HYDRATES AND CARBONATES OF THE ALKALI-METALS, ETC. 457 



suffices for causing it to vanish, as shown by the following entries for (4). Those for 

 (5) and (6) are calculated from the uncorrected numbers. 



Percentage Compositions of — 





(4) 



(5) 



(6) 



Carbonic acid, . 



. 36-320 



36193 



33-021 



Soda, Na 2 0, as carbonate, . 



. 51-270 



51-090 



46-612 



Soda, Na 2 0, as NaOH, 



9-479 



9-253 



15-533 



Water as NaOH, 



2-747 



2-682 



4-502 



Other water, 



. Nil 



0-782 



0-332 



99-816 100-000 100-000 



Calculated percentage of Na 2 in 



No. 4, . . . . 60-749 ; instead of 60-960 as found. 



Of the four products analysed, only No. (5) is proved to have contained a little "free" 

 water ; and its presence is only too easily explained, as having been absorbed on the way 

 from the crucible to the balance. A mixture of carbonate of and caustic soda is of course 

 highly hygroscopic. For the very same reason it is impossible to prove that the products 

 did not contain even a trace of anhydrous oxide of sodium, beside NaOH. But the 

 balance of probabilities is against this assumption. Assuming the absence of Na 2 to be 

 proved, it would not follow that the caustic alkali is formed from the carbonate only 

 by the chemical action of the hydrogen on the C0 2 ; thus : Na 2 C0 3 + H 2 = 2NaOH + C0 5 

 and none of it by a mere decomposition of the Na 2 C0 3 into C0 2 and Na 2 0. The question, 

 I thought, might be decided by a repetition of the experiments in an atmosphere of nitrogen 

 instead of one of hydrogen gas, and I thus came to carry out the following four experiments, 

 Nos. (7) to (10). In Experiments (7), (8) and (9) the modus operandi, apart from the 

 substitution of nitrogen for hydrogen, was the same exactly as in Experiments (4) to (6). 

 As the determinations of soda as sulphate ran away with a great deal of time, the last 

 experiment (No. 9) was started before the analysis of product No. 7 could be calculated ; 

 or else I should have found out sooner than I did what so clearly appears from the 

 following reports, namely, that the hydrogen of the flame in this mode of operating 

 diffuses through the platinum of the crucible, and produces a very tangible amount of 

 NaOH. After some meditation on means and ways for keeping this hydrogen away from 

 the alkali, I at last adopted (for Experiment (10) and later similar experiments) the method 

 which I took occasion to describe on p. 443, &c, of this memoir, in connection with 

 experiments on carbonate of lithia, i.e., I placed the carbonate of soda in a platinum 

 boat, and heated it in a muffle within a porcelaiu tube through which a current of perfectly 

 dry nitrogen was passing from beginning to end. In the first experiment of this series 

 (No. 7) the nitrogen was made ex tempore, by passing a current of dry air through a 

 column of red-hot copper wire gauze and thence direct into the crucible ; but it turned 

 out that what was supposed to be nitrogen contained about 10 per cent, of oxygen. To 

 my surprise the crucible at the end of the experiment was found to have been only very 



