August 15, 1901] 



NATURE 



6^/ 



a particular microscope or by any possible microscope. But 

 so much discontinuity, as compared with the grinding action, 

 has to be admitted in any case that one is inevitably led to the 

 conclusion that in all probability the operation is a molecular 

 one, and that no coherent fragments containing a large number 

 of molecules are broken out. If this be so, there would be 

 much less difference than Herschel thought between the surfaces 

 of a polished solid and of a liquid. 



Several trials have been made to determine how much material 

 is actually removed during the polishing of glass. In one ex- 

 periment a piece 6 inches in diameter, very finely ground, was 

 carefully weighed at intervals during the process. Losses oi 

 "070, '032, '045, '026, •032 gni. were successively registered, 

 amounting in all to '205 gm. Taking the specific gravity of the 

 glass as 3, this corresponds to a thickness of 3 6x lo-" cm., or 

 to about 6 wave-lengths of mean light, and it expresses the dis- 

 tance between the original mean surface and the final plane. 

 But the polish of this glass, though sufficient for most practical 

 purposes, was by no means perfect. Probably the 6 wave-lengths 

 would have needed to be raised to 10 in order to satisfy a critical 

 eye. It may be interesting to note for comparison that, in the 

 grinding, one charge of emery, such as had remained suspended 

 in water for seven or eight minutes, removed a thickness of glass 

 corresponding to 2 wave-lengths. 



In other experiments the thickness removed in polishing was 

 determined optically. A very finely ground disc was mounted 

 in the lathe and polished locally in rings. Much care was 

 needed to obtain the desired effect of a ring showing a con- 

 tinuously increasing polish from the edges inwards. To this 

 end it was necessary to keep the polisher (a piece of wood 

 covered with resin and rouge) in constant motion, otherwise a 

 number of narrow grooves developed themselves. 



The best ring was about half an inch wide. When brought 

 into contact with a polished flat and examined at perpendicular 

 incidence with light from a soda flame, the depression at its 

 deepest part gave a displacement of three bands, corresponding 

 to a depth of \\\. On a casual inspection this central part 

 appeared well polished, but examination under the microscope 

 revealed a fair number of small pits. Further working in- 

 creased the maximum depth to 2iA,, when but very few pits 

 remained. In this case, then, polish was effected during a 

 lowering of the mean surface through 2 or 3 wave-lengths, but 

 the grinding had been exceptionally fine. 



It may be well to emphasise that the observations here re- 

 corded relate to a hard substance. In the polishing of a soft 

 substance, such as copper, it is possible that material may be 

 loosened from its original position without becoming detached. 

 In such a case pits may be actually 

 filled in, by which the operation would 

 be much quickened. Nothing sug- 

 gestive of this effect has been observed 

 in experiments upon glass. 



Another method of operating upon 

 glass is by means of hydrofluoric acid. 

 Contrary to what is generally sup- 

 posed, this action is extremely regular, 

 if proper precautions are taken. The 

 acid should be weak, say one part of 

 commercial acid to two hundred of 

 water, and it should be kept in con- 

 stant motion by a suitable rocking 

 arrangement. The parts of the glass 

 not intended to be eaten into are, 

 as usual, protected with wax. The 

 effect upon a polished flat surface is 

 observed by the formation of Newton's 

 rings with soda light. After perhaps 

 three-quarters of an hour, the depres- 

 sion corresponds to half a band, i.e. 

 amounts to \\, and it appears to be 

 uniform over the whole surface exposed. 

 Two pieces of plate glass, 3 inches 



square, and flat enough to come into fair contact all over, 

 were painted with wax in parallel stripes and submitted to. 

 the acid for such a time, previously ascertained, as would 

 ensure an action upon the exposed parts of \\. After removal 

 of the wax, the two plates, crossed and pressed into contact so 

 as to develop the colours, say of the second order, exhibited a 

 chess-board pattern. Where two uncorroded, or where two 

 corroded, parts are in contact, the colours are nearly the same. 



but where a corroded and an uncorroded surface overlap, a 

 strongly contrasted colour is developed. The combination lend^ 

 itself to lantern projection, and the pattern upon the screen 

 [shown] is very beautiful, if proper precautions are taken to 

 eliminate the white light reflected from the first and fourth 

 surfaces of the plates. 



In illustration of the action of hydrofluoric acid, photographs ' 



were shown of interference bands as formed by soda light be- 

 tween glass surfaces, one optically flat and the other ordinary 

 plate, upon which a drop of dilute acid had been allowed to 

 stand (Fig. 2) Truly plane surfaces would give bands straight, 

 parallel and equidistant. 



Hydrofluoric acid has been employed with some success to 

 correct ascertained errors in optical surfaces. But while im- 



FlG. 5. 



provements in actual optical performance have been effected, 

 the general appearance of a surface so treated is unprepossess- 

 ing. The development of latent scratches has been described on 

 a former occasion. - 



A second obvious application of hydrofluoric acid has 

 hitherto been less successful. If a suitable stopping could be 



1 The plates were : 

 - Ptoc. Roy. Inst., 



iitised ii\ the laboratory with cya 



NO. 1659, VOL. 64] 



