[ 2 57 I

visions on the mercurial one were all equal, those of theother widened upwards ; as if heated spirit either ex-panded more, or heated mercury less, by a certain addi-tional heat, than the fame fluids do by an equal addi-tion of heat made to them in a colder state. Reaumurfays, that water from freezing to temperate expands onlyone tenth part as much as spirit does, but that fromfreezing to boiling it expands half as much as spirit inthe same interval. Though the difference in the propor-tion at different periods of the heat is doubtless veryconsiderable, I apprehend it does not amount to quiteso much as this ; and that the mistake arose from suppo-sing the full heat of boiling water to have been commu-nicated to the spirit thermometer immersed in it for alittle time; whereas spirit cannot bear so great a heat asthat in which water boils, and consequently, in this partof the experiment, the spirit was less heated than thewater it was compared with. These variations in theproportional expansions of different fluids seem to havebeen little considered by those who have given compari-sons of different thermometers, by reducing the divisionsof one to those of the other from only two correspondingpoints on each.

A more remarkable exception from the general law ofexpansion is observed in the freezing of water. Thoughwater shrinks more and more. its warmth diminishes,down to the pciiod of ics congelation ; yet, at the instantof its becoming ice, it expands into a larger volume, soas to burst the strongest vessels that have been employedfor confining it. The floating of ice in water is a neces-sary consequence, and a convincing proof, of ice beingless dense, or more expanded, than water in its fluid slate.M. de Mairan, in a dissertation on ice, attributes thisincrease of the bulk of the water chiefly to a differentarrangement of its parts ; the icy skin on water being com-fy 1 posed