86 
PROFESSOR R. THRELFALL AND MR. J. H. D. BREARLEY 
the reflected ray separated from the incident ray by an angle which is not too large. 
The method is suitable only for the observation of small angular deflexions, for we 
have not been able to engrave our scale with numbers finely enough to permit us to 
see the figures and divisions in the field at the same time. Consequently, a deflexion 
can only be observed by counting the images of the millimetre divisions as they flash 
past. With our present arrangements we can only count the divisions when they 
pass below a certain rate; when the whole deflexion amounts to more than about 
20 millims., the divisions pass too quickly to be counted. However finely we might 
succeed in eno’ravinof numbers on the scale, we fear that the interference images 
would be too blurred to be read. The best method would be to omit every tenth 
division, say. We did not require to trouble about this, however, for the “old” 
galvanometer was always set up at hand, and, as soon as the deflexions became too 
large for the method of observation mentioned, recourse was had to it. 
The diagram (Plate 4) will make the arrangement clear. A deflexion of one 
millimetre division is in angular measure oS'G”, and since each millimetre division of 
the scale covered about 8 micrometer scale divisions, each divisible into 5 parts, we 
may consider that the smallest deflexion which could be read with certainty was say 
one-fortieth of this, or 0'96”. Since the magnification depended on the adjustments, 
it varied slightly from day to day, but the general magnification was, as we say, 
such that the distance between two consecutive scale images covered from 8 to 9 
micrometer scale divisions. We ought to add that the definition of the mirror was a 
good deal better than we have ever seen before in a mirror of about the same weight 
and dimensions. It was worked for us by Mr. Cook, together with a large number 
of similar mirrors, all about equally good. The best unworked glass we have tried 
was incomparably worse. The difficulty in making such thin mirrors out of glass 
arises from the fret that they always change their shape when they come off the 
polishing tool; but in our experience the change of shape is very regular, thus a flat 
mirror almost always becomes concave or convex ; but it does not do so irregularly, 
and consequently the definition remains satisfactory. The glass, of course, is ground 
and polished on both sides. The window through which the light passes to and from 
the mirror, is a strip of patent plate-glass, selected by the method of observing the 
reflected images from the front and back, and cutting the glass so that it forms a 
prism of small angle along the top or bottom edge. Patent plate-glass can be got 
with fairly flat surfaces, which are, however, generally more or less inclined, and so in 
cases of this kind the glass must be cut in such a way that the inclination of the 
surfaces does not disturb the accuracy of the observations. For this method of 
selecting glass, we are indebted to Professor Wright, of Yale. 
(5.) The astaticism of the suspended system should he a maximum. 
This is a most tiresome condition to satisfy, and we have nothing to add to what was 
said about it by one of us (‘Phil. Mag.’ [5], vol. 28, p. 458). In order to magnetize 
the system astatically, four electro-magnets had to be .used ; an attempt that was 
