It appears that when arsenic is oxidized by nitric acid, 100 parts yield from 152 to 156 of acid, dried in a low red heat. The differences may in part be owing to the metal being partly oxidized at the commencement of the operation. On this account I should suppose 55 or 56 to be the proper quantity of oxygen united to 100 metal to form the acid. Proust and Thenard both found that 100 white oxide, when converted into acid by nitric acid, gave 115 or 116. I have found the same. Now if 116 ∶ 100 ∷ 156 ∶ 134; hence the white oxide of arsenic must contain 100 metal to 34 oxygen, if the data be correct; or the metal and oxygen are as 3 to 1 nearly. It is highly improbable that any inferior oxide subsists, as no traces of such have been found, if we disallow a conjecture of Berzelius on the subject. The white oxide of arsenic must then be considered as the _protoxide_, and the atom of arsenic must weigh 21 nearly, and that of the protoxide 28. It is plain the other is not the _deutoxide_, as it does not contain twice the oxygen of the protoxide; but as the proportion of oxygen in it is to that of the protoxide, as 5 ∶ 3, it may be a compound of 2 atoms of deutoxide, and 1 of protoxide; that is, it may be the _superarseniate of arsenic_, if we consider the deutoxide as the acid, and the protoxide as the base. According to this view, the compound oxide, or _arsenic acid_ of Scheele, is constituted of two atoms of the deutoxide, weighing 70, and 1 atom of the protoxide weighing 28, together making 98, for the weight of an atom of arsenic acid, = 63 arsenic + 35 oxygen: and 100 arsenic take 55.5 oxygen to form the acid, agreeably to the above recited experiments. Singular as this conclusion may appear, the truth of it is put beyond doubt, I think, by the following experiments. I have repeatedly found that 28 parts of white oxide in solution are sufficient to throw down 24 parts of lime, from lime water, so as to produce 52 parts of arsenite of lime, and leave the water free from both elements. This confirms the notion of the atom of protoxide weighing 28.