MR. 0. V. BOYS ON THE RADIO-MICROMETER. 
170 
melted, so distort the glass, even though it only touches near one edge, as to make a 
double image clearly visible. All difficulty is overcome by using instead the smallest 
^usible quantity of shellac varnish, and applying heat to a certain extent. Of course 
the light reflected from so small a mirror is not sufficient when the usual paraffin lamp 
is employed, but, with oxygen at its present price, there is no reason why a small lime 
light should not be used. I have found that with even a small supply of oxygen the 
light on the scale is abundant, and there is no difficulty in observing a deviation of 
5 - mm. The theoretical defining power upon a scaJe a metre distant of a mirror 
2f mm. in diameter is about "23 mm. 
The image of the cross wire given by a mirror of this size that was used in the 
radio-micrometer shown to the lioyal Society was a sharp line which could be read 
with an accuracy of mm. 
As the little mirror is plane, I have cemented a plano-convex lens of a convenient 
focus in the place of the usual plane glass window which must be used to protect the 
moving parts from currents of air. This is preferable to a double convex lens, because 
the flat surface is more convenient for cementing, but especially because this surface 
by reflection also throws an image on the scale which is invariable in position, and 
which may be used as a reference mark if the scale is moved. 
As the definition of the mirror is still so good that the pow'er of reading a deflection 
in the ordinary way is not materially reduced, no change will be practically neces¬ 
sary in the series of equations 14-18, which are only strictly applicable when the 
defining power is not affected by change of K. 
The junction is the only part of the suspended portion of the instrument which now 
remains arbitrary. I have provisionally assumed, for the sake of arithmetical results, 
a pair of bars of antimony and bismuth 5 X 1 X j mm. fixed parallel to one another 
at a mean distance apart of 1 mm. A less mean distance is impracticable, though it 
would be an advantage ; but'the length and sectional area may be modified if found 
necessary. 
Before considering the effect of varying the proportion of the antimony-bismuth 
bars, it will be convenient at this point to find numerically the value of the combination 
(see fig. 6 ) which has thus been developed. 
They are as follows : — 
1 
— *000387 sq. cm. 
V uv 
3'970 cm. 
= 4-284 X 10“^ 
^KCuv 
= 271-8. 
^ if 6 = 1 mm., 
J 
2 A 2 
