-1859] WRITINGS OF JOSEPH HENRY. 197 



solid state, as in the case of the shooting of spicules across 

 the surface of water in a metallic basin exposed to the cold. 

 It will be found on inspection that the filaments of ice ar- 

 range themselves at definite angles of either 60° or 120°, and 

 that the triangular openings are bounded by sides making 

 the same angles with each other. In reference to crystalli- 

 zation, there is an important law to be borne in mind, 

 namely, that the axis of the crystal always tends to be at 

 right angles to the surface of the cooling mass. For exam- 

 ple, if a quantity of melted zinc be poured into a cylindrical 

 hole in cold sand, and the bar thus formed be broken across, 

 the crystals will be found to be arranged in the form of radii, 

 with their bases in the circumference; and in some cases 

 there will be found a cylindrical hole along the axis, from 

 which the metal has been drawn away by the shrinking at 

 the time of cooling and crystallization. A precisely analo- 

 gous arrangement takes place in the freezing of water, which 

 may be observed by placing a quantity of this liquid in a 

 globular glass vessel, and submitting it to a temperature of 

 some 10° below freezing. We shall find then that the crys- 

 tallization will begin at all sides of the globe, and proceed 

 gradually towards the centre, expelling before it all the 

 air, and most of the foreign substances which may be con- 

 tained in the water. If the cold be continued, the freezing 

 will proceed toward the middle, until finally the process 

 would end by collecting at this point a quantity of air sur- 

 prising in amount. Before this takes place however, the 

 glass vessel will be broken by the expansion of the ice. The 

 crystallization at the upper surface of the water will be some- 

 what irregular at first; the spicules of ice around the margin 

 will tend to shoot out at right angles to the surface of the 

 glass ; but after a pellicle has formed over the top of the fluid, 

 this will serve as a point of attachment, and the crystalliza- 

 tion will go on, as in the other case, at right angles to the 

 surface; the air bubbles will be driven down before it, and 

 if the freezing be very gradual the air will be entirely ex- 

 pelled, and the ice assume a perfectly transparent and hom- 

 ogeneous structure. If the freezing be more rapid, the air 



