121) 



Fig. 8. Relation between crystallization velocity and temperature. (After 

 Tammann, 1925.) The crystallization velocities are plotted in ordinates, 

 increasing from the origin ; the temperatures are plotted in abscissae, decreas- 

 ing from the origin, the latter being the freezing point. The curve CDEF 

 represents experimental data (it is a composite curve drawn from several of 

 Tammann's observations). The curve AB represents the data that one 

 should obtain if one could avoid the experimental errors inherent in the 

 procedure. 



available thermoneedles have too high a heat capacity to 

 register this local temperature with any accuracy. 



Another source of error pointed out by Tammann is 

 that the heat liberated during crystallization at the grow- 

 ing surface of the crystals causes interruptions in the 

 crystallization process, the latter being resumed only after 

 the dissipation of the heat liberated in the preceding con- 

 gelation ; consequently, the duration of crystallization ex- 

 perimentally determined consists of a succession of crystal- 

 lization periods and interruption periods and the resulting 

 figure for velocity is too low. 



Taking these facts into consideration, Tammann con- 

 cluded that the relation between the velocity of crystalliza- 

 tion and the temperature should be of the type represented 

 by the curve AB, Figure 8. (hystallization would then 

 proceed more slowly at lower temperatures. Indirect 

 temperature determinations furnished an experimental 

 verification of this relation. 



It is of interest to notice that the physicists who, in gen- 

 eral, are better trained than other experimenters in the use 



