and aid the readlion of the recoiling 

 ftream, by making the lamp fail in the oppofite diredlion, as it 

 were down-bill. 



That the rarified oil under the bafe has really a conftant 

 tendency to rife above the general level, feems undeniable, from 

 the following fa<5ls, namely, that after any of the lamps has 

 burned a little while, and has got its bafe foaked with the oil, 

 as foon as the flame is blown out the lamp finks to the bottom } 

 and even a lamp, with its bafe made of a thin lamina of talck, 

 fails very well till the flame is extinguiflied, and then it imme- 

 diately finks. 



Agreeable to the explanation which has now been attempt- 

 ed, I fovmd, that when a topical heat was applied to the furface 

 of the oil, by bringing the point of a poker, dully red hot, near- 



