]\gt ; On the etical Solutioib by Finite of Physical Problems involving quations , to the a } .
L. F. ( Communicated ) Dr. E. T. Glazebrook , F.R.S. eccived in revised ovember 2 , 1909 , January 13 , 1910 .
) ( Al)stract .
In order to deal with ries , analysis is eplaced by arithmetic , continuous functions are represented by ) of nunlbels , differentials by central .
Problems then fall into two sses : ( A ) The relation -een the equation ) taining thouto the body , the boundary lition is such that the can be stepped out .
This class includes cquations of all rees .
] } been treated nritbmetical differences ) ) , W. F. Shcppard , Heun , .
Kutta , and .
Examples of it ] simple method are ( B ) The integral must be reference to tlJe b whole , as in l)irichlet 's ) .
The method iven only ) out for a limited ' of equations , \mdash ; namely for in connection which a to potential exists , }COIll ) minimum when , and only when , the difference equations satisfied .
Under this condition the ence ) } the and a function of the endents , aving the correct conditions otherwise albitr'ry , can be expanded in the here the constants and are " " principal odes of oscillation\ldquo ; defined by herc is the diffelence equation to be rated and is a we start with the tnble of nunlbcls and calculate .
Then as , we have both sides some number and ) the boundary lmbel S so the boundary condition is still have a new table may be called and Repeati process with , we get ae Now by a sufficient number of suitably chosen the polynomial in on the right can be made small throughout the from to .
Therefore the error of can be made small ; for , since I is one-signed , it is measured by the L.H.S. The process is arithmetical .
Under certain conditions the error duc to finite central differences is of the form , where is the co-ordinate difference and the 's are functions of position independent of .
If the integral has been found for two or more sizes of , more exact values of it can be extrapolated by this formula .
These methods have been applied in the ) aper to calculate the stressfunction in a nlasonry dam .
The 'otion of Systems of Extent , and the oduced by the Resulting Radiation .
By W. , A.R.C.Sc .
, M.A. ( Communicated by Prof. Sir J. Larmor , Sec. .
Received December 17 , 1909 , \mdash ; Read lnlary 2 , 1910 .
) ( Abstract .
) When the exact equations of motion of a dynamical system are known , it is in general possible , by a well-known ) rocess , to determine the initial mode of change from a steady to a variable state , even when the pl'imary equations cannot be completely rated in the reneral case .
The primary object of this ation was to show that the same process is applicable to the equations of motion of finite electrified systems .
That the results would be applicable to the question of the electric inertia of " " electrons\ldquo ; was constantly kept in view ; and it lvas felt to be undesirable to make any approximation deIJending on extl.eme smallness of the electrified system .
The va1ious expressions hitherto used for the electric inertia of " " eleotrons at spoeds parable u , depend , I think , without exception , on consideration of the energy of a steady state .
The " " quasi-stationary principle assumes that , when the in a steady state is known , it is

