M]0 



of tlu'iniometric* (Icxiccs, were induced iiilo error by a mis- 

 iiilerj)re1ati()ii of llie I'eadiii.ns of llierniometeis. Their at- 

 tempt to determine directly tlie tem})erature at the surface 

 of a growing crystal by the finest thermocouple placed the 

 nearest possible lo the surface, is comparable to an attempt 

 at measuring' the 1em])erature of a l)iiniiii,i>' match with a 

 giant thermometer (iiaviiig a bulb several inches in 

 diameter) separated from the match by a distance of a 

 centimeter or so. In the last analysis, the error is due to 

 the fact that too many experimenters believe in ther- 

 mometers; they take for granted that a thermometer is an 

 apparatus which gives the temperature at a point in space, 

 while a thermometer is a complicated system of bodies in 

 contact, through which heat flows, usually in an uncon- 

 trolled manner, and from the behavior of which we try to 

 guess (or to use a euphemism, to calculate) what the tem- 

 perature is at a point in the neighborhood of the system. 

 Since such errors probably escape the attention of the 

 investigators more often than is usually thought, we 

 deemed it justifiable to discuss the fundamentals of 

 thermometry, that is, heat conduction, in a preliminary 

 chapter of the present work. 



Strange as it may seem, then, the fact is that the lower 

 the temperature is, the slower is the freezing (crystalliza- 

 tion) and when the temperature reaches a certain mini- 

 mum, freezing becomes impossible because of the too 

 intense cold. 



Since the maximum of the curve of the numlier of crystal- 

 lization centers (Fig. 5) is at a lower temperature than 

 the maximum of the curve for the velocity of crystalliza- 

 tion (Fig. 8, CDEF), it is clear that at temperatures at 

 which the number of centers is high and the velocity low, 

 a relatively large numl)er of small crystals will be formed, 

 while at temperatures at which the number of centers is 

 low and the velocity high, there will be a small number 

 of large crystals. The latter condition is realized at higher 

 temperatures near the freezing point, the former at lower 

 temperatures. 



