1. It was possible to study in the ultracentrifuge by optical methods the behavior of yellow fever virus particles directly in the unaltered serum from infected monkeys. 2. The virus showed an extremely high light absorption in the spectral range of 320 to 440 mµ, which seemed to be its intrinsic property. In a 1 cm. thickness of fluid, the small amount of virus present in unaltered infective serum absorbed about as much light (approximately 25 per cent) in the middle of this range as did all the normal serum proteins present in a combined concentration some 1000 times as great. 3. The concentration of virus in the unaltered serum was found to be of the order of 0.00005 gm. per cc. 1 cc. of a 10–9 dilution, which, as has been shown, may constitute a minimal infective dose for monkeys, would contain approximately 10,000 virus particles. The probability that most of the virus particles were in the inactive form is discussed. 4. In infective serum having a viscosity of 14 millipoises, the particles sediment with a blurred boundary at rates lying between approximately 18 and 30 x 10–13 cm./sec./dyne. Evidence indicates that this spread is the result of an aggregation or association phenomenon. 5. Computations of size are in approximate agreement with those made from ultrafiltration studies. On the assumption that the density of the virus particle is near that of protein, its volume is computed to be at least that of a spherical particle having a diameter of 12 mµ. An assumed density of 1.15 gm. per cc. yields a diameter of 19 mµ, considering the shape as spherical.