PROFESSOR ROSCOE’S RESEARCHES ON VANADIUM. 
21 
When perfectly pure the boiling-point was accurately determined with 100 grins, of sub- 
stance; the temperature corrected for freezing-point of the thermometer and mean 
temperature of the column was found to be 126°- 7 under 767*0 millims. of mercury. 
(ii.) Vanadyl trichloride is obtained at once of a bright lemon-yellow colour, and free 
from the compounds which accompany its formation in the preceding reaction, by pas- 
sing dry chlorine over the trioxide gently heated. The reaction is as follows, viz. — 
3(V 2 0 3 ) + 6C1 2 = V 2 0 5 + 4(VO Cl 3 ). 
The liquid, thus obtained requires only to be distilled upwards for a short time in a cur- 
rent of carbonic acid, to remove excess of chlorine and hydrochloric acid, and then recti- 
fied once over sodium to obtain the substance chemically pure. 
The general properties of vanadyl trichloride have been already described by Ber- 
zelius. 
The specific gravity of this liquid was carefully determined at three temperatures, 
and found to be — 
at 14°-5 .... 1-841. 
17°-5 .... 1-836. 
24°-0 .... 1-828. 
The trichloride remains liquid at temperatures above — 15°C. As the coefficient of 
expansion of this liquid has not been yet determined above 24°, it is not possible to cal- 
culate the atomic volume of this liquid, or to compare it with that of phosphorus 
oxychloride (PO Cl 3 ). 
The vapour-density of vanadyl trichloride was determined according to Dumas’s method 
with the following results : — 
Weight of bulb and air at 11° C. and under 776-5 millims. . . 6-5172 grms. 
Weight of bulb and vapour at 18°t> C. and 780 millims. . . 7*0003 grms. 
Residual air =0-0 cub. centim. Capacity of bulb 135*13 cub. centims. 
Hence the vapour-density is found to be 88*20 (H=l) or 6*108 (air =1); the calcu- 
lated density is 86-8 (H=l) or 6-000 (air =1). 
Analysis of Vanadyl Trichloride. Determination of the atomic weight of Vanadium. 
Method 2. — The difficulty attending determinations of the atomic weights of the metals 
from analysis of their volatile chlorides has already been pointed out by Pierre * in the 
case of titanium. The errors here referred to, arising from absorption of moisture by 
these hygroscopic chlorides, may, however, be avoided by sealing up several portions of 
the chloride for analysis at once from a large mass of liquid. 
The following seventeen determinations were most carefully made, and the numbers 
obtained, closely coinciding as they do with the results of the reduction experiments 
already given, serve as a most important control of the reliability of the original method. 
Volumetric Determination of the Chlorine. — Before employing Gat-Lussac’s method to 
this determination, it was ascertained that when the chloride is decomposed by water in 
* Ann. de Ch. et de Phys. 3 ser. t. xx. p. 257. 
