188 PROF. H. E. ROSCOE ON METALLIC NIOBIUM 



such mixtures containing in addition some other metal, 

 such as tungsten, titanium, or iron. 



Troost determined the vapour-densities of niobium penta- 

 chloride and oxychloride, obtaining the number 9*6 for the 

 first and 7*87 for the second, agreeing closely with the cal- 

 culated densities when the atomic weight of niobium is 

 taken to be 94. 



Some time ago I obtained several crystals of Greenland 

 columbite, which possesses a specific gravity of 5*389, and 

 which contains, as Marignac has shown, little or no tan- 

 talum. From this material I prepared niobium pentoxide 

 (NbiOj), niobium pentachloride (NbClj), and niobium 

 oxychloride (NbOClj) by well-known methods. 



I have likewise prepared metallic niobium and a new 

 chloride of niobium. The results of these experiments are 

 contained in the following communication. 



The following results of analyses of the pentachloride and 

 oxychloride serve to show the purity of the material : — 



Pentachloride. Oxychloride. 



Calculated. Found. Calculated. Found. 



Niobium 34'62 34'38 Niobium 43'42 A'^'9'i 



Chlorine 65-38 65-55 Chlorine 49'i9 49'34 



Oxygen (diff.) 7-39 7-73 



99'93 



lOO'OO 



Metallic Niobium is difiicult to prepare ; and precautions 

 similar to those employed for the preparation of metallic 

 vanadium must be taken. It is obtained as a bright steel- 

 grey, shining, metallic crust by first passing the vapour of 

 the pentachloride mixed with hydrogen repeatedly through 

 a red-hot glass tube. The metallic crust then deposited on 

 the side of the tube is next strongly ignited in a porcelain 

 tube in a current of hydrogen until no further evolution of 

 hydrochloric acid occurs. 



In order to avoid as completely as possible access of 



