The problem of loose filler pressure on the walls of a silo shell has already served as a subject of theoretical studies over many decades [I, 2]. The Eeneral basis of these studies couslsts of the fact that all of them, starting from Janssen [3] and endlngwith the newest nonholonomlc theories [4,5], were carried out w~thln the limits of the deterministic approach. This approach makes it possible to predict charge pressure "on the averase," but it does not take account of raudom scatterln 8 observed in numerous experiments [6, 7, etc.].Consideration is riven below to another, i.e. the probabilistlc approach to calculating the pressure of a loose filler on the walls of a silo shell. A modlfied Janssen theory is used in whlch the lateral thrust coefficient is constant, and a stochastic field is formed.Using correlation theory for random functions, probability characteristics are deve/oped for charge pressure. The results obtained are compared with experimental data.1. We consider a round cylindrical silo shell with absolutely rigid vertlcal walls and a horizontal bottom filled with loose material. We ascribe the co1,---of loose material, having radius R and height H, to cylindrical coordinates z, r, and , (Fi 8. 1).Let Z be the tensor field of internal static stresses established in the charge after filling the shell. We denote in terms of u z and o R vertical and radial normal componeuts respectively of field Z in side surface S for the loose filler (with r -R, see Fi E . I).In the Janssen arrangement Here ~ is lateral thrust coefficient, f is coefficient of friction for the loose material on the vessel walls, 7 is specific mass of the charge; a prime indicates derivative for variable Z, Values of ~,, f, and 7 in (i) and (2) are ass,-,~d to be constant. Accordin E to the data of five tests by Janssen [3] for wheat on average ~i = 0.222, 7 " 0.8 kE/dm s. However, it is emphasized that this assumption about the constancy of parameters f, ~, and particularly ~ is often criticized from a deterministic standpoint [4,8]. Besides this, from a probabilistlc viewpoint this assumption is insufficiently risid. As numerous experiments have demonstrated, parameters ~ , f, and 7 are not constant, and they are subject to random variation. In the tests with wheat mentioned above, Janssen established that the value of ~I varies within the limits 0.203-0.235. These variations are caused by a series of random factors present in any actual loose material. We will touch on them brlefly.First, grains of loose materials as a rule have different dimensions, shapes, and physical properties, which it is impossible in principle to predict accurately.Second, partlcles of loose material during vessel filling form a random packln E in which the amount, shape, and locatlon of contact zones (points, lines, surfaces) vary in a random way from grain to grain.Third, silo shell walls have random unevenness.Higher Artillery Co,-M-d School. Odessa.