DENSITIES OF OXYGEN AND HYDROGEN. 97 



o( \\;itfr. After the tulie coiit.-iiiiiiii;- the clilorate and tlie whole apparatus and its 

 coniieeting tubes" bad heeii exhausted, the chlorate was heated till ox^'gen came off 

 t'reel}', when it was cooled, and the apparatus agaiu exiiausted. The gas passed 

 through three tubes each a metre long aud two and a half centimetres in diameter; 

 the tirst was filled with beads moistened with a sti'ong solution of potassium hy- 

 droxide, the second with beads and sulphuric acid, the third vvitli phosphoi'us [)eut- 

 oxide and olass wool so that no channel could form above the oxide. A reeulatino- 

 stopcock kept the pi-essure in the first part of the apparatus at about that of the 

 atmosphere. 



It was convenient to use in tlie combustion the oxygen contained in two globes; 

 they \vere commoidy filled at the same time. 



•^. SYNTHESIS OF WATER. BALANCE AND WEIGHING ON REVERSAL APPARATUS. 



All the weighings of this series were made on the Becker balance carrying 

 twelve hundied grammes, which was mentioned at page 28, and shown in Fi<j,-. 10, 

 page 30. It was mounted on the closet shown in Fig. 16, page 38. It was provided 

 with the apparatus for weighing by revei-sal which is shown in Figs. 1(3 to 19. 

 But at that time, some ti-ifling details were not cpiite as shown by the drawings ; 

 the operation of weighing was, however, effected exactly as by the apparatus in its 

 present condition. 



4. SYNTHESIS OF WATER. MANIPULATION OF TUBE CONTAINING PALLADIUM. 



These experiments were the first in which large quantities of hydrogen wei'e 

 weighed while absorbed in palladium ; so some changes were by degrees introduced 

 into the apparatus. As far as I see, these trifling modifications sim[>ly made the 

 manipulation shorter and more convenient, without affecting the accuracy of the 

 results. But opinions may differ on this matter, so that it is proper to describe the 

 three forms which the tube containing palladium successively assumed. 



In expei-iments 1 and 2, the palladium was contained in two separate hard glass 

 tubes, Fig. 33. To each was fitted a stopcock of common glass by means of aground 

 joint and paraffin. A stojicock of hard glass 

 cannot be made. The nianip)ulation of this ^ 

 tube would have been very simple if a stopcock 

 were anything more than a contrivance for 

 lessening the flow of a gas through a tube. But for an}- precision, together 

 with freedom from large constant error, in manipulation with a gas so light 

 as hydrogen, it was necessary that leakage should be rendered excessively improba- 



Fig. 33. — Palladium tube, lirst form. 



