After milling and/or drying, the seed must be washed and lightly dissolved to eliminate dust.

It is not easy to add the seed at exactly the moment that the mother liquor is saturated. If we act:

- too early (undersaturation), the seed is dissolved;

- too late (supersaturation), a wave of primary nucleation risks occur.

In reality, we must choose the instant at which supersaturation verifies Ac.

There is an advantage in seeding with a small mass of fine particles.

If there is not attrition, we accept that the number of crystals present in the crystallizer is invariable in time and equal to the number of seeds dispersed in the device at the beginning of the operation, but there is a dependence relationship between the number and the size of seeds, on the one hand, and the cooling or vaporization rate, on the other hand.

If supersaturation is constant, so is the growth velocity G. At a given moment, the apparition rate of the crystal mass is:

A_{c} is the crystal surface, all the same size (L_{s} + Gt):

N_{s} is the number of seeds that we assume to grow at the same rate:

M_{s} and L_{s} are the total mass and the seed dimension.

Hence:

On the other hand, if X is the mass of solute per kilogram of the solvent, we can write:

Hence, the relationship between M_{s} and L_{s}, and dX* / dT or dM_{solvent} / dt.

We observe that if dM_{c} / dt is maintained constant, the growth G varies, thus:

This is not easy to obtain, since we have hypothesized that the supersaturation was constant. If we adhere to this last hypothesis, we must necessarily alter dMc/dt throughout the crystallization process. This equation is readily integrated and the crystal mass is accordingly:

We observe in passing that P = 3a, if the crystals, while growing, remain similar to themselves.