110 
When once the goniometer axis has been put at right angles to the 
normal of the mercury surface, this is the case for each new surface *). 
The two following criteria, derived by Drunks, refer to the purity 
of a mercury surface: a. surface layers increase the reading of the 
compensator. 6. increase the restored azimuth at not too small angles 
of incidence’). To this the following criterion, found experimentally, 
may be added. For a pure mercury surface, which as such gives 
the lowest value, both for the phase-difference and for the restored 
azimuth, the reading of the compensator in the air may vary only 
slowly. If the mercury is not perfectly pure, a clearly perceptible 
rise of the compensator-reading, corresponding to a phase-difference 
of 21:150, is observed soon after the formation of the mercury 
surface, even in a space of greatly rarefied air. That this rise does 
not take place through formation of an adhering layer of air, appears 
also from this, that it proceeds incomparably much more quickly 
than is the case with that in consequence of anairlayer, which forms 
slowly, according to a relation given and tested by Haak *). The 
only conceivable explanation is the coming to the surface of impurities. 
No sufficient improvement is attained by rubbing off the mercury- 
surface with cotton wool, so that the upper layer is removed. 
Addition of clean mercury influences the surface-layer only toa very 
small degree. This experience leads to the conclusion that R6nrGun’s 
pouring-out metbod and that of the communicating vessels of DrupE 
cannot give a sufficiently pure mercury surface. Better is the method that 
WernickE applied to crystals and glass, in which asolution of collodion in 
ether is poured over the surface and the film, which is formed after 
half an hour, is detached from the side of the mercury vessel, and 
slipped off over the mercury. This procedure, however, also fails 
several times, and is often less successful on repeated use of the 
same glass and mercury. The best results were obtained by conveying 
double distilled mercury after filtering through a paper funnel] into 
a glass vessel with a fine, drawn-out point. This vessel fits with a 
ground rim into a cylinder glass, which is also filled with mercury 
through the fine point. The vessel itself is closed by means of a 
ground-in stopper. From the vessel the mercury is run into a glass 
dish with flat bottom and small depth, which had been carefully 
cleaned previously. The dish is kept between filter paper and never 
touched. Alcohol is never used for the cleaning, for fear of impurities 
and oxidation of the mercury. f 
1) C.f. for the centering Haak, loc. cit. 
*) Drupe, Wied. Ann., 39, 492. 1892. 
3) Haak, Thesis, p. 32. 
