It is with considerable diffidence that I suggest the following as an explanation of the multitude of facts to which previous reference has been made.

In a mixture of two substances, A and B, we have the following forces active, tending to produce solidification:

1. The cohesion between the particles of A.

2. The cohesion between the particles of B.

3. The cohesion between the particles of A and the particles of B.

With regard to this last factor, it will be seen that there are three cases possible:

1. The cohesion of the mixture A B may be greater than the cohesion of A + the cohesion of B.

2. The cohesion of A B may be equal to the cohesion of A + the cohesion of B.

3. The cohesion of A B may be less than the cohesion of A + the cohesion of B.

Now, since cohesion tends to produce solidification, we should in the first case expect to find the melting-point of the mixture higher than the mean of the melting-points of its constituents, or the curve of melting-points would be of the form given in a, Fig. 3. Here no eutectic mixture is possible.