Ewart has also observed that in the streaming of the endoplasm, there is a variation of velocity of streaming in different parts of the stream (Figure 31). The middle of the stream moves fastest while the layer near the ectoplasm moves very slowly and the layer in contact with the ectoplasm moves hardly at all. But the endoplasm in contact with the central vacuole moves only a little more slowly than the middle of the stream, and the effect of this is that the outer edge of the vacuole is dragged along with the moving endoplasm. This is an important observation and from it Ewart concludes that the energy which produces the streaming movement must be liberated, not at the boundary between the ectoplasm and the endoplasm, nor at that between the endoplasm and the vacuole, but within the endoplasmic stream itself. In this conclusion Ewart is undoubtedly correct, for as a physical phenomenon, no other conclusion is at present possible. Other experiments made upon the velocity of streaming in plant cells indicate that the streaming process obeys the laws of physics. The velocity varies with the proportion of water present in the endoplasm,--the more water, the faster the streaming (Ewart, ’03). The effect of temperature on streaming, noted first by Corti (’74), and studied by Velten, (’76), Schaeffer (’98), Ewart (’03) and other writers, is also such as would be expected if the endoplasm were a simple physical fluid. The rotational streaming in plant cells, such as those of _Chara_, is very similar to the rotational streaming in paramecium and numerous other ciliates. In these organisms it is often called cyclosis. A paramecium differs, however, from a plant cell exhibiting rotational streaming in that no central vacuole is present. This space in paramecium is occupied by the gullet, the nucleus and some endoplasm which is not in the main stream. The effect of this difference seems to be one affecting velocity only, slowing it down, for in the _Chara_ cell the endoplasm meets with much less friction when moving in contact with the vacuolar wall than when moving in contact with the ectoplasm. Its velocity is still further reduced by the large food vacuoles which are almost always carried by the endoplasm, for these vacuoles behave like solid bodies in the endoplasmic stream. During streaming these vacuoles are often seen coming close to the limiting ectoplasm, when they act as obstructions to the flow of the endoplasm. The velocity of the endoplasmic stream in paramecium is relatively slow, ten to twenty minutes being required for a complete revolution.