1904 ] Colours in Metal Glasses and Films .
443 1 The membrane , owing to the fact that there is no glazed cap to the tube , goes right up to the ends * * * S 2 .
There is no necessity to have pressure-tight joints .
3 Greater speed in working ; the actual experiment takes from 2 to 3 hours only .
, 4 The form of tube used is such that it will withstand very high pressures_it may be of interest to mention that one of the tubes and membrane stood a pressure of 120 atmospheres without apparent harm .
" We publish this preliminary notice as it will be some time before the experiments can be continued\#151 ; a new apparatus has to be cast , and new porcelain tubes are required .
We hope , by means of the new apparatus , to reduce greatly the guard-ring leak . . . .
We are glad to avail ourselves of this opportunity to thank Mr. W. C. I\gt ; .
Whetham for the kindly interest he has taken in the research and Mr. H. Darwin for designing the pressure apparatus .
\#171 ; Colours in Metal Glasses and in Metallic Films .
" By J. C. Maxwell Garnett , B.A. , Trinity College , Cambridge .
Communicated by Professor Larmor , Sec. R.S. Received April 19 , \#151 ; Read June 2 , 1904 .
( Abstract .
) The first part of the paper is devoted to coloured glasses .
The phenomena which it seeks to explain were observed by Siedentopf and Zsigmondy . !
Expressions are first obtained for the electric vector of the light scattered from a small metal sphere when a tram of plane polarised light falls upon it , the investigation following 01 Rayleigh.f By means of these expressions it is proved , from diagrams and statements given by Siedentopf and Zsigmon y , a the metal particles which they observed in gold glass are spherical in shape when the diameters are less than 10 5 cm .
The fact that such particles are spherical throws light on the manner in which metals crystallise out of solution , the particles taking first a spherical form under the action of surface tension , and later , when they become too large for the forces of surface tension to overcome the crystallic forces , becoming amenable to the latter .
Mi . . .
ei A previously arrived at similar conclusions .
S An investigation into the optical properties of a transparent medium * Cf .
Adie , loc. crt .
t ' Ann. der Pliys .
, ' January , 1903 .
+ ' Phil. Mag. , ' vol. 44 , 1897 , and ' Collected Papers , vol. 4 , p. 305 S * Brit. Assoc. Report , ' Southport , 1903 .
444 Colours in Metal Gla and Films .
[ Apr. 19 containing metal spherules , so that the average distance between two neighbouring spheres is considerably less than a wave-length of light , i- ; .
is next undertaken .
It is shown that every such medium has a perfectly definite colour by transmitted light , depending only on the optical constants of the metal of which the spheres are made , on * the refractive index of the substance in which they are embedded , and on the quantity of metal , but not on the size or distance apart of the |lt spheres .
The intensity of the absorption of light of each colour is proportioned to / x , the volume of metal per unit volume of medium .
It is calculated , by means of the metal constants given by Drude , * that , with glass of refractive index 1-56 , gold glass is more red than yellow , silver glass a little more yellow than red , copper glass considerably more red than yellow , and " potassium-sodium " glass much more blue than yellow , provided always that the average distance ' between two neighbouring particles of metal in the glass be considerably less than one wave-length ; in which case , as stated above , the particles must be spherical .
Metal glasses for which this provision is ii satisfied will be called " regular .
" It is next proved that the presence of metal spheres accounts for the optical properties of regular gold ruby glass , and that the irregularities in the effects of colour and polarisation , sometimes exhibited by gold glasses , are due either to excessive distance between adjacent gold particles or to excessive size of such particles\#151 ; the latter , however , involving the former .
Experiments are described , proving that this regular colour can be produced in a colourless metal glass , containing the metal in solution ( which is the state in the manufacture of gold or copper ruby glass before the second heating ) by the / I-radiation from radium .
Thus , a piece of clear gold glass and a piece of clear soda glass were exposed to the emanation for two days , when the gold glass had acquired an unmistakable pink tint , while the soda glass had turned an intense blue-violet .
In the second part of the paper , the optical properties of media built up out of metal spheres as before , but now so that the volume of metal may have any value between zero and unity , instead of remaining very small , as in metal glasses , are investigated .
The changes of colour , and the final change to almost complete transparency , observed by Mr. G. T. Beilbyf in gold and silver films , are accounted for .
Explanations are also given of the changes of colour on heating , observed by Professor R. W. Wood } , in potassium and sodium films deposited on the insides of exhausted glass bulbs .
The # ' Phys. Zeitschrift/ January , 1900 .
t 6 Roy .
Soc. Proc. , ' vol. 72 , 1903 , p. 226 .
t ' Phil. Mag./ 1902 , p , 396 .
f 1904 .
] The General Theory of Integration .
I increase in strength of colour , which was generally observed in the light transmitted through these films when the plane of polarisation of obliquely incident light was changed from that of incidence to a | perpendicular position is accounted for .
In Part III some evidence is brought to show that the allotropie I ; silvers obtained by Carey ]Lea* are particular cases of the media which I have been considered in the second part .
" The General Theory of Integration .
" By W. H. Young , Sc. D. , St. Peter 's College , Cambridge .
Communicated by Dr. E. W. Hobson , F.E.S. Received April 23 , \#151 ; Read May 19 , 1904 .
( Abstract .
) The paper begins with a recapitulation of the well-known definitions of integration and of upper and lower integration ( integral par exces , par defaut ; oberes , unteres Integral ) .
The theorem on which the Darboux definition of upper ( lower ) integration is founded is stated and proved in the following form :\#151 ; Given any small positive quantity we can determine a positive quantity e , such that , if the fundamental segment S be divided up in any manner into a finite number of intervals , then , provided only the length of each interval is less than e , the upper summation of any function over these intervals differs by less than e\ from a definite limiting value ( the upper integral ) .
Next follows a discussion as to whether it is admissible to adopt a more general mode of division of the fundamental segment than that f used by Riemann , Darboux and other writers , when forming summations ( upper , lower summations ) , defining as limit ' the integral ( upper , lower integral ) , of a function over the fundamental segment .
It is shown by examples first that the restriction as to the finiteness of the number of intervals into which the fundamental segment is divided .
cannot be removed without limitations ; but that it can be removed , provided the content of the intervals is always equal to that of the fundamental segment .
Secondly it is shown that the error introduced by taking the summation over an infinite number of intervals whose content is less than that of the fundamental segment , is not in general corrected by adding to the summation the content of the points external to the intervals multiplied by corresponding value ( upper , lower limit ) of the function .
Similarly it is shown that the more * ' Amer .
Journ. of Science/ 1886 .
2 i VOL. LXXIII .

