336 Mr. G. W. Walker .
Initial Accelerated [ Dec. 17 , Now a function I exists such that \#163 ; bIP*2 = 1 , \#163 ; bIP*P* = 0 , where denotes a summation throughout the region .
Therefore ^i(\lt ; /)OT+i-^)2 = X[AA(i-\#171 ; r1x\amp ; 2 ) ... ( i-am-%2)]2 .
Now by a sufficient number of suitably chosen a 's the polynomial in X2 on the right can be made small throughout the range from to Therefore the error of ( f)m+ican 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 due to finite central differences is of the form e2h2 + eji* + eJA + etc. , where h is the co-ordinate difference and the r 's are functions of position independent of If the integral has been found for two or more sizes of h , more exact values of it can be extrapolated by this formula .
These methods have been applied in the paper to calculate the stress-function in a masonry dam .
The Initial Accelerated Motion of Electrified Systems of Finite Extent , and the Reaction produced by the Resulting Radiation .
By George W. Walker , A.R.C.Sc .
, M.A. ( Communicated by Prof. Sir J. Larmor , Sec. E.S. Received December 17 , 1909 , \#151 ; Read January 27 , 1910 .
) ( Abstract .
) When the exact equations of motion of a dynamical system are known , it is in general possible , by a well-known process , to determine the initial mode of change from a steady to a variable state , even when the primary equations cannot be completely integrated in the general case .
The primary object of this investigation 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 " was constantly kept in view ; and it was felt to be undesirable to make any approximation depending on extreme smallness of the electrified system .
The various expressions hitherto used for the electric inertia of " electrons , " at speeds comparable with that of radiation , depend , I think , without exception , on consideration of the energy of a stead}7 ' state .
The " quasi-stationary principle assumes that , when the energy in a steady state is known , it is Motion of Electrified , etc. 1909 .
] possible to draw true inferences as to the motion of a slightly disturbed system .
The logical incorrectness of this has been generally recognised , but there appears to have been a hope that the inferences would not be far from the truth .
The principle involves two fallacies : ( 1 ) It practically involves the fallacy of differentiating an expression which is true only for a steady state , ( ii ) It involves the fallacy that when the energy of a steady state is found to contain higher powers of the velocity than the squares , the usual process of getting an equation of motion is valid .
It may , of course , happen that a correct result is obtained , but the correctness is only proved by a correct process .
The possible failure of the quasi-stationary principle is readily shown by consideration of a dynamical case .
When a solid body moves through a gas , any change of its velocity sets up a disturbance radiated into the gas .
This disturbance is exactly what is neglected in the quasi-stationary principle , because the energy of this disturbance is not properly included in the energy of a steady state .
In the electrical case a similar disturbance occurs , but compared with the dynamical example the motion in the medium is now the dominating influence .
It seems clear that , to get correct results one must proceed from fundamental equations by what I venture to call the Newtonian method of building up equations of motion .
Prof. Love 's method of analysing the electric vibrations on a sphere supplied the necessary weapon in this method of attack .
The results show that the initial motion from rest in the case of an electrified sphere is governed by two linear equations .
Under certain limitations , they lead to the single equation of motion of an " electron " proposed by Lorentz .
They thus include the circumstances under which his equation is valid , but possess a greater generality which becomes important in optical applications .
In passing to the case of disturbance from the steady motion at high speed , a difficulty arose as to surface conditions .
The point was whether the tangential component of setherial electric force or of electrodynamic force was continuous at a surface separating a charged body from free aether .
It was found to be impossible to arrive at a definite conclusion as to which was correct .
There appear to be arguments for and against each view ; and so the consequences of each condition have been examined .
The new method , using the first of these conditions , confirms Thomson 's expression for transverse inertia , while with the second condition we find confirmation of Abraham 's expression for longitudinal inertia of a " rigidly " electrified sphere , but not of his expression for transverse inertia .
These results for a conducting sphere are compared with the experimental results obtained by 338 Initial Accelerated Motion of Electrified , etc. Kaufraann in the case of Becquerel rays .
Analysis of the experiments shows excellent agreement with Thomson 's formula , hut inferior agreement with Abraham 's formula or the corrected expression proved by the new method .
It appears , however , that a considerable proportion of inertia of ordinary kind is indicated by Kaufmann 's experiments .
The method was applied to the linear and rotary accelerated motion of charged insulating spheres at slow speeds .
It is shown that if m ' is the initial electric inertia for linear motion , the arrangement \m ' at the centre of the sphere and \on'uniformly distributed over the surface gives the inertia effect in rotational motion .
Passing to vibratory motion , we still have two linear equations to determine the motion of a charged conductor , and four linear equations in the case of a charged insulator .
Examination shows several new features which appear to have a bearing on the problem of fluorescence .
It appears that a charged particle under the influence of incident periodic radiation absorbs the radiation for a certain range of wave-length , and outside this range simply radiates it .
General scattering of the incident radiation has no special new significance , but the possibility of absorption raises an interesting point .
Accumulation of the energy could not proceed indefinitely , and re-emission must supervene , although the equations do .
not show how .
Doubtless the first approximation ceases to be sufficiently accurate .
It is of interest that with present estimates only positively charged ions of molecular dimensions are likely to produce observable effects , Lastly , the effect of speed of a moving system on the frequency and damping of fundamental vibrations is considered .
The results indicate a possible bearing on the problem of luminosity .
Note.\#151 ; The paper was practically completed towards the end of 1907 , but publication has been delayed for private reasons .
During the interval I have had the advantage of consulting Sir Joseph Larmor on the points involved .
While he is in no way responsible for any of the views expressed in this paper , his valuable suggestions and criticisms have enabled me to remove many obscurities in the work , and I desire most gratefully to record my appreciation of his encouragement and interest .

