404 Mr. Cricltton on [June, 



nor more, but at present, I am rather inclined to place it a very 

 little above 42 ; a trial I made in very favourable circumstances 

 a few days ago, gave for the first appearance of the rising of a 

 ball 37*5, and for that of its sinking 46'3, these make the point 

 in question 41*9; the local temperature was 46*8, but the baro- 

 meter having been at only 29-4, the above 41*9 may be held 

 perhaps too low. These experiments were made with balls 

 adapted to all the intervals from 33 — 51 to 39 — 45, yielding 

 however great uniformity of results. 



As I cannot anticipate what objections, or if any, can be made 

 to this method of ascertaining a curious and not unimportant 

 point, I shall allude to one only ; that is error from expansion of 

 the ball, and consequent increase of its volume ; but as the 

 whole range required does not exceed 4° or 5°, on either side of 

 a starting point, and though it were granted that the expansion 

 of glass is the same for 4° about temperature 42, as it is for 180°, 

 that is from freezing to boiling of water, as determined by M. de 

 Luc and others, the expansion for these 4° must be so extremely 

 little, as not by any means to affect the decision in any consi- 

 derable part of a degree. 



But were the expansion of glass in the above range even ten 

 times what it is, still it must in effect be cancelled, for taking 42 

 as the point where this expansion in the present case must be 

 assumed as incipient, and granting that at say 33, a ball just held 

 in poise has become less, that is heavier specifically, some degree 

 above 33, for example 34, where water is denser, must really be 

 what the ball virtually indicates; again, if at 51 the same ball 

 poises, then by a parity of reasoning the ball is now said to be 

 increased in volume beyond what it was at 42, or it is too light , 

 therefore it must indicate too high a degree, or it really shows 

 that the ball, supposing it inexpansible, would have stood at a 

 lower degree or denser medium, which call for instance 50 ; so 

 that by the one extreme thus correcting the other, the conclusion 

 to be drawn is the same as in the case of altogether neglecting 

 the expansion. 



The low temperature of the atmosphere when these experi- 

 ments were made, gave confidence that no current upward or 

 downward moved the water ; besides, a few very minute parti- 

 cles of dust, just visible in different parts of it, remained entirely 

 motionless during the whole operation. 



My first trials on this subject were made with spherical balls, 

 half an inch in diameter, having a depending stalk of about an 

 equal length ; but to obviate the possibility of error from dissi- 

 milarity of the extremities, I latterly used solids resembling in 

 shape a buoy or parabolic spindle, sharp at the ends, of about an 

 inch in length and 4-10ths in diameter. This shape gave another 

 apparent advantage, that is of meeting less resistance than a 

 sphere when moving in a fluid, and in order to ensure perpendi- 

 cularity of the axis, before such a ball was hermetically sealed, a 



