50 CORRELATION OF PHYSICAL FORCES. 



always manifested by an expansion of matter. One class of 

 these exceptions is only apparent: moist clay, animal or 

 vegetable fibre, and other substances of a mixed nature, 

 which contain matter of different characters, some of which is 

 more and some less volatile, i. e. expansible, are contracted 

 on the application of heat ; this arises from the more volatile 

 matter being dissipated in the form of vapour or gas ; and the 

 interstices of the less volatile being thus emptied, the latter 

 contracts by its own cohesive attraction, giving thus a prima 

 facie appearance of contraction by heat. The pyrometer of 

 Wedgwood is explicable on this principle. 



The second class of exceptions, though much more limited 

 in extent, is less easily explained. Water, fused bismuth, 

 and probably some other substances (though the fact as to 

 them is not clearly established), expand as they approach 

 very near to the freezing or solidifying point. The most 

 probable explanation of these exceptions is, that at the point 

 of maximum density the molecules of these bodies assume a 

 polar or crystalline condition ; that by the particles being 

 thus arranged in linear directions like chevaux de frise, 

 interstitial spaces are left, containing matter of less den- 

 sity, so that the specific density of the whole mass is dimin- 

 ished. 



Some recent experiments of Dr. Tyndall on the physical 

 properties of ice seem to favour this view. When a sun- 

 beam, concentrated by a lens, is allowed to fall on a piece 

 of apparently homogeneous ice the path of the rays is in- 

 stantly studded with numerous luminous spots like minute air 

 bubbles, and the planes of freezing are made manifest by 

 these and by small fissures. Stars or flower-like figures of 

 six petals appear parallel to the planes of freezing, and seem- 

 ingly spreading out from a central bubble. These flowers 

 are formed of water. When the ice is melted in warm water 

 no air is given off from the bubbles, so they seem to be va- 

 cuous ; it is, however, possible that extremely minute parti- 



