176 PROCEEDINGS OP THE AMERICAN ACADEMY. 



oxidizing it more ; and worse than this, the gold gets inside and alloys 

 with the iron, bringing the latter eventually to the surface. Before 

 soldering it is better to cover the steel rod below the box with pieces 

 of asbestos, binding them on tightly by means of a wire, so as to diminish 

 the oxidation. 



Before adopting quartz crystal as the insulator various other sub- 

 stances were tried. Mica was tried first of all, using both of the 

 methods subsequently employed by Knipp * in his work on surface 

 tension. Our experience agreed with his, — that it is impossible to 

 secure an absolutely tight joint with mica because of the formation of 

 radial cracks. Nor was it an ideal substance chemically. Carnelian, 

 flint, and agate were next tried because of their known toughness. It 

 was with the last-named substance that we developed the method finally 

 employed for making an absolutely tight joint ; the substance itself, 

 however, proved to be chemically unsuitable, since the hydrated silica 

 which it contains dissolves readily in the hot water. The method which 

 we finally employed for securing a tight joint put very little strain on the 

 agate, so that there was no longer any reason for avoiding a substance 

 because of its brittleness. Quartz crystal was then the natural substance 

 to try. Japanese quartz, however, proved a failure, owing to included 

 water or carbon dioxide, which caused it to crack upon heating ; but the 

 Arkansas quartz which we next tried was not affected by heat, and for 

 work with neutral-salt solutions up to 306°, to which we have as yet 

 confined ourselves, it leaves nothing to be desired. Since the thermal 

 coefficient of expansion is so different in the directions parallel to and 

 perpendicular to the main axis of the crystal, it seemed safer to make 

 the axis of the cup parallel to the main axis of the crystal, and this was 

 done. 



The process employed for making the quartz cup is as follows: A 

 crystal is selected which is perfectly clear and free from imperfections. 

 A slice, in thickness a little greater than the height of the finished cup, 

 is then sawed out at right angles to the main axis. For this operation a 

 thin tinned-iron disk, whose edge is charged with diamond powder, is 

 rotated in the lathe : and the piece, supported on a sawing table, is 

 pressed lightly against the saw by hand, a wet sponge being held against 

 the edge of the saw with the other hand. This operation of sawing is 

 discussed at some length by ThrelfalLf It is both easy and rapid 

 if the saw is in good condition. Care must be taken in this and 



* Phys. Rev., 11, 129-154 (1900). t On Laboratory Arts, pp. 187-189. 



