exposed. In the case of the great rift-valley of the Red Sea, which, according to Triulzi and Hecker, is already isostatically compensated, development may have gone so far that at the deeper parts the sima is uncovered. With a further separation of the blocks, the pieces broken from the margins remain behind as islands. It is to be noticed that these pieces, even if in their highest portions they reach or exceed the level of the continent, need not have the same thickness throughout as the continental blocks. Instead, they need only be essentially broader in the submerged than in the elevated portion. Thus here the only condition which needs to be fulfilled is the one that the proportion between the weights above and below the level of the deep-sea floor is the same as in the great continental platforms. All these views as to the nature of the rift-valleys are not contradictory to those now current, but merely supplement them.
Just as a single rift can be resolved occasionally into an extensive meshed net of smaller rifts (the system of the East African troughs, which pass over into a solitary rift in the Red Sea, forms an example of this), so instead also of a single rift-valley the breaking down of an extensive area can be accomplished. The Ægean Sea is the best example of this. Here a large area was broken up in quite late geological time into single blocks, which were submerged to dissimilar depths. We must assume that the lower layers of the lithosphere have stretched so that the fault-rifts gradually disappear downwards. The amount of horizontal extension can be measured on the inclined fault planes, shown diagrammatically in Fig. 36, so far as they are free. At many other places a land connection has obviously been submerged in a similar way; for example, in the Bass Straits between Australia and Tasmania. But it is easily seen that there is a limit to
