Cryo-electron microscopy of vitrified specimens was just emerging as a practical method when Richard Henderson proposed that we should teach an EMBO course on the new technique. The request seemed to come too early because at that moment the method looked more like a laboratory game than a useful tool. However, during the months which ellapsed before the start of the course, several of the major difficulties associated with electron microscopy of vitrified specimens found surprisingly elegant solutions or simply became non-existent. The course could therefore take place under favourable circumstances in the summer of 1983. It was repeated the following years and cryo-electron microscopy spread rapidly. Since that time, water, which was once the arch enemy of all electronmicroscopists, became what it always was in nature – an integral part of biological matter and a beautiful substance.

Ice formation in aqueous solutions and suspensions involves a number of significant changes and processes in the residual liquid. The resulting effects were described concerning the redistribution of dissolved salts, the behaviour of gaseous solutes and bubble formation, the rejection and entrapment of secondphase particles. This set of conditions is also experienced by biological cells subjected to freezing. The influences of ice formation in that respect and their relevance for cryopreservation were considered as well.

A model of transient heat conduction and solute diffusion with a planar ice front, propagating through a system of finite length was found to be in good agreement with measured salt concentration profiles. The spacing of the subsequently developing columnar solidification pattern was of the same order of magnitude as the pertubation wavelengths predicted from the stability criterion. Non-planar solidification of binary salt solutions was described by a pure heat transfer model under the assumption of local thermodynamic equilibrium.

The rejection of gaseous solutes and the resulting gas concentration profile ahead of a planar ice front has been estimated by means of a test bubble method, yielding a distribution coefficient of 0·05 for oxygen. The nucleation of gas bubbles has been observed to occur at slightly less than 20-fold supersaturation. The subsequent radial growth of the bubbles obeys a square-root time dependence as expected from a diffusion controlled model until the still expanding bubbles become engulfed by the advancing ice-liquid interface. The maximum bubble radii decrease for increasing ice front velocities.

The transition between repulsion and entrapment of spherical latex particles by an advancing planar ice-front has been characterized by a critical value of the velocity of the solidification interface. The critical velocity is inversely proportional to the particle radius as suggested by models assuming an undisturbed ice front. The increase of the critical velocity for increasing thermal gradients shows good agreement with a theoretically predicted square-root type of dependence. Critical velocities have also been measured for yeast and red blood cells.

The effect of freezing on biological cells has been analyzed for human lymphocytes and erythrocytes. The reduction of cell volume observed during non-planar freezing agrees reasonably well with shrinkage curves calculated from a water transport model. The probability of intracellular ice formation has been characterized by threshold cooling rates above which the amount of water remaining within the cell is sufficient for crystallization. The cooling rate dependence of viability exhibits largely different maxima for lymphocytes and erythrocytes at about 30 and 4700 K/min, respectively. The decrease of viability above these values has been attributed to the damaging effect of intracellular ice formation.