Studies with the Rous sarcoma in tissue culture in correlation with electron microscopy will be dealt with more at length, since they involve our own work (29,40, 78). Here, again, the most important observations are concerned with the agent itself. Whereas the study of the cancerous cells obtained from tumors in the bird revealed particles of only one type (fig. 24), as described by several authors (60, 61, 82), the particles found in tissue culture display new aspects. In the tumors from the bird, particles are always extracellular, and even when apparently inside the cell, they are still always extracytoplasmic, contained in vacuoles often communicating with the surface membrane (fig. 23). No absolute proof then can be gathered from these pictures as to the actual site of virus formation. Such images could illustrate engulfment of particles in the external medium as well as their production and release.

After infection of cultures with Rous sarcoma virus, from preparations of supernatant fluids of previous passage tumor-cell cultures, the infective titer, as measured by titration on the chorio-allantoic membrane, reached its highest level at 10 days. Electron microscopy showed an absolutely remarkable increase in this period in the number of virus particles of "classical" type. This increase in particle number can be correlated with the titer by direct counting (78).

In addition, two new morphologic aspects of the particles become apparent. Some particles unequivocally intracytoplasmic are revealed (fig. 26). They occur in dense clusters so that the units are not always easily discerned. Each particle possesses a large nucleoid less dense than that of the classical particle. The outer membrane is delicate and surrounded by distinct ribonucleoprotein granules. The mean diameter of the whole particle measures approximately 40 and that of the nucleoid 33 my,. Since all the viral particles are tightly pressed together with no space between, they appear as if embedded in an agglomeration of ribo-nucleoprotein granules. Such groups of particles have been found thus far very rarely and only at the height of "classical" virus production. Bernhard, Oberling, and Vigier (82) had already observed in some tumors a "granular mass" which they believed to be a kind of "viroplasm" from which typical virus could be formed. Difficulties in the preservation of chicken sarcoma tissue prevented observation of the details seen in tissue culture, yet there is no doubt that these were the same formations. This constitutes, therefore, some evidence in favor of a relationship between this new type of particle and the "classical" Rous virus particle, though the possibility that they are another unrelated virus will have to be ruled out.

The other phenomenon observed in tissue culture was concerned with the "classical" virus particle. Again, at the height of infection, the extracytoplasmic viral particles are very numerous and can be observed not only in typical vacuoles well limited by a membrane, as is the rule, but embedded in a gray homogeneous "matrix" in which a membrane is difficult to detect (fig. 25). It is necessary to consider whether the gray matrix is not due simply to tangential sections of closely packed viral particles, whether it corresponds to modified mitochondria as in erythroblastosis (39), or to a formation similar to the gray bodies found in myeloblastosis (87, 38) which is a likely hypothesis. A relationship between these formations-the intravacuolar virus, the intracytoplasmic particles, and the sequence of a possible cycle of virus synthesis'-is not yet clear.

As may be seen already, there is now no doubt that the application of tissue culture to the study of the Rous sarcoma, though in its very beginning, will provide a working system of most significant value for the attention of the electron microscopist. Already it may be said that:

1. Richness in viral particle number is sufficient to meet the requisites for easy study in the electron microscope.

2. No doubt exists that cytologically some evidence of developmental aspects of virus formation has been revealed.

3. The preparations obtained provide the possible means for developing a morphologic method for direct count of cell-associated virus particles which may immediately be related to biologic activity of similar samples.