536.5.081.0 S. F. Gerasimov, and V. M. FuksovMethods are considered for initiation of the inner interface between metal liquid and solid phases. The dependence of structure and temperature in an indium fi xed point cell on initiation method is studied. Estimates are obtained for the density of heat fl ow with different inner boundary initiation methods. Optimum initiation conditions are determined in order to exclude undesirable thermal connections between a thermometer channel and cell walls.Implementation of the international temperature scale ITS-90 is based on reproducing the temperature of phase transitions for pure substances, i.e., fi xed points. The main reproduction conditions are: purity of the substance used, i.e., not less than 99.9999%; pressure during phase transition of 101.325 kPa; thermal equilibrium between substance phases and thermometer during measurements corresponds to determination of a fi xed point temperature [1].The optimum method for reproducing metal freezing fi xed points used extensively in practice by national metrological institutes contains an additional condition for performing crystallization [2,3]. This relates to the formation of two interfaces of metal solid and liquid phases: outer, arising at a cell wall with a reduction in furnace temperature, and inner, developing in a thermometer channel due to short-term introduction of cold rods after supercooling molten metal. The inner boundary is required in order to exclude slow recalescence after metal supercooling, and a reduction in thermal resistance between a phase interface and a thermometer. It forms with immersion for several minutes of one or two cold quartz rods into the thermometer channel of the cell after recording the start of recalescence within metal. For example, for indium introduction of two quartz rods for 3 min is recommended [2].A model of the metal crystallization process in a cell with two boundaries has been given in [3] in the form of formation within a channel of an unchanging even layer of solid metal and a layer of solidifying metal growing from the cell wall. Layers are separated by liquid metal. The temperature is close to the substance liquidus temperature, and being an optimum result of measurements is achieved during the fi rst 25% of metal crystallization time. This model is used in order to estimate the uncertainty of temperature measurement results, caused by the effect of impurities within metal and thermal equilibrium between phases.In order to estimate the reliability of the crystallization model adopted, the authors have obtained photographs of the structure of the inner phase interface after initiating an indium freezing point within a cell [4]. The choice of this point is due to the fact that its temperature makes it possible to extract the thermometer from the thermometer channel at different freezing stages and to record the form of the inner boundary. From visual observation of the inner boundary immediately after initiation in accordance with the recommended procedure, it follows that...