VOL. LXXIII.] PHILOSOPHICAL TRANSACTIONS. 451 



the afternoon it was observed at — 1 ]6°, and by 8 the same evening it had risen 

 to— 31°. Though the quicksilver," continues M. Tornsten, "thus fell to 

 — 1 16° on the first of January in the afternoon, I am of opinion that its des- 

 cent ought not to be ascribed to a proportionable increase of cold, but on the 

 contrary proceeded from the sudden change to milder weather, which came on 

 that afternoon. For the preceding evening, when the thermometer was stand- 

 ing at — 54°, I remarked, that on bringing it into a warm room, the quick- 

 silver fell on a sudden entirely into the ball, which was about 130 degrees below 

 O. This experiment I repeated several times with success, but observed the fol- 

 lowing difference, that if I had not kept the thermometer in the heat long 

 enough for the quicksilver to begin to rise again after it had sunk into the ball, 

 it never ascended above the 130th degree by continuing in the cold, but on being 

 carried back into the warm room it contracted still more in the ball by a quantity 

 which, however visible, could not be measured. On the other hand, if the in- 

 strument had been kept in the room till the mercury had risen above — 54°, it 

 became stationary at that degree in the open air. Now, though I did not, on 

 the 1st of January, bring the thermometer within doors before it had sunk of 

 itself to — 11 6°, yet this fait likewise seems to have been occasioned by the 

 change to milder weather which was then taking place. For at 8 in the evening, 

 when the external cold was at — 31°, I found that hoar-frost formed on the ball 

 and stem of the thermometer as before, on its being brought into a warm room ; 

 but the mercury did not sink, on the contrary it began immediately to rise. 

 Some quadrupeds perished by the intense cold,, and a great number of small birds 

 were found dead." 



M. Tornsten certainly judged right when he concluded, that the fall of the 

 thermometer to — 11 6° rather indicated a diminution than an increase of the 

 cold. Though he knew nothing of the cause, yet his observation led him to a 

 just inference, in which he displayed more sagacity than M. Hellant on a similar 

 occasion. All the phenomena which so much perplexed these gentlemen are ex- 

 plicable in the following manner. When the air becomes sufficiently cold to 

 freeze quicksilver, that metal must be standing about — 3Q°, or in the common 

 way of marking the boiling point, somewhere between — 40° and — 50°, in the 

 tube of a thermometer exposed to it. As the small thread of mercury in the 

 tube must be more easily affected by the cold, it will probably congeal before any 

 other part, and stick fast about the above-mentioned degrees. The remainder of 

 the mercury will then go on to freeze, and as it suffers such a great contraction 

 in becoming solid, must leave a considerable vacuity in the bulb of any common 

 thermometer. Consequently, when the cold, from whatever cause, comes to 

 be less than is required for keeping the metal in a solid state, the small thread 

 that was frozen in the tube immediately melts, and sinks down into the vacuity 



3 M 2 



