446 Messrs. Marignac, Deville, and Troost on Niobium. 



ing double fluorides of tin and titanium. In all eases the group 

 Hub Fl 3 (if Hnb represents the atomic weight of niobium in the 

 hyponiobic compounds) is exactly equivalent to a molecule of 

 fluoride of tin, Sn Fl 4 , or fluoride of titanium, Ti Fl 4 . 



To bring this into accordance with the doctrine of isomor- 

 phism, • two suppositions are possible. If hyponiobium is an 

 element, then it must be assumed that hitherto unknown fluo- 

 rides of tin and titanium, Sn Fl and Ti Fl, are compound radicals 

 which play the part of metallic elements, just as ammonium 

 replaces potassium. 



- Or hyponiobium is not a simple body as Rose had supposed, 

 but contains one atom of a metalloid which is isomorphous with 

 fluorine. 



Now not only are the fluoride-of-hyponiobium double salts iso- 

 morphous with those of fluoride of titanium, but also with those 

 of oxyfluoride of tungsten. All these compounds acquire analo- 

 gous formulae, if it is assumed that hyponiobium is an oxide, 

 NbO. These fluorine double salts stand then, as regards their 

 composition, intermediate between the potassium fluoride of 

 titanium and the potassium oxyfluoride of tungsten, thus — 



Ti K 2 F1 6 , H 2 0, 

 KbK 2 Fl 5 0, IPO, 

 W K 2 F1 4 2 , H 2 0. 



On this view the chloride of hyponiobium becomes an oxy- 

 chloride, NbOCl 3 ; hyponiobic acid, or properly oxyniohic acid, 

 has the formula Nb 2 5 , or perhaps (NbO) 2 3 . 



Marignac found for the equivalent of niobium, on this view, the 

 number 93 (H = l, = 16) (Hose found 97-6). If hyponiobium 

 is an oxide, several peculiarities in Rose's investigation are 

 cleared up. In his chloride of hyponiobium he always found 

 oxygen, which he ascribed to an admixture of a certain quantity 

 of an oxychloride. But on determining the amount of oxygen, 

 he obtained three times as much as he could have done on this 

 supposition, and he ascribed the result to an error of experiment. 

 His analyses agree, however, better with the formula NbOCl 3 

 than with NbCl 3 . On the new supposition, too, it is easy to 

 see why Rose failed in converting the chloride of hyponiobium 

 into chloride of niobium by excess of chlorine. 



Marignac's formulae have acquired a confirmation in the 

 experiments of Deville and Troost, who have determined the 

 vapour-density of oxychloride of niobium, as they had previously 

 done that of the chloride. 



They found that of the oxychloride of niobium to be 7*87 at 

 440° and 7*89 at 860°. The formula NbOCl 3 requires 7-5. 



To demonstrate directly the presence of oxygen in oxychloride 



