Considering compounds of two elements, he divides these into binary, ternary, quaternary, etc., according as the compound atom contains two, three, four, etc., atoms of the elements. He then proceeds thus-- "The following general rules may be adopted as guides in all our investigations respecting chemical synthesis:-- "1st. When only one combination of two bodies can be obtained, it must be presumed to be a _binary_ one, unless some cause appear to the contrary. "2nd. When two combinations are observed, they must be presumed to be a _binary_ and a _ternary_. "3rd. When three combinations are obtained, we may expect one to be _binary_ and the other two _ternary_. "4th. When four combinations are observed, we should expect one _binary_, two _ternary_, and one _quaternary_," etc. Only one compound of hydrogen and oxygen was then known; hence it was presumed to be a binary compound, _i.e._ a compound the smallest particle of which consisted of one atom of hydrogen and one atom of oxygen; and hence, from the data already given on page 130, it followed that the atomic weight of oxygen was 8. Two compounds of carbon and oxygen were known, each containing six parts by weight of carbon, in one case united with eight, and in the other case with sixteen parts by weight of oxygen. From Dalton's rules one of these was a binary, and the other a ternary compound; but as the atomic weight of oxygen had already been determined to be 8, that compound of carbon and oxygen containing eight of oxygen combined with six of carbon was decided to be binary, and that containing sixteen of oxygen (_i.e._ two atoms) to be ternary; and hence the atomic weight of carbon was determined to be 6. In the second part of the "New System" Dalton, guided by these rules, determined experimentally the atomic weights of a great many substances; but this was not the kind of work suited to Dalton's genius. His analytical determinations were generally inaccurate; nevertheless, he clearly showed how the values of the atomic weights of elements ought to be established, and he obtained results sufficiently accurate to confirm his general theory. To make accurate determinations of the relative weights of elementary atoms was one of the tasks reserved for the great Swedish chemist Berzelius (see pp. 162-170). When we examine Dalton's rules we must confess that they appear somewhat arbitrary. He does not give reasons for his assertion that "when only one combination of two bodies can be obtained, it must be presumed to be a binary one." Why may it not be ternary or quaternary? Why must the atom of water be built up of one atom of hydrogen combined with one atom of oxygen? Or, when two compounds are known containing the same pair of elements, why must one be binary and the other ternary?