This theory is open to many objections. First, there is no reason why the blood wave should not produce a dicrotic elevation when it flows back to the aorta. Second, the narrow lumen of the arterioles cannot be an obstacle to the flowing blood, because if an artery splits up into small branches, the sum of the lumina of the branches is greater than the lumen of the artery. Lack of space, therefore, cannot be the cause of the reflection of the pulse wave. Marey, finally, is mistaken in his conception of the effect of the blood pumped into the aorta by the action of the left ventricle. He supposes that the entering blood pushes before it the whole column of blood in the arteries. This view is refuted by the actual measurements of the velocity of the pulse wave, because if it were true the pulse would appear at the same moment in every part of the body.[48] These are the more obvious of the arguments against Marey's theory. Other investigators have tried to state a more correct theory of the central origin of the dicrotic wave. Landois's theory belongs to this type of improved theories of the central origin. The action of the left ventricle, according to Landois, causes the primary pulse wave which travels down the arterial system, until it is extinguished in the arterioles. The walls of the arteries are expanded by the arriving blood wave, and, when the valves close, they force the blood onward by their elasticity. There is a free way to the periphery, but the blood pushed towards the heart finds the semilunar valves closed and is reflected. In this way a new positive wave originates which may produce in the same way a secondary or tertiary wave.[49] It seemed necessary first to decide between the theories of the central and of the peripheral origin of the dicrotic wave. Many investigations have been carried on for this purpose, and some of them bear witness to the high ability of the investigators. It is, however, remarkable that the arguments which have been brought forward in favor of one hypothesis chiefly consist in reasons why the other hypothesis should not be accepted. These experiments can be divided into two classes. The first class comprises all the experiments which study the relation of the pulse curve to other functions, or its dependence on various conditions. The above mentioned observations of the pathological changes of the pulse curve belong to this class. The object of frequent studies of this type has been the relation of the sphygmographic curve to the curve of the apex beat. The papers of Otto and Haas,[50] Garrod,[51] Traube,[52] Rosenstein,[53]