vestigator must take his solution into the open air, but here again he must be careful, for his clothes and hair may carry enough powered material to inoculate the solutions.
Fig. 2. Crystals Growing in a Tube of Undercooled Sodium Acetate Solution. The amount of material required to inoculate a solution is very, very small, far beyond the sensitiveness of the most delicate balance. Nowhere in science is the importance of the fact that "the very small is as real as the very great" better illustrated than in the case of undercooled liquids; a human hair lightly brushed against a crystal of thymol will collect enough of this material to inoculate a flask of the undercooled liquid thymol. A tube of undercooled sodium acetate may be divided by a piece of parchment paper into two parts (Fig. 3). The inoculation of the solution in A causes the separation of crystals which ultimately appear in B via the parchment. The pores in the parchment, which are of microscopic size, become filled with the undercooled solution, and as the crystals forming in the pores can not be larger than the pores it is evident that these crystals are of microscopic dimensions only.
How far can a liquid be cooled below the temperature at which crystals ought to separate? This depends entirely upon the substance used. With some liquids if the undercooling is greater than a degree or two, crystals at once separate spontaneously from the solution. In the case of water the undercooling has been carried to as low as twenty degrees below zero before crystals of ice separated.
Sodium acetate on being strongly undercooled shows an interesting property, for as the temperature becomes lower and lower the liquid becomes less mobile until at fifty degrees below zero just before spontaneous crystallization the liquid assumes a viscous glassy appearance. Its similarity to glass is more than superficial. When molten glass 
Fig. 3. is cooled it gradually becomes more and more viscous until finally it has all the appearances of a solid. But at no definite temperature does it suddenly harden, as would be the case in cooling mercury or molten lead. The glass differs from the undercooled sodium
