454 Mr. Griffin on the Constitution of 



Observations on Table XV. — Ammonia. 



The measure of the test atom of ammonia, as existing in 

 various solutions, varies from 44*20 septems to 43*26 septems. 

 The mean of 14 examples is 43*73 septems. The differences 

 among the determinations are all so small and so irregular as 

 to be apparently mere errors of experiment. Ammonia there- 

 fore differs from the fixed alkalies, and from the acids, in 

 having a fixed atomic volume in solution. If we compare this 

 volume with the measure of the test atom of anhydrous liquid 

 ammonia condensed by 6i atmospheres, we find an apparently 

 constant expansion of about 3^ septems ; but of this expan- 

 sion we have no corroborative evidence. It may occur, or it 

 may not. 



Reflecting upon the causes that could give to ammonia 

 this peculiar property of a fixed atomic volume in solution, it 

 occurred to me that there might be some relation between 

 the corpuscular constitution of ammonia and that of water. 

 The number of ultimate atoms in an equivalent of ammonia 

 being 8, and those in an equivalent of water being 3, these 

 numbers bear a near proportion to the apparent atomic mea- 

 sure of ammonia, 43*73, and the atomic measure of water, 

 16*07, the last of these two numbers being very nearly three- 

 eighths of the preceding number. I thought it possible that 

 solutions of ammonia might consist of compounds in which 

 water and ammonia were united in the proportions indicated 

 by those numbers. I tried to construct a table of solutions 

 of ammonia on that principle, using the atomic numbers of 

 Berzelius both for water and ammonia, but I did not succeed 

 by this means in obtaining consistent results, and I consi- 

 dered the hypothesis to be untenable. Yet, previous to 

 abandoning it, I thought it proper to try whether better re- 

 sults would ensue if the same principle was applied to the 

 atomic numbers for water and ammonia that have been pro- 

 posed by Thomson and Dumas. It gratified me to find that 

 this threw a new light on the difficulty, and enabled me to 

 calculate the following Table : — 



