arrangements are to fit together, they must be of the same size, whether they consist of bees, or cells, or balls, or molecules. In the same way Mr. Barker has proved that Iceland spar and nitrate of soda must consist of materials which are not only arranged in the same way, but are of the same size, although we do not know in the least what that material is, nor what its actual size may be.

But this, of course, is not the end of the whole mystery: for not only do two such structures fit together, as might be expected, but each has the remarkable property of making the other crystallize and grow, and that means, as I have just explained, that it draws the other out of the solution where it is liquid into the crystal where it becomes part of a solid structure and lays it down in the exact position in which it fits, just as one bee’s cell is added to another in the growth of the comb. We have advanced a step, but only one step, towards the better understanding of the mystery, and I would beg you to note how we are continually led on by analogies which may be quite false, but which are at any rate fruitful.

Now let me pass to another series of researches which were conducted by Miss Isaac and myself for a few years before I left Oxford and have since been carried on by her with conspicuous success. Still experimenting with the same apparatus, and endeavouring to trace how a solution changes in strength while it is crystallizing, we came across some curious and unexpected results. It is, of course, well known that if a crystal, say of alum, is placed in a weak solution of alum it is dissolved, and only has the effect of making the solution stronger, but that at last a stage is reached at which the solution becomes ‘saturated’ and can dissolve no more, just as a stage is reached at which a soaking sponge will hold no more water. At this point a crystal, if put into it, remains unchanged. But it is quite easy to make the solution still stronger, not by adding alum to it, but by taking water away from it by evaporation: it then becomes oversaturated or ‘supersaturated’, and now a crystal of alum dipped into the liquid will at once begin to grow and to make the solution weaker. In fact, this is an unfailing test by which we can tell whether a solution is saturated or supersaturated; and, more than this, until a bit of solid crystal gets into it the liquid does not crystallize. Keep it in a closed vessel so that no speck of alum can fall into it, and it will remain liquid for weeks or years or as long as you please. But let the smallest possible grain of an alum crystal fall into it, and crystallization will be started; inoculate it with an invisible germ of alum dust, such as must be flying about in the air of any room where dry alum is or has been kept, and you will see the life and growth of a crystal begin when that germ is introduced. This is the most extraordinarily sensitive test, and one that can easily be applied.

In many of our experiments we found that during the first day when we were working with some new substance it would not crystallize from an exposed solution; but on the second day, when the air of the laboratory had become impregnated with crystal germs, an exposed solution would begin to crystallize at once.

Dr. Tutton, one of the most accomplished investigators of crystals, whose refined and beautiful researches were for many years carried on in Oxford, has fully described these effects in his two books on crystals just published, and confirms them from his own experience.

Now, Miss Isaac and I have found in the course of our researches that, as the solution becomes stronger and stronger—say, by the evaporation of some of its water, it continues to be in the ordinary state of supersaturation in which it does not crystallize save by inoculation with a germ of solid alum crystal, but at last it suddenly reaches a condition in which it can crystallize spontaneously, and at this moment it is enough to stir or shake the solution, and you will at once witness the birth of thousands of tiny crystals which appear as a cloud in the liquid and begin to grow rapidly.

A very easy way in which to make this experiment is to dip a clean needle into a drop of evaporating solution on a glass plate and to scratch the glass on which the drop lies: for a time nothing happens, and then suddenly, as the liquid passes into the new condition, a chain of tiny crystals appears along the line of scratch.

That this suspended crystallization has a fixed limit was suspected, and had been predicted by the German chemist Ostwald, but could never be proved until we made our experiments, and it was Mr. Hartley who first helped us to interpret our results. He has subsequently, with his pupils, made a number of investigations on the same subject.

I can show you the two conditions in one and the same drop by using a solution of common potassium bichromate. Crystals begin to grow at the edge of the drop where the liquid first becomes sufficiently strong, and continue to grow slowly in the evaporating liquid which is only slightly supersaturated. But in a few moments other parts of the drop which are thinner become so strongly supersaturated that they begin to crystallize spontaneously; and there you can witness the birth of new crystals which grow rapidly in all directions, because they are growing in a solution which is much stronger than that in which the first crystals are growing slowly.

Does not all this set one thinking? What is taking place we cannot tell, but we can only think of it as in some way analogous to the birth of a cloud; and in both instances we have to picture to ourselves minute invisible particles, of whose shape and size we know nothing, coming together and coalescing till they grow into a drop or a crystal that we can see. But why they do not begin to coalesce as soon as the liquid is supersaturated it is difficult to say. We have to conceive the alum solution as made up of moving particles of alum and of water, and it may be that the particles are constantly coalescing into minute groups, but as rapidly being broken up again, until a moment arrives at which the alum particles are sufficiently dense to cohere permanently; but how they attract one another and arrange themselves into the wonderful structure which makes a crystal, of this we are entirely ignorant. The question brings us back again to our initial mystery, how does the crystal actually grow?

But this is not all. I have said that all solutions seem to behave in the same way, and among them nitrate of soda, which we have already seen growing in perfect regularity on Iceland spar. It appears, however, from experiments made by Mr. Barker, Miss Isaac, M. Chevalier (who was another of my pupils), and myself, that Iceland spar behaves in this respect also exactly like nitrate of soda. In a solution which is supersaturated, but is not strong enough to crystallize spontaneously, not only will inoculation with a crystal of nitrate produce instant crystallization; but inoculation with a crystal of Iceland spar produces the same result. So we have a still more convincing proof of what I suggested a short time ago, that two crystals, which have structures so nearly identical that they can fit together, possess also the power of drawing each other from the liquid state into the solid form of a crystal. Whatever it is which conditions the fitting together of two structures must then also confer upon them this extraordinary power of making each other grow.

If I had more time, I should like to give an account of some of the more important discoveries that have been made about crystals during the last fifteen years, for they would make it easier to