37 
arrived at the I’elation between the atomic weights of 
hydrogen and oxygen given in the table as 1 to 5 ’5 (but 
altered to 7 in 1808). The composition of water by weight 
had been ascertained by the experiments of Cavendish and 
Lavoisier to be represented by the numbers 15 of hydrogen 
to 85 of oxygen, and the result was generally accepted by 
chemists at the time, amongst others doubtless by Dalton. 
That in those early days Dalton had actually repeated or 
confirmed these experiments appears improbable. At any 
rate he formed the opinion that water was what he called a 
binary compound, i.e., that it is made up of one atom of 
oxygen and one atom of hydrogen combined together. Hence 
if he took the numbers 85 to 15 as giving the composition 
of water, the relation of Hydrogen =1 to Oxygen would be 
as 1 to 5 ’6, or nearly that which he adopted. It does not 
appear possible to explain why Dalton adopted 5*5 instead of 
5*6 for oxygen; it may perhaps have been a mistake or a 
misprint, as there are two evident mistakes in the table, viz., 
18‘7 for nitrous oxide instead of 13‘9, and 9’3 for nitrous gas 
instead of 97. 
Let us next endeavour to ascertain how he obtained the 
number 4*3 for carbon (altered to 5 in 1808 and 5 '4 later on). 
Lavoisier, in the autumn of 1783, had ascertained the com- 
position of carbonic acid gas by heating a given weight of 
carbon with oxide of lead, and he came to the conclusion 
that the gas contained 28 parts by weight of carbon to 72 
parts by weight of oxygen. Now Dalton was not only 
acquainted with the properties and composition of carbonic 
acid, but he was aware that Cruikshank had shown in 1800 
that the only other known compound of carbon and oxygen, 
carbonic oxide gas, yields its own bulk of carbonic acid 
when mixed with ox}^gen and burnt; and also that 
Desormes^' analysed both these gases, finding carbonic oxide 
to contain 44 of carbon to 56 of oxygen, whilst carbonic 
* Ann de Cliimie, T. .39, r>. 38, 
