Moreover, it is not correct to say that the simple juxtaposition of the surfaces of fracture will reproduce the initial body. The fragments, so far as analysis has obtained them, are not absolutely what they might have been in the original structure. There they adhered the one to the other, not only by the mere contact of their surfaces of fracture, as is supposed, but in a slightly more complex manner. The fragments of the molecule are joined by bonds. We can picture them to ourselves by supposing these bonds to be like hooks. The hooks, which could only be broken by violence, are called by the chemists _satisfied atomicities_. These atomicities, set free by the breaking up, cannot remain in this condition; they must be satisfied anew. The hook tries to attach itself. In Schützenberger’s experiment the addition of water provides for this necessity. A molecule of water (H⌄{2}O) splits into two, the hydrogen (H) on the one side and the hydroxyl (OH) on the other. These two elements cling to the liberated bonds of the fragments of the molecule of albumin, and thus the bodies were found complete. Schützenberger’s experiment was too violent, too radical, and it gave too large a number of fragments, with their free hooks and atomicities unsatisfied, for rather a large proportion of the water added disappeared during the experiment. In one case this quantity was as much as 17 grammes per 100 grammes of albumin. The molecules of this water were employed in the reparation of the incomplete fragmentary molecules of the albumin. It follows that Schützenberger’s experiment gave too large a number of very small pieces corresponding to far too great a pulverization. The very small fragments are the molecules of acids such as acetic acid, oxalic acid, carbonic acid, molecules of ammonia, and even of hydrogen, which we know we are setting free. But, apart from these products which represent a quarter of the molecule of albumin submitted to analysis, the other three quarters represent larger fragments which may be considered as the real constituents of the building. Thus we find four kinds of groups which may be accepted as natural. The first of these groups is that of the leucins or amido-acids. It proves the existence in the molecule of albumin of compounds of the fatty series. There is also an aromatic group—a pyridine group—and a group belonging to the category of sugars. Imagine a certain grouping of these four series. This would be the nucleus of the molecule of albumin. If we graft on to this nucleus, on to this framework as it were, so many annexes, or lateral chains, the building will be loaded with embellishments; it will have been made unstable and _ipso facto_ appropriate for the part that it plays in the incessant transformations of the organism.