The regularity and simplicity expressed by the exact laws of crystalline form repeat themselves in the aggregation of the atoms to form molecules. Here, as there, there are but few forms which are essentially different, and their apparent diversity reduces itself to a few fundamental differences of type. There the molecules aggregate themselves into crystalline forms; here, the atoms aggregate themselves into molecular forms or into _the types of compounds_. In both cases the fundamental crystalline or molecular forms are liable to variations, conjunctions, and combinations. If we know that potassium gives compounds of the fundamental type KX, where X is a univalent element (which combines with one atom of hydrogen, and is, according to the law of substitution, able to replace it), then we know the composition of its compounds: K_{2}O, KHO, KCl, NH_{2}K, KNO_{3}, K_{2}SO_{4}, KHSO_{4}, K_{2}Mg(SO_{4})_{2},6H_{2}O, &c. All the possible derivative crystalline forms are not known. So also all the atomic combinations are not known for every element. Thus in the case of potassium, KCH_{3}, K_{3}P, K_{2}Pt, and other like compounds which exist for hydrogen or chlorine, are unknown. Only a few fundamental types exist for the building up of atoms into molecules, and the majority of them are already known to us. If X stand for a univalent element, and R for an element combined with it, then eight atomic types may be observed:-- RX, RX_{2}, RX_{3}, RX_{4}, RX_{5}, RX_{6}, RX_{7}, RX_{8}.