with were soak were [4].Experience at the Nizhne-Tagil Metallurgical Combine (NTMC) with the development and introduction of forsterite refractories for steel casting shows that the use of these materials instead of periclase is sensible because they are less prone to wetting by the metal, and they have a lower thermal conductivity [i]. Thus, we were faced with the task of developing gas-permeable forsterite refractories instead of periclase, designed for ensuring shortterm blowing of steel in the ladle with inert gases at 1590-1640~ [2].Experience with the development of gas-permeable refractories for the above purpose, both in our country [3] and also abroad, shows that the most stable properties are attained by forming gas-permeable structures from bodies with a granular composition close to a single fraction distribution. This principle was the basis for developing gas-permeable forsterite refractories.Studies were made using mathematical methods for planning the experiment. The starting materials consisted of calcined dunite fractions 3-2 and 2-0 mm, and finely milled, sintered magnesite used for making forsterite refractories at NTMC (Table i). The experiment plan (Table 2) included the preparation of nine batches of three components, and here the limits of variation of the proportions of components were taken, starting from the formation of coarsely grained structures: the mass part of calcined dunite fractions 3-2 mm (X I) was varied from 70 to 100%; mass part of calcined dunite fractions 2-0 mm (X 2) -from 0 to 30%; and mass part of finely milled sintered magnesite (X 3) -from 0 to 20%. The fourth independent variable was taken to be the fabrication pressure p.Bodies were prepared in laboratory runner mills using a solution of lignosulfonate (LST) a density of 1.21 g/cm 3 for moistening. The specimens of diameter and height 50 mm pressed on a hydraulic press at 40, 55 and 70 N/mm 2, and fired in a tunnel kiln with a of 4.5 h in the zone with a temperature of 1640-1690~The properties of the specimens determined by standard methods, and the channel porosity -by the attestation method with of forsterite materials and Following the treatment of the trial batch (27 experiments on three parallel specimens) regression analysis methods we obtained the following relationships for the properties the batch composition and fabrication pressure Q~ap= 2,123X, + 1, 674X2+3,99Xa+O,