18 Temperature of the Terrestrial Globe. 
theoretischen Physik, printed in 1811, properly the true point of con- 
gelation, that at which the water begins to congeal in infinitely deli- 
cate and invisible crystals and the augmentation of these crystals produ- 
ces the successive augmentation of volume which attains its maxi- 
mum at the precise congelation which furnishes the zero of our ther- 
mometers. But we have seen above, that congelation precipitates 
the greater part of the salt contained in salt water, by which its spe- 
cific gravity is diminished. An approximative calculation easily 
proves that these two opposite effects cannot compensate each other, 
and still less produce an augmentation of volume on the approach of 
perfect congelation, the augmentation of density by commencing con- 
gelation being inferior to the diminution by the removal of the salt. 
This theoretic proof has been confirmed by the very exact experi- 
ments of M. Erman the younger, which have proved that salt wa- 
ter increases in density up to the moment of complete congelation. 
Other partisans of central fire, again have recourse to the waters 
which flow from the polar regions by the melting of the ice, and which 
being heavier than the warmer water, descend towards the bottom, 
and passes to the tropical regions. After what has been said of the 
water of the polar seas, we shall not be disposed to allow much stress 
to this argument. We proceed to demonstrate that it is of no val- 
ue whatever, however specious and palpable it may appear. First, 
we only object that to think of cooling the whole ocean in this man- 
ner, would be like attempting to cool the Lake of Geneva with a cu- 
bic fathom of water of the temperature of melting ice.* Let us ex- 
amine the matter a little closer. 
Let us take for a basis the temperature of 30° C., for that of fics 
water of the sea under the equator and at the ewatnee The differ- 
ence of specific gravity between this water and the same water there 
where the temperature is =0, will be ;1,, if we admit according to 
MM. Dulong and Petit, a variation of density equal to ;3;; for 1° 
C. abstraction being made for the dilatation of the glass. On the con- 
trary the superior degree of saltness of the tropical water produces, 
according to M. Horner, a difference of ,1, in favor of this water 
over the water at the latitude of 60°; and we can without risking any 
sensible error, shew this difference to be ,1, between the tropical 
* An attempt was thus made some years ago, fo explain the general fall of the 
mean temperature of our whole atmosphere, by the melting of ice detached from the 
shores of Iceland. 
