100 



SCIENCE 



[N. S. Vol. XXXVI. No. 917 



two and two, for if A fits B and C, and B 

 fits C all over each surface they must all 

 be true planes. However, I shall only 

 speak of the figuring of concave mirrors. 



The roughly hollowed glass disk is now 

 laid on several layers of Brussels carpet 

 centrally on a massive horizontal turn- 

 table. The convex iron tool just described 

 is suspended by a universal joint from a 

 lever, and it is counterpoised so that only 

 a portion of the weight of the tool will rest 

 on the glass when it is in use. A com- 

 plicated system of cranks and levers is so 

 arranged that the tool can be driven by 

 machinery to describe loops or curves of 

 any ai'bitrarily chosen size over the glass, 

 and as these loops are described by the tool . 

 the turn-table turns round slowly. In this 

 way every part of the tool is brought into 

 contact with every part of the glass disk 

 in a systematic way. "When working near 

 the edge a large part of the tool projects 

 beyond the edge of the glass. 



Emery powder and water are supplied 

 in a way I need not describe, and the tool 

 is lowered gently on to the glass. The 

 motive power is then applied, and the 

 grinding is continued for many hours until 

 the preliminary rough depression has been 

 hollowed to nearly the desired shape — 

 namely, that of the standard concave iron. 



For finer grinding a change of procedure 

 is now adopted, and very finely powdered 

 emery is used. Another convex tool is 

 formed, by grinding with the standard con- 

 cavity ; the working face of the tool is, how- 

 ever, now cut up into small squares by a 

 criss-cross of narrow and shallow channels. 

 Such channels are found to be necessary 

 in order to secure an even distribution of 

 the emery and water all over the surface. 

 The grooved tool is now used for many 

 hours, and the surface is tested at frequent 

 intervals with a spherometer. The work 

 ceases when it is no longer possible to 

 detect errors of curvature in this way. 



The next stage is polishing. The thick- 

 ness of the layer of glass worn off in polish- 

 ing is to be estimated in ten-thousandths of 

 an inch, and can scarcely be detected 

 even with the finest spherometer. For 

 polishing the iron tool is discarded and the 

 work is carried on by hand. As lightness 

 is essential, the tool is built up by a stiff 

 lattice-work of wood with a continuous 

 wooden working face. It is obvious that 

 however carefully the face may be turned 

 it can not be made sufficiently true, and 

 the requisite accuracy is obtained by means 

 of the plastic properties of rosin or pitch. 

 A number of squares of rosin about a 

 quarter of an inch thick and an inch 

 square are made, and these are glued in 

 rows on the convex face of the wooden tool, 

 with a narrow space intervening between 

 each rosin square and its neighbors. The 

 tool is then warmed slightly so as to soften 

 the rosin a little, and it is then pressed 

 lightly on to the glass disk. By means of 

 this "warm-pressing" a nearly perfect fit 

 is attained. 



Each of the rosin squares is then painted 

 with hot melted wax. This is done because 

 wax is harder than rosin and affords a 

 better working face. Finally, when the 

 tool is quite cold, the surface of the glass 

 is painted all over with very finely pow- 

 dered rouge and water, and the tool is 

 placed gently on the glass with some addi- 

 tional weight resting on it. It is left thus 

 for several hours, but is moved slightly 

 every ten minutes to ensure an even distri- 

 bution of the rouge and water. By means 

 of this "cold-pressing" a perfect fit is 

 secured of the wax-coated rosin squares 

 with the glass face. Cold-pressing has to 

 be repeated every day before the work 

 begins. 



The polishing is now carried on in much 

 the same way as the grinding, but by hand 

 instead of by machine power. The turn- 

 table can be made to tilt so as to bring the 



