ABSTRACT:In order to clarify the mechanism of removing viruses with cuprammonium regenerated cellulose hollow fiber (BMM hollow fiber), monodisperse gold particles were used. The dependences of the concentration of gold particles in the filtrate on the particle concentration and the particle size were investigated. The particles were considered to be caught by BMM through two mechanisms, that is, plugging of capillaries and trapping within voids. Here, the capillaries stand for the narrow pathway among neighboring cellulose particles which construct the membrane and the voids the bulky space surrounded by the aggregated cellulose particles. In the initial stage, the plugging of the capillaries causes decrease in the particle concentration. On the other hand, trapping leads to the occupation of spaces within the voids. When the voids near the inner surface are occupied by gold particles, the particles proceed inwards through channels formed by voids and finally flow out from the outer surface of the membrane. This leads to increase in the particle concentration in the preceding stage. The removability of the particles depends both on the relative size bewtween the capillary (or the void) and the particle and on the trapping capacity.KEY WORDS Regenerated Cellulose I Hollow Fiber I Gold Particle I Virus I Membrane Structure I Ultrafiltration I In the 19th century, viruses were named a filtrable microbe because of their filtrability through a porcelain plate, In 1937, Elford used a membrane filter prepared from collodion solution in order to decide the size of viruses. 1 From this result, a virus was known to be a particle larger size than protein molecule. He imaged the membrane was constructed with the straight through cylindrical pores and viruses were caught on the membrane surface not in its inside. Recently, this sieving mechanism of viruses with the membrane was again applied to decide the dimension of nonAnonB hepatitis virus. 2 That is, the straight through cylindrical pores in the membrane were coated with albumin so as to minimize the adsorption of virus by the polymeric membrane. He assumed that the pore size distribution was very sharp such as being expressed as <5 function and that all viruses having larger diameter than pore size were caught only on the membrane surface. When an appropriate membrane having smaller pore size than the size of virus is used in a laboratory the filtrate obtained may become virus-free, but this membrane is not adequate for industrial usage. This is because it cannot meet the following demands to the filtration performance:1) The virus logarithmic rejection coefficient cf>v defined by eq 1 is not less than 4, 304 Polym.