In 1958, SENO et al has successfully fixed the dye granules in the supravitally stained bone marrow cells and Yoshida tumor cells by the use of mercuric potassium iodide solution which is known as the reagent for the quantative determination of NH(3). But, the fixation method is not perfect for electron microscopy, because the resulted precitants of mercuric potassium iodine-dye complex is easily soluble in pure alcohol, and there is a possibility that the complex is lost during dehydration. Hereupon, the author tried to substitute the dye complex with mercuric sulfide which is insoluble in water, alcohol and ether and by this substitution the granules should be retained safely through dehydration for electron microscopy. Observations revealed that the stable complex of the dye. HgS can be produced in the site of dye granules by exposing the cells, which were previously stained supravitally and treated with mercuric potassium iodide, to (NH(4))(2)Sx solution. The fine structure of the protoplasma proved to be retained through the treatment with (NH(4)) (2)Sx solution giving almost the same picture as those treated only with mereric potassium iodide solution added an equal volume of 2% osmic acid solution (pH. 7.4). Electronmicroscopic observation on the cells stained supravitally with Janus green B and fixed by the method just mentioned revealed that Janus green B granules appear as opaque masses of uniform in density or rings having thick opaque walls and scanty in contents. Morphologic observation proved granules are the mitochondria themselves, because some of them showed the inner structure like cristae. Treatment with (NH(4)) (2)Sx seems to increase the opacity the dye granules, and the prolonged treatment results in the formation of black masses irregulare in shape. The result obtained is almost the same as demonstrated by SENO et al, but in my experiment the possility of solving out of dyes from some cytoplasmic area has been complexely devoided. And now it can safely be said that Janus green stains mitochondria selectively. Dye, even in its lenco type, is not detected in the ground substance of cytoplasm. These observations refute the theory proposed by LAZAROW and COOPERSTEIN in which they claim that the Janus green B enters into the cytoplasma diffusely, but mitochondria only can be recognized as the stained granules by the oxidation of Janus green B. There are some of mitochondria remain without staining, but any other cytoplasmic elements are rat stained by Janus green B as far as the cells the author observed.