Chemistry and Physics. . 237 
these results, and that chlorine was entirely absent, the conclusion 
is that this gas is either fluorine itself or is a higher fluoride of 
hydrogen. 
In a third paper, Moissan describes the production of this gas 
by the electrolysis of carefully dried hydrogen-potassium fluoride 
kept in fusion at 110°. But the apparatus is rapidly attacked. 
In electrolyzing the liquid hydrogen fluoride, a yield at each 
electrode of from one and a.half to two liters of gas per hour, is 
readily obtained. To test the question whether this gas contained 
hydrogen, it was passed over red-hot iron. The delivery tube 
was connected first with a platinum tube containing dry potas- 
sium fluoride, to absorb any hydrogen fluoride; and second, with 
a similar tube, also 20°" long, containing a bundle of iron wires 
and previously tared. To the end of this tube is attached, by 
means of a rubber joint, first a test tube and then a flask both 
inverted and filled with pure carbon dioxide. The liquid hydro- 
gen fluoride is cooled to —50° by passing a rapid current of air 
through the surrounding methyl chloride, and the current is 
turned on. Immediately the platinum tube containing the iron is 
raised to incandescence by the chemical action going on within 
it, the form of the brilliantly burning wires being visible through 
the walls. After ten minutes, the operation was closed and the 
tube weighed. The iron was found as white crystallized fluoride, 
and on examining the collected gas remaining after absorption 
of the CO, by potassium hydrate, it was found to be only the air 
which the platinum tubes had contained. -On the negative side 
78° of hydrogen was collected, weighing 0:006942 gram. Maulti- 
plying this by 19, to give the corresponding weight of the fluorine 
disengaged, gives 0°132 gram; the increase in the weight of the 
iron having been 0°130 gram. Hence the gas obtained must be 
fluorine.— C. F., cii, 1543; cili, 202, 256, July, 1886. GF. B. 
2. On Methods of avoiding errors due to polarization, in the 
use of double image prisms.—In order to measure the relative inten- 
sity of two images, a double image prism is often employed. 
Two images of A and B are thus brought near together and 
polarized at right angles. These images are observed with an 
analyzer, which is attached to a divided circle, and which allows 
the equalization of the images. When the original bundle of rays 
is polarized, the photometric measures are complicated, since the 
double image prism acts itself as an analyzer and alters unequally 
the intensity of the two sources of light. Thus it is*necessary to 
make a preliminary determination of the amount ef error thus 
introduced. M. Cornu believes that the cases in which the bun- 
dle of rays are partially polarized are more frequent than is gene- 
rally supposed and gives several methods of avoiding this error. 
These methods are simple and are explained in detail in his 
article.— Comptes Rendus, Dec. 20, 1886, p. 1227. a tat 
3. Hlectromotive force due to the Voltaic arc.—Many efforts 
have been made to determine this contrary electromotive force ; 
but the results obtained by different observers have not been concor- 
